Project P917-GI BOBAN – Building and Operating Broadband Access Network Deliverable 10 DWDM technologies for access networks Volume 1 of 4: Main Report Draft waiting for Board of Governors' Approval Suggested readers: Managers, strategic planners, researchers and consultants involved in planning the evolution towards broadband access networks, and in the work on the associated standards. EDIN 0049-0917 Project P917 For full publication November 2000 EURESCOM PARTICIPANTS in Project P917-GI are: Finnet Group British Telecommunications plc. Swisscom AG Cyprus Telecommunications Authority Deutsche Telekom AG France Télécom Matáv Hungarian Telecommunications Company Ltd. Telecom Italia S.p.A. Koninklijke KPN N.V. Telenor AS Hellenic Telecommunications Organisation S.A Portugal Telecom S.A. eircom plc This document contains material which is the copyright of certain EURESCOM PARTICIPANTS, and may not be reproduced or copied without permission. All PARTICIPANTS have agreed to full publication of this document. The commercial use of any information contained in this document may require a license from the proprietor of that information. Neither the PARTICIPANTS nor EURESCOM warrant that the information contained in the report is capable of use, or that use of the information is free from risk, and accept no liability for loss or damage suffered by any person using this information. This document has been approved by EURESCOM Board of Governors for distribution to all EURESCOM Shareholders. 2000 EURESCOM Participants in P917-GI page 1 (73) Deliverable 10 WDM technologies for access networks Preface Broadband access network introduction has been discussed extensively during the last ten years within different research programmes in Europe. The search for a common and ultimate strategy, guidelines and set of technologies somehow hampered real field deployment. In any case the number of users who can really enjoy broadband services are not in proportion with the efforts put into standardisation over the years and V.34 modems still represent the gate to the info highway for most of us. On the other hand, as a result of the extensive research efforts, a number of technologies are ready today or on the verge to be effectively exploited in implementing the broadband access infrastructure. Unfortunately, and despite of the standardisation efforts, the variety of technologies is made even more complex by the differences among vendor specific implementations. After the successful experience of EURESCOM Project P614 “Implementation strategies for advanced access networks”, that addressed introduction scenarios and technology appraisal, the BOBAN Project takes on board to investigate a wide range of issues insufficiently covered so far, such as testing, installation and operation of broadband access networks. The quick evolution of the involved technologies and the major changes under way in the telecommunication industry warrant a short study period to review the state-of-theart, after which the laboratory trials and demonstrator implementations can start. It is also understood and incorporated into the BOBAN approach, that the access networks will be based on a variety of technologies, and a major challenge will be to assure consistent operation across different systems. BOBAN also aims to reflect the ideas raised and proposals discussed during the EURESCOM Senior Managers Conference (4th June 1998). The main objectives of BOBAN are to: Obtain experience with the operation and management implications of some broadband access systems; Develop methodology and experiment with procedures for access network monitoring and supervision; Demonstrate low-cost fibre based access systems; Test and evaluate commercial and pre-commercial low cost DSL systems (particularly ADSL lite); Understand the viability and the performances of power-line modems; Provide a comprehensive and reliable assessment of equipment and systems available or under development; Understand the opportunities and the challenges of new powering solutions; Develop requirements and demonstrate a prototype of broadband access cabinet including powering and mechanical parts; Identify possible application scenarios and evaluate WDM systems for the access network; 2000 EURESCOM Participants in P917-GI page 2 (73) WDM technologies for access networks Deliverable 10 Elaborate the specification for a common broadband access network planning tool; Identify guidelines for optical access systems deployment; Define scenarios and guidelines for a cost-effective migration of FTTH in the access network. It is expected that the findings of the BOBAN Project will significantly contribute to the Shareholders strategic decisions on how to upgrade their access networks to accommodate broadband services. This Deliverable is one out of a series of Deliverables BOBAN is producing to summarise its findings in the different aspects it is studying. This Deliverable reports on the results of the investigations done regarding the use of DWDM technology in the access network. It consists of a public main part (volume 1), and 3 EURESCOM confidential Annexes published as separate volumes. The results summarised in this Deliverable are particularly relevant for managers, strategic planners, researchers and consultants involved in planning the evolution of access networks. Adam Kapovits, Project Supervisor (EURESCOM) 2000 EURESCOM Participants in P917-GI page 3 (73) WDM technologies for access networks Deliverable 10 Executive Summary This report is about the introduction of DWDM technology in the access network. It evaluates the potential of the technology and how it can add value to the telecom business. The study was conducted to find answers to questions such as: What services does DWDM technology enable? What are the benefits of deploying DWDM in the access network for a telecom operator? What are the limitations of DWDM applicability in the access network? The Benefits for Your Company Although current demand for broadband services does not necessarily justify the deployment of DWDM in the access network today, demand forecasts indicate the need for DWDM technology in the near future. This Deliverable provides up to date and operator oriented information about the status and the potential of DWDM technology for deployment in the access network. Main results In our study we confirm that large business customers are typically served today by optical fibre access rings using STM-1 (155 Mbit/s) connections. However, the overall penetration of fibre in the access remains low and most of the business customers are still served via copper. Looking at the residential customer market we found that digital TV and video has not taken off yet, and high speed Internet access at present is limited by the core network, therefore providing more bandwidth in the access would not result in a better service experience of the customers. However, the situation concerning the backbone is rapidly changing; a number of large-scale projects are underway to increase the bandwidth available. Clearly the next step is to provide more capacity in the access, especially, because demand forecasts predict an exponential growth of the traffic. Such a growth in the demand cannot be met by any other means than deploying DWDM in the metro networks. A range of application fields of DWDM in the access is defined based on real or foreseen needs. It is important to note that some of these applications could be implemented in some forms either today using available DWDM technology or in the very near future (probably in 2001) taking advantage of emerging more flexible network elements such as configurable OADMs. The following key applications fields have been identified: Wavelength leasing Wavelength on demand Intra-Enterprise Connectivity / Optical VPN From the operator point of view DWDM technology is an option to introduce flexibility and add new features to their networks that represent value to their customers. DWDM provides the following advantages: page 4 (73) 2000 EURESCOM Participants in P917-GI WDM technologies for access networks Deliverable 10 Simultaneous access to several service optimised networks from one or different operators. Virtual private networks on shared fibres through wavelength dedicated optical ADM with enhanced security. Cost efficient upgrade of PON systems. Increased network flexibility that allows rapid provisioning of bandwidth to satisfy unexpected demands and dynamic configuration and assignment of QoS. Higher fibre plant efficiency through fibre sharing, allowing the mixing of different technologies in the same area (different protocols and transmission formats, analogue and digital with different transmission codes). Technology maturity is another key factor of deployment. Our study shows that for the time being the immaturity of DWDM technology poses a serious limitation to the introduction of DWDM in the access network. Although DWDM offers huge transmission capacities and supports a wide variety of input signal formats, it still suffers from limited flexibility and networking functionality compared to its full potential. For example available transponders emit at pre-selected wavelengths and only fixed OADMs are available, that cannot be dynamically configured. Moreover, optical cross-connects that would enable the implementation of all-optical fully configurable networks are still in the development stage. Ultimately, cost will determine the use of DWDM in the access network. Here quick and inexpensive provision of services is essential. The responses to the RFI we have circulated gave very little information regarding hardware/system costs. One aspect, however, that should be taken into account is that almost all vendors offer “metro-oriented” WDM solutions that should allow the development of WDM access network at a lower cost. Our investment study revealed that today the deployment of DWDM technology in the access cannot be justified purely on a cost basis. We also verified, that the commercially available DWDM access/metro systems are fully compatible with the legacy SDH systems. Furthermore, in the framework of a laboratory trial we demonstrated the feasibility of the PON upgrade using commercially available DWDM components. We have studied a special aspect of deploying DWDM in the access, the convergence with the core network. Ultimately this can lead to an all-optical, end-to-end DWDM network. This technology convergence supports the blurring and disappearance of the traditional demarcation of access and core. Our study shows that the critical point of DWDM convergence is the access node that interconnects the metropolitan and access network domains. The significance of the access node stems from the fact that it interfaces various technologies and architectures, is cost sensitive, must be flexible to accommodate new services / demands and has a large impact on OAM. To enable DWDM deployment in the access network management systems should also evolve from the presently available simple network element monitoring to wavelength channel monitoring and provide all the management functionalities in the optical layer that the SDH layer offers in today's networks. DWDM convergence is envisaged to have a large impact on network structure and planning, from the optical networking point. DWDM convergence is mandatory for applications, such as: 2000 EURESCOM Participants in P917-GI page 5 (73) WDM technologies for access networks Deliverable 10 Wavelength VPNs capable of supporting customers’ native signal formats, Wavelength leasing and Wavelength on demand. Convergence should be expected when the technology becomes mature enough and other options for capacity increase are exhausted, allowing DWDM to migrate from core to metro and then to the access network. However, when these barriers are eliminated DWDM convergence will constitute a real paradigm shift in optical networking, changing the perception we presently have of optical networks. page 6 (73) 2000 EURESCOM Participants in P917-GI WDM technologies for access networks Deliverable 10 List of Authors Marcelino Pousa PT Inovação Teresa Almeida PT Inovação Erik Weis DT -T Nova Mirko Adamy DT -T Nova Piero Bradley IT - OTC Lucas Soldano IT - CSELT Fabrice Bougart FT- France Telecom R&D Serge Mottet FT- France Telecom R&D Anjali Bhatnagar Riise NT – TELENOR R&D André-Fossen Mlonyeni NT – TELENOR R&D Risto Riihimaki AF – ELISA Communications Marco Smit NL – KPN Research Zacharias Ioannidis OTE Thomas Sphicopoulos OTE Klio Kallimani OTE Makis Pagiatakis OTE 2000 EURESCOM Participants in P917-GI page 7 (73) WDM technologies for access networks Deliverable 10 Table of Contents Preface ......................................................................................................................................... 2 Executive Summary .................................................................................................................... 4 The Benefits for Your Company ............................................................................................. 4 Main results ............................................................................................................................. 4 List of Authors............................................................................................................................. 7 Table of Contents ........................................................................................................................ 8 Abbreviations ............................................................................................................................ 10 1 Introduction ....................................................................................................................... 13 2 DWDM in the access network ........................................................................................... 14 2.1 General Network structure ......................................................................................... 14 2.1.1 Metro networks .................................................................................................. 14 2.1.1.1 Structure ........................................................................................................ 14 2.1.1.2 Requirements ................................................................................................. 15 2.1.1.3 Present state ................................................................................................... 15 2.1.1.4 DWDM Solutions .......................................................................................... 15 2.1.2 Access networks ................................................................................................ 16 3 DWDM solutions for business customers ......................................................................... 17 3.1 General trends in telecommunications ....................................................................... 17 3.2 WDM based applications in the business environment ............................................. 17 3.2.1 Wavelength leasing............................................................................................ 17 3.2.2 Wavelength on demand ..................................................................................... 17 3.2.3 Storage Area Networks ...................................................................................... 18 3.2.4 Intra-Enterprise Connectivity / Optical VPN ..................................................... 18 3.3 State-of-the-art in the DWDM systems for business applications ............................. 18 3.3.1 Definitions ......................................................................................................... 19 3.3.2 Evaluation of Available Systems ....................................................................... 19 3.3.3 System and architectures ................................................................................... 21 3.4 DWDM architectures for business customers............................................................ 21 3.4.1 Existing access network solutions for business customers ................................ 21 3.4.2 DWDM installation driving forces .................................................................... 22 3.4.2.1 Interest for a service perspective ................................................................... 22 3.4.2.2 Interest from a networking perspective.......................................................... 23 3.4.3 DWDM introduction constraints ....................................................................... 23 3.5 DWDM network topologies applicability.................................................................. 24 3.5.1 Point to point links ............................................................................................. 24 3.5.2 Point to multipoint links .................................................................................... 25 3.5.3 Ring structures ................................................................................................... 25 3.5.3.1 Upgrade concept with DWDM ...................................................................... 25 3.6 Network evolution paths towards DWDM networks for business customers ........... 27 3.6.1 Enhanced star structured copper networks (xDSL) ........................................... 27 3.6.2 Enhanced Point to Point optical links (SDH) .................................................... 28 3.6.3 Point to multipoint links (ITU-T G.983 ATM – PON) ...................................... 29 3.6.4 Enhanced cable Ring architectures .................................................................... 29 3.6.4.1 Particular issues regarding CR configurations ............................................... 29 3.6.4.2 Proposed ring configuration .......................................................................... 30 3.7 Conclusions ............................................................................................................... 30 4 DWDM Upgrade of existing PON .................................................................................... 32 4.1 Introduction ............................................................................................................... 32 4.2 Passive optical Networks (PON) ............................................................................... 32 4.2.1 General PON architecture .................................................................................. 32 4.2.2 Limitations of Power splitting PON .................................................................. 33 4.3 Drivers for the WDM upgrade ................................................................................... 33 4.3.1 Further requirements on future networks: ......................................................... 34 4.3.2 Basic Upgrade concepts ..................................................................................... 35 4.3.3 Upgrade constraints, restrictions and limitations ............................................... 38 page 8 (73) 2000 EURESCOM Participants in P917-GI WDM technologies for access networks Deliverable 10 4.3.3.1 System constraints ......................................................................................... 38 4.3.3.2 Technological constraints .............................................................................. 38 4.3.3.3 Economic Constraints .................................................................................... 38 4.3.3.4 Operator Constraints ...................................................................................... 39 4.4 WDM Upgrade solutions ........................................................................................... 39 4.4.1 Solutions for business customers ....................................................................... 39 4.4.1.1 Downstream dedicated links in a single-fibre PON ....................................... 40 4.4.1.2 Bi-directional dedicated links ........................................................................ 41 4.4.1.2.1 Bi-directional upgrade based on discrete components ............................. 41 4.4.1.2.2 Bi-directional upgrade based on AWG wavelength routers..................... 41 4.4.1.2 Conclusions.................................................................................................... 42 4.4.2 Solutions for increased demands at single ONUs .............................................. 43 4.4.3 Solutions for increased gross Downstream bit rate ............................................ 43 4.4.4 Wavelength per Service: .................................................................................... 43 4.4.4.1 Wavelength per ONU: ................................................................................... 43 4.4.5 Solution for increased bi-directional capacity demand ...................................... 44 4.4.6 Future networks ................................................................................................. 44 4.4.7 Component issues .............................................................................................. 45 4.4.8 Management and protection aspects .................................................................. 46 4.5 Evaluation of the upgrade solutions and migration path ............................................ 46 4.6 Guidelines for Network planners ............................................................................... 49 4.7 Conclusions ............................................................................................................... 50 5 Convergence of DWDM Networks ................................................................................... 52 5.1 Reference Optical Metropolitan Access Architecture................................................ 52 5.2 Functional description of DWDM access node ......................................................... 53 5.2.1 Use of OXC instead of OADM .......................................................................... 54 5.2.1.1 Optical-electrical-optical conversion ............................................................. 54 5.2.2 Network level interfaces .................................................................................... 54 5.2.2.1 Ring-Ring interconnection ............................................................................. 54 5.2.2.1.1 Protection strategies ................................................................................. 55 5.2.3 Possible functional/physical implementation of ring interconnections .............. 56 5.2.3.1 OXC/Configurable OADM functionality ....................................................... 56 5.2.3.2 OADM/EADM(EXC)/OADM ......................................................................... 56 5.2.4 Feeder/SDH access ring interconnection ........................................................... 57 5.2.5 Feeder/FSAN PON interconnection .................................................................. 57 5.2.5.1 Distributive service PON ............................................................................... 57 5.2.5.2 Interactive services PON ............................................................................... 58 5.2.6 Wavelength routing policy................................................................................. 59 5.2.7 Aggregation of traffic in the distribution network for WDM applications ........ 61 5.2.7.1 Non-transparent solution for traffic aggregation in the distribution network, Example : Wavelength per customer ............................................................................. 61 5.2.7.2 Transparent solution for aggregating traffic in the distribution network, Example: Wavelength per service ................................................................................. 62 5.3 Use of DWDM for the reduction and concentration of network switches ................. 62 5.4 Network migration concepts towards all optical networks ........................................ 64 5.4.1 Migration Phases ............................................................................................... 64 5.4.1.1 Advantages .................................................................................................... 65 5.4.1.2 Disadvantages ................................................................................................ 65 5.5 Time frame for the DWDM application .................................................................... 66 5.5.1 DWDM network element maturity .................................................................... 66 5.5.2 Network Management Systems maturity. .......................................................... 67 6 Conclusions ....................................................................................................................... 68 7 References ......................................................................................................................... 72 2000 EURESCOM Participants in P917-GI page 9 (73) WDM technologies for access networks Deliverable 10 Abbreviations ACTS Advanced Communications Technologies and Services ADSL Asymmetric Digital Subscriber Line APS Automatic protection switching AN Access Network ATM Asynchronous Transfer Mode AWG Arrayed Waveguide Grating BER Bit Error Rate BW Bandwidth DEMUX Demultiplexer DWDM Dense Wavelength Division Multiplexing E/O Electrical to Optical Conversion ESCON Enterprise System Connection EUR Euro (€) EURESCOM EUropean institute for RESearch and strategic studies in teleCOMmunications FCF Fused cascaded fibre technology FDDI Fibre Distributed Data Interface FGR Fibre Grating Router FICON Fibre Connectivity Channel FSAN Full Service Access Network FSR Free Spectral Range FTTB Fibre To The Building FTTCab Fibre To The Cabinet FTTEx Fibre To The Exchange FTTH Fibre To The Home FTTC Fiber To The Curb FTTx Fiber To The x HHI Heinrich Hertz Institute InP Indium Phosphide IP Internet Protocol LAN Local Area Network LED Light Emitting Diode IP Internet Protocol ITU-T International Telecommunications Union – Telecommunications page 10 (73) 2000 EURESCOM Participants in P917-GI WDM technologies for access networks Deliverable 10 MAC Medium Access Protocol MFL Multi Frequency Laser N-ISDN Narrowband ISDN NTE Network Termination Equipment OADM Optical add-drop Multiplexer OAM Operation, Administration and Maintenance O&M Operations and Maintenance OAN Optical Access Network OCH Optical Channel Layer network ODF Optical Distribution Frame ODN Optical Distribution Network OLT Optical Line Terminal ONT Optical Network Termination ONU Optical Network Unit OMS Optical Multiplex Section Layer network OSA Optical Spectrum Analyser OTS Optical Transmission Layer network PDL Polarisation Dependence Loss PLOAM physical layer operation administration and maintenance PON Passive Optical Network POTS Plain Old Telephone Service PS Power Splitter PSPON Power Splitting Passive Optical Network PSTN Public Switched Telephone Network QoS Quality of Service RFI Request For Information SDH Synchronous Digital Hierarchy SLS Shared Mulitchannel Light Source SM Single Mode SME Small / Medium Enterprises SNPM Simple Network Protocol Management SOHO Small Office / Home Office TDM Time Division Multiplexing TDMA Time Division Multiple Access UBR Unknown Bit Rate VDSL Very high bit rate Digital Subscriber Line 2000 EURESCOM Participants in P917-GI page 11 (73) WDM technologies for access networks Deliverable 10 VBR Variable Bit Rate VoD Video on Demand WIC Wavelength Insensitive Coupler WDM Wavelength Division Multiplexing /-er WGR Waveguide-Grating Router WL Wavelength WR Wavelength Router WSF Wavelength Selective Filter xDSL generic Digital Subscriber Line page 12 (73) 2000 EURESCOM Participants in P917-GI Deliverable 10 1 WDM technologies for access networks Introduction This Deliverable presents the results of the evaluation of the opportunities DWDM technology offers in upgrading the telecommunications access networks. Besides evaluating the maturity of the technology, the drivers and the constraints for its introduction are also presented. Today DWDM is widely deployed in the core network, because due to the high distances and the existence of fibre the economics is in favour of it. Regarding the access network the use of DWDM is still an issue of the future, because the technoeconomic analysis shows that the conventional solutions are still favourable. In this Deliverable we summarise the existing situation in the Large Business Networks and assess the potential use of DWDM technology. A special attention is paid to the maturity of DWDM technology, the possible network architectures and to the identification of the possible migration paths towards a full optical network. Drivers and constrains are identified and a simplified investment study is presented, that compares the use of DWDM with the traditional space division multiplexing. In some countries PON systems are used to provide the narrow band services to residential and SOHO customers. The Deliverable looks at how such PONs can be upgraded using DWDM. We also take a look at the issue of interfacing the access and core networks. Here we consider both opaque and transparent networks, with the focus being on the transparent networks. Finally, we provide guidelines on how to migrate towards an end-to-end all optical network based on DWDM technology (the so-called DWDM convergence phenomena). We elaborate the required interfaces and time frame for this development. For who has interest in a deep study on these topics the annexes I, II and III give all information that support this document, these annexes are EURESCOM confidential. 2000 EURESCOM Participants in P917-GI page 13 (73) WDM technologies for access networks Deliverable 10 2 DWDM in the access network 2.1 General Network structure In general the whole network can be subdivided in the three domains: Backbone network, Metro network and Access network. Figure 1 shows schematically these three network domains. Backbone-Network Metro-Network Access-Network Figure 1 - General network domains The requirements and functions of the above network parts are quite different from each other. The Metro network serves as connection between the backbone (transport network) and the access network. Therefore, at each "border", it has to fulfil the requirements of the connected networks. The core network will not be defined because it is out of the scope of this work and was studied in other EURESCOM projects (P709, P918) 2.1.1 Metro networks 2.1.1.1 Structure In the present document special attention was devoted to the metro networks because they are in fact the support infrastructure for the large business users. The term "Metro network" is not very well defined and several interpretations can be found in literature. But it is most common to consider the Metro network as a segment in the access network, with traffic aggregation functionalities, in the study we will also use the term feeder network to refer it. But in the view of the convergence aspect, the metro access network will grow from the core network towards the last drop. In the first implementation, it might connect large business customers and concentration/aggregation points residing in the access network such as access multiplexers, access routers, OLTs etc. Based on the high requirements for such connections, one can assume that, for this segment, ring and mesh structures are promising solutions, though different network topologies are also possible. In the presence of cost-effective fibre based access solutions, the metro access network will be expanded to connect the large segment of small and medium enterprises customers. Therefore the metro access network will grow in relation with the move of the fibre deeper to the customer. page 14 (73) 2000 EURESCOM Participants in P917-GI Deliverable 10 2.1.1.2 DWDM technologies for access networks Requirements The deployment of DWDM and increased transmission rates in the backbone network has a strong impact on the metro networks. But the metro network is confronted with a large variety of services, clients with different protocols and bit rates and customer requirements delivered by the access / customer premises networks. The changing paradigm from switched circuits to packet routing and from voice to data centric traffic enhances the pressure on the metro network. Figure 2 shows the complexity of requirements on the metro network. Backbone network (transport ) Survivability Managability Dynamic provisioring Capacity Scalability Metro Low first- cost Service transparency Data centricity Access / Customer premises network Figure 2 - Requirements on Metro networks 2.1.1.3 Present state Today’s metro networks are generally based on SDH rings and do not employ DWDM technology. Typically these networks are installed to connect LEX, CO, ISPs and POPs. These SDH networks are confronted with the described requirements and have to response on the growing requirements. Some of the main problems are: 2.1.1.4 Scalability: versus fixed Multiplex hierarchy Transparency - ”native” data signals: versus SDH frame Flexibility - Set up of new high bit rate connections: versus adding new SDH connection and/or rebuilding of SDH equipment Emerging new customer clients: versus supported by SDH - fitting or not efficient in the SDH hierarchy Capacity upgrade - reusing fibre infrastructure and equipment: versus installing new fibre rings and SDH equipment DWDM Solutions DWDM technology offers the possibility to satisfy the enhanced requirements on metro networks and overcomes the limits of a pure SDH solution. The availability of multiple different wavelength channels offer additional degrees of freedom such as high capacity, flexibility, scalability and transparency. 2000 EURESCOM Participants in P917-GI page 15 (73) WDM technologies for access networks 2.1.2 Deliverable 10 Access networks In today’s access networks ( also referred here as distribution), there exists a great variety of access technologies. The main part of access networks in Europe are determined by copper solutions. However, the growing demand for broadband services as well as competition is pushing broadband solutions, apart from xDSL systems, to broadband radio systems, PONs and AONs etc.. In general, such broadband systems contain a multiplexing function, which multiplexes the different customer bit rates to a higher bit rate. In some systems, traffic separation, for instance IP/ ATM, is also implemented. Also, the existence of a variety of systems results in a variety of interfaces, which can be either standardised or proprietary. Figure 3 shows a variety of client formats which have to be transported over the access network and integrated into the metro network. Figure 3 - Variety of customer client formats To cope with the multiple protocols and the considerably different bit rates of different client formats DWDM seems a promising solution due to its transparency, high capacity and the possibility to run different services and serve different clients on the same (existing) fibre. Our study shows that the DWDM technology in the access is in disadvantage when compared with the conventional TDM or space division multiplexing as far as the cost is regarded, it has been stated that the price of DWDM equipment should come down by some 40% before this technology is worth deploying in the urban environment. However even if carriers cannot afford it at present, they cannot ignore the fact that DWDM could be the key to the grail of carrier services: just-in-time provisioning of bandwidth and transparency. It is capable of handling different types of signals (SDH, ATM, IP) simultaneously, by using the independent characteristics of the multiwavelength network. This transparency in protocol and signal format/bit rate is one of the major advantages of the optical network concept. page 16 (73) 2000 EURESCOM Participants in P917-GI Deliverable 10 DWDM technologies for access networks 3 DWDM solutions for business customers 3.1 General trends in telecommunications The amount of data traffic has been increasing significantly since the beginning of the introduction of IP-protocol and Internet-services. This trend is going to be continued also in the future. Access bandwidth demand of companies increases as the bandwidth demand of various business applications grows. As an example it can be stated that some SOHO companies will reach the access bandwidth demand of a current large enterprise. This trend is shown in Figure 4. Figure 4 - Development of access bandwidth demand Due to the fact that the access bandwidth demand is increasing companies are moving to faster connection types as the number of 2 – 34 Mbit/s connection slowly decreases and contraversary number of faster connection types, e.g.155 Mbit/s and 622 Mbit/s, increase. This increase in bandwidth demand is a signal indicating the need of high bandwidth networks (DWDM networks) as a mean to overcome the existing fibre systems. 3.2 WDM based applications in the business environment The range of applications of DWDM in the metro/access network depends on a variety of issues including technology maturity, capacity demand and need for new services. Since many components of the above issues are still vague and uncertain, it is difficult to predict with accuracy the actual range of applications. However, a list of application fields can readily be defined based on real or foreseen needs [11]. 3.2.1 Wavelength leasing Resembles to dark fibre leasing but adds the advantage of “carrier manageability” of the optical channel. The carrier should provide the customer with optical channel performance monitoring, alternate path routing of signals, failure recovery and isolation. Thus the value of a leased channel is being increased. 3.2.2 Wavelength on demand A service in which an end-user has access to wavelength channels on demand. This requires fairly sophisticated optical systems and management systems, but has been 2000 EURESCOM Participants in P917-GI page 17 (73) WDM technologies for access networks Deliverable 10 envisioned by a number of vendors as an application that may be useful to many large organisations or smaller operators. Wavelengths on demand could be particularly useful to customers who require vastly different amounts of bandwidth throughout the year or throughout the day, and have the capability of bringing up and tearing down optical channels from their own management systems. This service will not likely be available for at least two years. 3.2.3 Storage Area Networks Storage area networks (SANs) have emerged in the past few years in major corporations and institutions requiring massive levels of data storage or data "mirroring." These SANs are high-speed networks which establish direct connections between storage elements and servers and perform various functions: Disk Mirroring Disaster recovery application that creates a second disk image of critical data. Can be recovered instantaneously. Backup and Restore Disaster recovery application that creates copies of data or tape to protect against human error, equipment failure or catastrophic event. Archival and Retrieval Store of critical data on low-cost media. This can be accomplished internally or through an arrangement with a data warehousing provider. Data migration Moving data from one storage system to another Shared Storage Open and legacy access to common physical storage resources Data Sharing Extraction, movement and loading of data between environments Source: ADVA, CNT, 1999 The play for DWDM in the SAN market is through interconnection. SANs that exist within an enterprise tend to be interconnected via ESCON, SCSI or Fibre Channel today. When SANs need to be interconnected, however, dark fibre per link is typically required, because these protocols do not map onto a TDM format well. 3.2.4 Intra-Enterprise Connectivity / Optical VPN Intra-enterprise connectivity has already been discussed earlier in this report and refers primarily to the demand for interconnection of buildings within a campus or a metropolitan area of a large enterprise. Microsoft is an example of an early adopter of DWDM as a means of increasing its capacity on its corporate-wide network. Metropolitan access WDM solutions can provide a low-cost means of transporting the various data protocols common to an enterprise network across a common fibre optic infrastructure (Optical VPN), sometimes eliminating the need for a SDH intermediary multiplexing stage, greatly simplifying enterprise networks. 3.3 State-of-the-art in the DWDM systems for business applications The chapter presents definitions for the different components involved in the DWDM infrastructures, as well an overview of the existing systems on the market, data obtained with the RFI for DWDM large business solutions carried out by the project. page 18 (73) 2000 EURESCOM Participants in P917-GI Deliverable 10 3.3.1 DWDM technologies for access networks Definitions Functional breakdown to be built with a set among: Transponder: network element performing wavelength converter and DWDM adaptation (Optical) Amplifier: network element performing signal amplification Multiplexer/demultiplexer: network element performing merger/separation of signals into/from one stream. OADM: network element performing wavelength add/drop function Optical switch: network element enabling to connect any of its input/output port to any other, according to a switching signalling. Optical router: network element similar to the switch but acting dynamically after analysis of the incoming information routing label. Optical cross connect: network element enabling to connect any of its upstream input to any of its downstream output. These connections are fixed at first installation and can be modified by upgrade. Passive devices may enter this type of NE enabling the sharing (broadcast) of data or combine the data flows from N port/channel onto M others and vice versa. Filter: Passive elements able to isolate one or a group of wavelength Fibre: As far as long distance transmission is concerned, there is little uncertainty, with the physical medium chosen, as most parts of the networks have been built with G.652 single mode fibres. Nevertheless the new development G.652 still demonstrate a good balance between the dimension of the mode propagation area and the dispersion, providing the best balance between non linearities and dispersion accuracy, being in this sense the most promising option for the access network to carry DWDM signals. 3.3.2 Evaluation of Available Systems Comparing the received information in the RFI, with information provided by some alternative sources referring manufacturers that didn’t answer to our RFI, the results shows that, despite, the small number of the answers received, BOBAN’s market research gives a fairly representative view of the current status of DWDM technology. The most relevant facts about the described DWDM systems are summarised below: Maximum capacity, in most systems, ranges from 40 to 80 Gb/s. This includes up to 16 to 32 individual signals of 2,5 Gb/s each (wavelength separation 100 – 200 GHz). Further increase of this capacity is to be expected and some DWDM manufacturers either have already announced the multiplexing of 80 STM-16 channels and 40 STM64 implemented in near-future versions. In any case, the 40 – 80 (or 160) Gb/s capacity offered by available systems can be considered sufficient for the current as well as the future demands for transmission bandwidth. Electro-optical transponders are used at the entrance of all described DWDM systems (some manufacturers also offer integrated SDH/DWDM solutions employing coloured SDH terminals, at a lower cost). These accept a wide variety of tributary signals (PDH, SDH, ATM, IP etc.) and have optical outputs that generally conform to ITU-T Rec. G.692. Mainly owing to the use of transponders, described systems have SDH compliant and open interfaces these are to a large extend bit-rate and signalformat transparent. Scalability in the transmitted information is accomplished by using the appropriate number of transponder cards. 2000 EURESCOM Participants in P917-GI page 19 (73) WDM technologies for access networks Deliverable 10 Optical Add-Drop Multiplexers (OADMs) are offered by all manufacturers to enhance the connectivity of their DWDM systems. Though, in the products offered, the number of add-dropped channels ranges from 1 to 16, all OADMs are similar in that the above number is fixed as well as in that dynamic reconfiguration of adddropped channels is not yet possible. Optical Cross-Connect (OXC) products are not yet available although most manufacturers have stated that are in the process of developing such products. The following table presents a summary of the actual status of the DWDM technology according our RFI results. Table 1 - Key network elements for DWDM DWDM Network Elements Status of technology Time frame Line systems: Up to 32 multiplexed wavelengths. 80 wavelengths are expected by more manufacturers by the end of 2000. Number of wavelengths Some manufacturers offer 64 wavelengths. 128 wavelengths may be available in 2001. Optical add-drop multiplexers Fixed OADMs available, up to 32 wavelengths. (OADMs) No configurable OADMs available yet, under development by some manufacturers. Optical crossconnects No product available yet. Configurable OADM technology is still immature, possibly no product available before 2001. Wavelength crossconnects based on transponders are expected in 2001. (OXCs) Under development by some manufacturers. Optical switches No product available yet Immature technology, no safe forecast Wavelength routers No product available yet Immature technology, no safe forecast Optical amplifiers (Wideband) 1) 64 x 2.5 Gbps 1) currently available 2) 32 x 10 Gbps (ERICSSON) 2) 2001 3) 80 x 10 Gbps (SIEMENS) 3) 2000 (C+L band, 40 channels / band) Laser modules DFB lasers used in transponders. Fixed-wavelength output (1550 nm window). Transponders Electrooptical, mostly with fixedwavelength DFB lasers. Input signals: either 1310 or 1550 nm window, output signals: 1550 nm window. page 20 (73) Tuneable lasers have already been announced Transponders with tuneable lasers have already been announced. 2000 EURESCOM Participants in P917-GI Deliverable 10 DWDM technologies for access networks DWDM Network Elements Status of technology Time frame Wavelength converters No product available yet. Immature technology, no safe forecast. Filters 100GHz spacing available. 50 GHz spacing is expected by more manufacturers in 2001 50 GHz (64 channels) claimed by at least one manufacturer (OSICOM) 3.3.3 System and architectures All available systems support point-to-point and ring topologies. Protection mainly refers to the aggregate transmission path though some products also anticipate protection at the transponder or multiplexer/demultiplexer levels. System supervision is performed by means of a dedicated Optical Supervisory Channel, outside or inside the wavelength zone used for the transmission of the DWDM aggregate signal (at 1310, 1485, 1510 or 1625 nm). However, due to the lack of configurability of modules, management is usually limited to fault reporting. All offered management systems have “open” interfaces to the TMN. In view of the above, the following remark can be made. Though DWDM offers huge transmission capacities and supports a wide variety of input signal formats (conditions that are essential for the migration of DWDM to the access part of the network), DWDM technology still provides limited configurability and networking functionality. Another important issue regarding the migration of DWDM to the access network is that of cost. In the access area, where usually, there is high demand for quick and inexpensive provision of services by means of mature and standardized technologies, the cost issue is much more complex and renders necessary the techno-economical comparison between DWDM and non-DWDM solutions (such has xDSL and SDH/FITL). In the answers to the RFI, there is very little information on product costs, however one should take into account that almost all vendors offer “metrooriented” DDWDM solutions that are simplified versions of a DWDM core solution. 3.4 DWDM architectures for business customers 3.4.1 Existing access network solutions for business customers European operators appear to have adopted similar solutions to serve Business Customers (BCs). Nearly all of them make use of optical access networks based on a physical fibre ring topology and SDH transmission. BCs are connected through pointto-point or ring logical connections. This creates a rather common “present situation” from which future networks, such as DWDM networks, will evolve. In spite of the commonly claimed “booming” increase of BC capacity demand, most BC connections still appear to be in the under-STM-1 range. This means that considerable further increase of capacity demand may be needed to boost DWDM deployment in AN. It is common practice among operators to develop optical fibre infrastructure in physical ring topology, thus future optical AN will have to make use of ring legacy infrastructure. However, It should be noted that penetration of optical access lines - as percentage of total access lines - is presently well below 1% in the majority of 2000 EURESCOM Participants in P917-GI page 21 (73) WDM technologies for access networks Deliverable 10 operators. This means that new fibre will have to be installed in any case leaving space for new network concepts. Large divergence on the availability of services exists among operators (claimed availability ranges from 99.9% - 99.9999%), although most of them use the same basic architecture and type of optical AN. This means that availability of services depends largely on other parameters, such as operations efficiency. Therefore, to retain or enhance availability when introducing DWDM, OAM issues need careful examination. 3.4.2 DWDM installation driving forces It has already thoroughly described that the services perspective and the need for bandwidth upgrade is for sure the major reason for DWDM introduction. Nevertheless, DWDM may feature some other functions that are to be investigated as switching for network flexibility and service differentiation. For this purpose, the benefits for all actors will be envisioned. 3.4.2.1 Interest for a service perspective Since the bit rates required to deliver digital video to the residential client are quite high, it must be quoted that the bit rate differentiation between business and residential, is no more so clear. So that the main difference remains on the asymmetry of the bit rates since the residential where mainly supposed to be downloading information rather than issuing some. So where’s the difference then? Business clients are more quality and service availability sensitive than residential customers, they also have telecom engineers that are able to have a very accurate service value appreciation and express an extremely focused cost sensitivity. Therefore, the access network operator has to provide a range of solutions according to the customer’s specifications. To answer to some possible customer demand, the following applications have been identified: Service capability enhancement (N channels per fibre) higher bandwidth: When the aggregate bandwidth exceeds the electrical channel capacity, DWDM is an option. (The alternative solutions are the electrical upgrade from STMx to STMx*2 or the fibre enrichment of the link). Simultaneous or selectable access to several operators: the second one would be to be able for any service to send out an RFP to operators and select the best offer, without any change in the access network infrastructure. It must be kept in mind that the latter aim also matches somehow the regulators willingness to introduce competition. Simultaneous access to several service optimised networks: This solution aims at segregating the services in order to link the service multiplexers to a telecom network, the IP router to an IP access network thus enabling an per service/technology throughput and translation optimisation, and thus minder the clients need to upgrade the existing CPE. Virtual private networks on shared fibres through wavelength dedicated optical channel. Enhanced security: While the passive optical wavelength extractors OADM, are located in the Telco’s network. The single client signal would only be available on the fibre drop which is likely to be very short thus less easily accessible. page 22 (73) 2000 EURESCOM Participants in P917-GI Deliverable 10 3.4.2.2 DWDM technologies for access networks Interest from a networking perspective Though the aim for the operator is to provide services to its clients, some benefits from new technologies can be taken for internal purposes, such as operational cost savings. But this implies either the absolute transparency of technology to the end user, or the sharing of the benefits through new features or cheaper services. The major features of DWDM for the operator are the following: Minor cost PON system evolution. From the infrastructure point of view, unless extra fibres have been laid from the start, the lowest bandwidth upgrade possible for the clients is to skip from a FSAN 155/155Mbps system to a 622/155Mbps which are both included in the G.983 standard, then move to the 622/622Mbps which is likely to get studied within the next months. Beyond, no plans have been drawn on the electrical upgrade. Higher fibre plant efficiency through fibre sharing: The operator’s interest is to provide a maximum of services through its existing infrastructure, due to the cost of new fibre laying and the fee per civil work meter introduced in some countries. Individual service management on a shared infrastructure: While in TPON or APON systems, the bandwidth and the user management is strongly correlated through the sharing, in DWDM the bandwidth are nearly independent from one user to another, at least it isn’t anymore physical layer dependent. Each wavelength channel’s bandwidth may evolve at its own pace, the bottleneck is likely to become the operator’s first electrical access NE. Mixing of technologies in the same area: 3.4.3 Depending on the market, ATM or IP or DTM NEs will be the cheapest ones either for the Access Network operator and the end user, if he’s the DWDM modem owner. So it will be of major importance to be able to accommodate the latest and cheapest technology with minor infrastructure changes. It is also important to have a technology which will fit into most of the FTTx (X= O, H, B, C, R, ) architectures. DWDM introduction constraints It is clear that unless some strong access network issues are solved, the DWDM in the access benefits listed before won’t convince any operator to take all risks involved. Management of a set of fixed wavelength network terminals: If the DWDM used is a point to multipoint or point to point scheme with a passive wavelength splitting device, each drop will then be coloured. That does mean that the end user NE should match the wavelength or wavelength range allocated. If the NE wavelength is fixed, then it implies to the operator to have a full set of NE to manage, which induces important costs. OAM procedures for DWDM networks monitoring: the introduction of DWDM technologies for business clients are possible only if the operator is given the tools to monitor each channel’s status, detect faults, identify and localise the default. This is of importance for point to point links, but even more for DWDM fibre sharing schemes. Wavelength allocation management in the infrastructure (fixed, dynamic): DWDM introduction in the access, introduces a new parameter to be managed into the OAM procedures, the wavelengths. This induces an upgrade of all supervisory 2000 EURESCOM Participants in P917-GI page 23 (73) WDM technologies for access networks Deliverable 10 software, which might be very light in case of point to point channel upgrade or complex for point to multipoint schemes especially if these allocations are dynamic. 3.5 DWDM network topologies applicability 3.5.1 Point to point links All solutions based on DWDMmux and DWDMdemux can be classified under this category. They all deal with one user and the central office, it is independent whether the physical fibre link uses a physical ring or a straight path between user and LEX building. If linked through a ring, these links can be secured with exactly the same mechanisms as in SDH: G783 defined K1, K2 Automated protection switching. All suppliers have a solution for point-to-point topologies, since it implies no intermediate passive plant feature and is the lowest grade DWDM. Any upgrade in this scheme requires the replacement of the NE on both sides of the fibre. One very special issue, is whether the outcoming modems developed for the LAN world will satisfy the access network operators or not, since they most will probably offer few monitoring functions and quality monitoring but at the same time be the ones at the lowest costs. These lowest costs will be due both to the low functions available and to the higher market than the raw access network. The cost has a major impact on the future deployment of DWDM in the access network . From our studies, considering two different scenarios for a point to point architecture and the two options presented in the following pictures: duplication of systems (Space Division Multiplexing) or the increase of bite rate in the electrical equipment, was tested. N x 1Fibre 1 Fibre Fiber optic terminal 1 1 10 Gbit/s 1 1 Fibre Fiber optic terminals ouo equivalente Fiber optic terminals 1 Increase of bndwidth 2.5 Gbit/s Fiber optic terminal 3 1 1 W D M 1 Fibre 4 Fiber optic terminal a) WDM versus SDM 1 W D 3 M 3 4 4 F iber optic terminal Fiber optic terminal W D M 1 Fibre 1 2.5 Gbit/s Fiber optic terminal 1 10 Gbit/s Fiber optic terminal 1 W D 3 M 4 Fiber optic terminal b) WDM versus bite-rate increase Figure 5 - Point to point and ring architecture used today for business customers In the case a) DWDM versus SDM we considered 3, 6 and 9 terminals using the same cable over the average distance of 8 Km, and STM-1 lines for each customer. The figure shows that for this scenario DWDM is not a competitive solution even considering the cost of the fibre in the SDM case. In case b) DWDM versus increase of bite rate, we considered the case with 3, 6 and 9 circuits ( STM-1) per node and average distance of 8 Km. the results shows that from a economic point of view the DWDM solution never is advisable. The case with 3 circuits show the same value for both solutions, this is due to the fact that we are overdimensioning the equipment, use a ADM STM-16 with protection to support only 3 circuits, so the investment curve for this case has a very slight increase. page 24 (73) 2000 EURESCOM Participants in P917-GI Deliverable 10 3.5.2 DWDM technologies for access networks Point to multipoint links Point-to-multipoint solutions are adapted to residential and SoHo markets, where the demand for broadband services (bandwidth) are not developed, so the presence of this type of systems is yet residual, only some operators are using it for narrow band applications. The identification of limitations regarding the downstream capacity, and the importance of this system for the fibre deployment in the access justified our decision to devote a dedicated section to this field of PON upgrade. 3.5.3 Ring structures Among the various components the quality of service expected by business customers, transmission quality, availability and confidentiality are the most basic requirements. Due this presently, taking advantage of the SDH ADM availability, ring topology is broadly used as a cost effective survivable network architecture allowing bandwidth sharing and improved survivability. SDH ring architectures are widely installed in the transmission network and in the upper level of the access network (metro network). A techno-economical evaluation made in EURESCOM P413 Task 2 project shows SDH STM-1 ring solution has lower cost, for large customers, than point to point SDH STM-1 links. 3.5.3.1 Upgrade concept with DWDM The concept of a Coloured Node Ring (CNR) architecture which is a logical star configuration implemented on a physical ring (A Hamel et al "Coloured Node architecture for business customers", OAN’97) is the most favourable option. The driving idea of the CNR architecture for business customer application is to assign a dedicated wavelength to each customer or node. This solution combines the advantages of point to point links (confidentiality, security) and the advantages of SDH (end to end performance monitoring, protection and restoration capacities, civil work shared between several users). This architecture is potentially cost effective because the Optical Add Drop Multiplexer (OADM) are passive components, and simplified terminal SDH equipment can be used instead of more expensive SDH ADMs, while using the same fibre pair as in a conventional SDH ring. This architecture can be upgraded with the introduction of ATM enabling statistical multiplexing. Transmission Network ADM Network Manager 1, 2, 3, 4 Acces Network 4 STM1 customer 4 protection fibre 3 1 customer 1 working fibre 2 customer 3 customer 2 Wavelength Add Drop Multiplexer Figure 6 - Architecture with Coloured Node Ring (CNR) The ring full length is according to the number of OADM within 28 dB budget loss and without optical amplification. An optical connector (0.3 dB) is installed at the end of each fibre section. The LEX OADM includes an optical coupler for multiplexing 2000 EURESCOM Participants in P917-GI page 25 (73) WDM technologies for access networks Deliverable 10 and a grating or multilayer demultiplexer. The distance between OADM and CP is assumed to be less than 300 meters. Our investment studies evaluate from today cost if DWDM is an attractive alternative to SDH. Transmission Netw ork Transmission Netw ork OADM SDH ADM Netw ork Manager Netw ork Manager STM-x SDH-ADM 1, 2, 3, 4 OADM Acces Netw ork STM-x STM1 4 STM-x Acces Netw ork 1 STM-X customer 4 protection fibre customer 1 w orking fibre customer 4 protection fibre 3 STM-x STM-x customer 3 customer 2 customer 1 w orking fibre 2 customer 3 SDH Add Drop Multiplexer customer 2 Wavelength Add Drop Multiplexer a) SDH ring b) WDM ring Figure 7 – SDH ring and the correspondent WDM solution Basically the comparison of the SDH ring with the WDM ring was considered, where each customer is served by one wavelength, in fact the ring is converted in a point to point topology. The study starts from the assumption that the SDH STM-4 ring is saturated and at this time, a change from the SDH ring to a WDM ring is a favourable situation for WDM as far as a technical aspects comparison is concern. WDM is used to duplicate the SDH rings capacity, considering that fibre is not available in place. SRD ring / WDM ring 100.000 90.000 80.000 70.000 60.000 Cost/customer 50.000 40.000 30.000 20.000 10.000 0 3 n SDH 1 customer/node 3n WDM 6 n SDH 6 n WDM Nº of Nodes 2 Customers/node 8 n SDH 8 n WDM 4 Customers/node Figure 8– Cost comparison, SDH ring / WDM ring The graphic in the Figure 8 shows that for both technologies SDH/WDM the cost per customer decrease with the number of customers. With exception of the SDH case with three nodes and one customer per node, this was due to the fact that we are overdimensioning the SDH, in any other case we couldn’t find a DWDM solution less expensive than the electric option, revealing again that the motivation to install DWDM can not be only the solution cost. page 26 (73) 2000 EURESCOM Participants in P917-GI Deliverable 10 3.6 DWDM technologies for access networks Network evolution paths towards DWDM networks for business customers Looking for the existing situation we identified three main applications of DWDM in the access network: The SDH point to point links capacity multiplication or service multiplexing by DWDM. Referred to as case 1. The proprietary coloured loops as an alternative of MAN network, to join one client’s different locations. Referred to as case 2. The fibre sharing between several client/application. Referred to as case 3. Of course these three cases don’t preclude of any other interest that might emerge especially with appliances towards full optical networks featuring wavelength continuity with the core network. This will most probably occur when optical routing/switching will be available and affordable. The lower level time domain existing switches will then be likely to disappear. 3.6.1 Enhanced star structured copper networks (xDSL) While the two first cases were referring to very big business client the one here rather refers to small and medium business. Therefore it is much more likely to generate volumes sufficient to induce learning curves for the technology and significant cost reductions. Therefore, even if it is further away both in time and technical feasibility, than the first cases it should be given a special care. Due to fierce competition, some operators are making service offers through ADSL, thus all know that within a short time frame the capacity will be too limited and while migrating towards other DSL technologies, the reach and technical constraints will lead to much higher costs. Therefore, there is a need for a graceful migration path towards optics at lower costs with just in time providing constraints, compatibility with the existing architectures, services, management and interfaces. So what we identified were the following 3 steps scenario: Step one: deploy the DSL technology The shot has been already fired. Both DSLAMs and client modems have been defined, dimensioned, service offers and requirements been published and interfaces are known. Step two: opticalize the DSL existing systems and areas, and enhance them. In order for the network planner to avoid blocking situations due to rapid growth of the demand, it could be wise for the operators to gradually install fibre in the access directly to some clients, under the conditions of not being obliged to roll out a fully new technology. Step Three: densify the use of existing optical links If both step1 and step2 are successful, it is time to envision as soon as possible to ameliorate the use of the pioneer optical links through the use of DWDM. Here again the challenge is to remain transparent to the clients already linked and the existing systems, architecture and NE already deployed. The plan is then to accommodate several clients or NEs (including why not radio base stations ) on the same fibre, through the use of a passive DWDM (Phasar or grating) in the plant, each branch/drop after the DWDM would then be dedicated to one 2000 EURESCOM Participants in P917-GI page 27 (73) WDM technologies for access networks Deliverable 10 wavelength. This solution might fit as well the evolving path of the early PON implementations. Constraints are: Insertion of passive DWDM, provided the optical budget are respected some technologies used in the first generation modems should further work. Replacement of early customers modems unless they were achromatic or wavelength tuneable from the start. Installation of the additional network terminations (Modems, ATU-R whatever they are called). The operators aren’t likely to be willing to manage wavelength dedicated NEs so that this aspect of the technology should be hidden to the plant crews. Two ways are then possible, either self-tuning devices extracting their operating window from the downstream information or to be wavelength insensitive. Limitations brought and to overcome: 3.6.2 At the present early stages, a passive DWDM dedicates one wavelength per drop, so that transmission has to be duplex to work on a single fibre. No coarse diplex or Diplex compatible devices are foreseen till now. Due to the same fact further evolution should be preferred in the time domain, since wavelength allocation will be fixed! Here again some standardising work could ease the availability of a reduced set of solutions. Enhanced Point to Point optical links (SDH) This case has been already used for point to point SDH systems upgrade within several access network operators. Impacting only between the end user and the central office where 2 modems are to be placed, two alternative provider sources can be found: One by the traditional Telecom product manufacturers: who propose expensive but reliable material. And the other by the LAN and computing manufacturers: who propose efficient low cost systems. If the first group will bring reliable products, the costs of the LAN world will be likely to be very attractive, both of them are to be considered provided they make use of the single mode G.652 fibre. No special requirements are to be expressed as long as no interworking between products of different providers but the following features will be particularly valued: The respect of the ITU-T grid: because it induces possibly the lowest cost solutions through the use of common laser sources. A standard remote modem management facility for easy interfacing with the operator’s TMN, which enables him to justify the quality of service offered to the end user, thus avoiding expensive hassles. The scalability is to be provided especially as long as the price per wavelength is as high as it is. The operator and client should be able to upgrade the system to their actual need through plug-in units, with no other hardware operations. DWDM transparency to the operating and maintenance and installing crews: This page 28 (73) 2000 EURESCOM Participants in P917-GI Deliverable 10 DWDM technologies for access networks issue covers mainly the unawareness of technology for the operators. They do not have to know about the wavelength affected to any end user. The system should afford for self start-up & calibration, wavelength recognition and associating of user identity or channel allocations with wavelengths at system level. Other types of technologies may be forwarded transparently through the access network from user site to user site, provided that the global path optical budget is respected and that the DWDM plan is compatible on both links. Examples are FDDI, ESCON, FICON, but not many others are available in telco’s core network only SDH VCs and FR are likely to be accepted. The case of several channels only to have a broader bandwidth hasn’t been shown since it’s obvious. 3.6.3 Step 1: is a self standing one: Diplex working of existing 2 fibre systems Step 2: is the upgrade of wavelength unaware systems into coloured ones and the introduction of DWDM combiners and splitters on both sides of the optical link. Step 3: eventually upgrade of the number of channel. Point to multipoint links (ITU-T G.983 ATM – PON) This architecture and topology is comprehensively addressed in the next section, where all the upgrading schemes and options along with the related issues are given. 3.6.4 Enhanced cable Ring architectures This case refers to cable ring infrastructures where one fibre pair all the way long is dedicated to one customer/node (physical point to point topology). On each customer/node, an active dedicated terminal multiplexer has to be inserted. This implies for the operator in front of each of the customer/node locations to extract one or two fibre pairs from the passing cable, as close as possible to the users’ premises. Of course this is very expensive and of relative poor revenue. Therefore, if several clients could take advantage of the same fibre through DWDM, the cost efficiency would raise. Some obvious constraints are to be met: Presence of several client locations on common spots, else no fibre sharing will be cost effective or possible. Flexible and accurate use of the wavelength allocation plans, passive and tunable add/drop multiplexers, in order to extract from the fibre loop the wavelength needed on each spot It should be a technique affordable both from the cost point of view and technical (low loss and easy to handle) point of view. Most of the constraints on the active elements quoted in case 1 have to be respected. 3.6.4.1 Particular issues regarding CR configurations In order to propose specific ring architectures for DWDM access networks, the following issues have to be addressed: The number of wavelengths within the CR and its usage (e.g. one wavelength per BC, one wavelength shared by a number of BCs, one wavelength per type of service etc.). The type of protection and restoration (electrical or optical) type of signal clients (SDH or others). 2000 EURESCOM Participants in P917-GI page 29 (73) WDM technologies for access networks 3.6.4.2 Deliverable 10 The possibility of using more advanced network elements (e.g. Optical CrossConnects -OXCs-, configurable OADMs etc.) on the basis on their (future) availability. Proposed ring configuration In the configuration to follow, the following assumptions have been made: Only optical distribution networks (ensuring path-protection to the corresponding BC) are considered, The ADD and DROP function should be optical (use of OADM in the nodes) A tributary wavelength, may be either dedicated to a specific BC or shared by a number of BCs. In case that a number of BCs use the same wavelength, those BCs may be either connected to different OADMs (forming a logical "sub-ring" within the CR) or form a separate "sub-ring" connected to a single WADM. The rate of the SDH signals carried by the individual wavelengths of the CR can be STM-1, STM-4 or STM-16, depending on the requirements of the particular BCs, served by the ring. The transmission of different SDH rates by the various wavelengths is possible. OADMs providing BCs with the assigned wavelengths are installed either in an outdoor cabinet or preferably in an indoor location controlled by the local PTT. Each BC is served by an Optical Network Unit (ONU) installed in the BC premises. This ONU must be equipped with an SDH ADM (of the appropriate level n= 1, 4 or 16 depending on the STM rate carried by ?i) and connected to the OADM that provides the BC with the pre-allocated wavelength. OADM (proving optical or electrical protection) Transmission Network Network Manager 1, 2, n Acces Network n 1 STM-1 to 16 BC ONU m BC ONU 1 protection fibre 3 working fibre 2 BC ONU 2 BC ONU 3 WADM sharing ?2 Figure 9 – Proposed final step in the evolution (ring structure) 3.7 Conclusions The demand to justify WDM in the access network is not there yet, but an initial huge deployment of fibre in the access is required. Equipment available is not primarily developed for the access network, so is expensive and is more adapted for point to point topologies, the cost of technology is high even considering the short distances in the access. Components for flexible all optical networks are not yet available. page 30 (73) 2000 EURESCOM Participants in P917-GI Deliverable 10 DWDM technologies for access networks The increase of data traffic brings a paradigm shift in the network structure concept and DWDM will be a key technology as far as transmission is concerned. This fact is a signal for the operators that should plan their infrastructures to enable a DWDM deployment in a smooth manner towards to the all-optical networks. 2000 EURESCOM Participants in P917-GI page 31 (73) WDM technologies for access networks Deliverable 10 4 DWDM Upgrade of existing PON 4.1 Introduction In 1995 the Full Services Access Network (FSAN) initiative was founded to support the co-operation between operators and equipment vendors aiming to recommend and establish a standard fibre access system widely accepted among the involved operators and cost efficient by minimising the custom requirements on the system.. This work leaded to the FSAN recommendation, which was incorporated in 1998 into the International Telecommunications Union (ITU) standard G.983 “BROADBAND OPTICAL ACCESS SYSTEMS BASED ON PASSIVE OPTICAL NETWORKS (PON)”. This standard is the basis to make the attractive ATM PON solution for optical access networks a mass-market technology. First operators (NTT) started with implementation in their access network and many other operators plan to follow in a short time frame. It is assumed that the FSAN PON concept will determine the optical access network for the next years and therefore the fibre infrastructure. Due to this fact network operators intend to build new future proof networks avoiding significant cost expensive changes on the installed fibre infrastructure for many years. It is expected that the predicted data traffic for the next years, driven mainly by the internet access, will cause capacity bottlenecks in the access networks so that upgrade concepts for the installed FSAN PONs are necessary. Because of the high flexibility and the capability to provide very high bandwidth DWDM is a very attractive technology for upgrading existing PON systems. Compared with the possibility to upgrade the PON in the time domain (TDMA/TDM systems, e.g. 2.5 Gbit/s PON), DWDM could be the more flexible and cost effective solution. 4.2 Passive optical Networks (PON) 4.2.1 General PON architecture In PON architectures an Optical Line Termination (OLT) is connected with multiple Optical Network Units (ONUs) by a passive optical distribution network (ODN). This point to multipoint connectivity is established by a passive optical power splitter (1 x N star coupler) serving as passive remote node (RN). In general the OLT is located in the CO and provides the interface between the access network and the service node. ONUs can resides on different locations of the Access Network (FTTH, FTTC, FTTCab). Depending on the ONU’s location, the customers are connected either directly or by a copper or radio drop. OLT Passive Branching Device at the Remote Node 1,5 m DFB Laser Burst - modereceiver m m ONUs m m 1 x 16 Power splitter m m Receiver Fabry-Perot Laser Receiver Fabry-Perot Laser Figure 10 - Example of a schematic PON layout compliant with ITU G.983. page 32 (73) 2000 EURESCOM Participants in P917-GI Deliverable 10 DWDM technologies for access networks Extracted from the ITU recommendation G.983.1, in Table 2 some key parameters describing the Physical Layer requirements are listed. Parameters Architectures Fibre type Optical path loss Requirements FTTH, FTTB/C, FTTCab single mode 1260nm – 1580nm (ITU G.652) 10 – 25 dB (G.982 class B) 15 – 30 dB (G.982 class C) Differential path loss 15 dB Fibre distance between S/R 20 km (OLT) and R/S (ONU/ONT) points Split ratio Restricted by path loss and ONU addressing limits 1:16 or 1:32 with passive splitters (G.983.1) Overall reflection in ODN at -32 dB R/S reference points Downstream Upstream Bi-directional transmission single fibre PON: 1550 nm (1480-1580 nm) 1310 nm (1260-1360 nm) dual fibre PON: 1310 nm (1260-1360 nm) 1310 nm (1260-1360 nm) Nominal bit rate: Option 1 (FTTCab/C/B/H) 155.52 Mbit/s 155.52 Mbit/s Option 2 (FTTCab/C/B) 622.08 Mbit/s 155.52 Mbit/s Point-to-Multipoint Time Division Multiplex Time Division Multiple transmission (TDM) Access (TDMA) Table 2 - Key parameters of Physical Layer of FSAN ATM PON extracted from ITU Recommendation G.983.1 4.2.2 Limitations of Power splitting PON There are some serious limitations due to the passive power splitting approach: 4.3 The optical power intended for one ONU is split between all ONUs, leading to a N-fold power budget penalty. Because the PON bandwidth is shared between all subscribers it is possible to allocate the bandwidth in a flexible and dynamic way. This important advantage has the counter part, that the OLT and all ONUs must work on aggregate bit rate. Since the downstream information is broadcast to all ONUs concerns about privacy will be stimulated. Network diagnostics and fault localisation of the outside plant are problematic. Drivers for the WDM upgrade Despite its great potential the ATM PON has also some serious limitations, explained in before, which will limit the system performance and make the upgrade of such a system difficult. With WDM it is possible to construct wavelength based point to point links on a shared PON fibre infrastructure and therefore to overcome these limitations. The high flexibility, capacity and transparency, also makes WDM an attractive upgrade solution offering additional degrees of freedom due to the insertion of various wavelength to support both entertainment services and more demanding business applications as LAN-LAN connections. In summary the main drivers for FSAN PON upgrading are: 2000 EURESCOM Participants in P917-GI page 33 (73) WDM technologies for access networks Deliverable 10 Increased bandwidth demands by business customers - The main drivers for business customers are similar to the generic ones driving other sectors of the market bandwidth increase. This will in most cases be a bi-directional requirement so that it is possible to construct a virtual LAN connecting various sites of the same company, requiring high levels of security and availability. WDM utilise the PON infrastructure but implements a point-to-point dedicated link. Increased bandwidths demands at single ONUs - Additional to the described requires by specific business customers (e.g. SME) in the early phase of arising capacity shortage the more and more increasing variety of data based services and the steady growing number of residential customers cause the need to upgrade certain ONUs. DWDM could solve the problem delivering bandwidth for this point. Increased gross downstream bit rate demands on the PON – The expected demand for internet browsing and digital TV may bring a bottleneck in the downstream PON capacity, WDM is an option to overcome the problem with minor changes in the network. 4.3.1 Further requirements on future networks: The hard competition in access market will lead to a pressure on the cost of future networks. This means equipment and installation cost as well as the operation and maintenance cost of such a network. The cost pressure will bring demands like: Increased network flexibility - The competition will force dynamic response to customer requests. So the Network operator has to react and to provide the requested services faster and faster. This can be solved only by a flexible and “more intelligent” system and aspects like dynamical consideration of various QoS aspects, rapid provisioning of unexpected traffic demands, and the possibility to quickly provide services and introduce new services will become of great importance. Also for the customers these aspects will have a strong impact on their decisions to select service and network provider. Reuse of the existing fibre infrastructure - This demand covers two parts, network upgrade and optimal exploitation of the existing network infrastructure. Network upgrade must take into account at least the reuse of the fibre infrastructure without significant changes, because the installation of the fibre plant is a main cost factor. Therefore it requires a long life of the fibre plant. So an upgrade will mainly affect the installation of new equipment at the OLT, RN and ONU locations. The second part aims optimal exploitation of the existing fibre infrastructure by means of delivering all services on same fibre infrastructure and combining different networks. Technological advances have opened the possibilities of new services and service types, while fibre-optic advances have permitted improved networks, such as fibre-reinforced cable networks, to evolve. These newer networks signal the possibility of competition between service providers to provide ever-improving services at ever-decreasing costs. Technical remedies can be found for this case and utilise WDM technology. In such a system, passive power splitters are used to convey multiplexed information between the head-end or central office (CO) and the subscribers. At the same time different service providers use the different regions of the optical spectrum, sharing the same physical network but operating with distinct services and even distinct terminal page 34 (73) 2000 EURESCOM Participants in P917-GI Deliverable 10 DWDM technologies for access networks equipment. In a similar way, WDM can further be used to separate distinct services or/and service types (leased lines). This will reduce network, operation and maintenance costs. Therefore the following aspects are of great importance: Combination of broadcast and point to point links. Handling of different QoS aspects. Multiplex / Segregation of services, service types and service providers. Service Segregation/Multiplex - WDM systems can be used for multiplexing different services on the same architecture. For example it has been proposed by NTT [19] to multiplex narrow band ISDN (N-ISDN) services with broad band ISDN (BISDN) services and video distribution services. A two way service uses two different wavelengths for downstream and upstream. A video distribution service needs only one wavelength. In total six wavelengths are multiplexed in the 1.5µm wavelength range. Customer segregation -Customer segregation could be another application for WDM. Here the technology presents interesting advantages, allowing independent overlay networks sharing the same fibre infrastructure and differentiating customer per gross bite rate used as well as by the quality of service required. Upstream capacity upgrade - The maturity and cost reduction in the WDM laser allow the use of WDM solution to be implemented in the upstream direction, in point to point fibre infrastructures, Passive optical networks or CATV networks. Segregation of different QoS - With the announced evolution in optical switching it could be interesting to use it to guaranty quality of service in the IP environment. This may open another field of application for the technology with a dynamic WDM allocation network. Dynamic selection of service provider on optical layer - From a user perspective the freedom of selecting service provider may be regarded as a service. Every service provider is connected to an OLT covering a segment of the AN. Whether the service given by the SP is a broadcast or point-to-point service will decide the means for transmitting the signal downstream from the OLT. Also multicast services can be provided if similar signals from one SP is modulated on a number of tuneable lasers. The ONU are able to select the correct wavelength. 4.3.2 Basic Upgrade concepts WDM PON networks can be divided in two broad categories: Broadcast and Select Networks Wavelength routing networks. The first category - Broadcast and Select Networks is based on the passive power splitter at the remote node. The wavelength insensitivity of the power splitter allows the broadcasting of signals at different wavelengths from the input port to the output ports. Figure 11 shows this behaviour for 1 x 4 splitter. The WDM signal at input port appears at all output ports with a reduced optical power by factor 4. 2000 EURESCOM Participants in P917-GI page 35 (73) WDM technologies for access networks Deliverable 10 1 1, 2, ... , N 2 1, 2, ... , N ... ... 1, 2, ... , N Splitter M=1, N=4 ... ... ... 1, 2, ... , N 4 ... Figure 11 - Wavelength broadcast behaviour of a power splitter Due to the passive optical splitting there is N-fold power budget penalty like in the PS-PON. Additional losses have to be considered, since filters must be implemented to select the allocated wavelengths in the ONUs and in the OLT wavelengths must be multiplexed in the transmitter and de-multiplexed in the receiver. The upstream and downstream separation can be established by a dual fibre PON or by Coarse WDM diplex in the one fibre PON. Because of the broadcast approach the network has the same limits like the PS-PON, except that the transmission bit rate and protocol per wavelength is not shared between all ONUs. Therefore the bite rate and transmission protocol per wavelength is devoted to the allocated ONU. A great advantage of the concept is that the temperature dependence of the power splitter can be neglected, because of no significant impact on wavelength transmission behaviour. In Figure 12 a generic layout of the single fibre Broadcast and Select-PON is given. OLT Passive Remote Node Source Array Mux ONUs 1 x16 Power splitter Router Source Receiver Demux Array Receiver Receiver Source Figure 12 - Generic layout of a single fibre WDM Broadcast and Select PON. The second category - Wavelength routing PON – is based on a passive device at the remote node, known as waveguide grating router, arrayed waveguide grating multiplexer or optical phased array. For simplicity, it will be designated in the following consideration as Wavelength Router (WR). The WR in general is a M (input) x N (output) device, but for our application it is an 1 x N device such as the power splitter. One of the remarkable properties of this device is the capability to direct downstream light from the input (1) to unique output ports (N) as function of optical wavelength. In this direction it works as a wavelength demultiplexer. Figure 13 1, 2, ... , 4 Wavelength Router 1 1 2 2 ... ... M=1, N=4 N=4 4 Figure 13 - Wavelength router as demultiplexer page 36 (73) 2000 EURESCOM Participants in P917-GI Deliverable 10 DWDM technologies for access networks WR designs permit WDM bi-directional links in two ways. In first approach two parts of the 1,5µm transmission window are used for the downstream and upstream transmission. In the second approach, WR works as wavelength router in the downstream direction in the 1.5 m window and as a simple power combiner in the upstream direction for the 1.3µm window. Figure 14 shows schematic layout of the general WDM PON architecture based on Wavelength Routing concept. OLT Source Array Mux ONUs Passive Remote Node Wavelength Router 16x Receiver Demux Array Receiver Source Receiver Source Figure 14 - Schematic layout of the general WDM PON architecture based on the Wavelength routing concept In this architecture, wavelength based point to point connections are established using the properties of the WR, improving the network integrity and the privacy significantly. The power budget penalties of the optical path are lower when compared with a power splitter approach, even if the number of connected output port is high. Further on the optical transmission in downstream and upstream direction is bit rate and transmission format transparent. The architecture offers huge bandwidth and high flexibility. Considering the architecture of FSAN PON and assuming that at least the installed fibre infrastructure is the fixed part of the network, there are 3 points to upgrade the network - OLT, Remote node and ONU. The WDM implementation on the PON topology generally requires multi wavelength devices. The OLT has to transmit and receive different wavelengths, the ONU has to transmit and receive one or more wavelengths. At the remote node a passive multipex / demultiplex and/or a broadcasting component has to be realised and an appropriate approach for allocating and selecting the different wavelengths to the individual ONUs have to be defined. Taking into account the upgrade drivers, four categories of upgrade concepts can be identified: WDM upgrade only of the downstream transmission and maintain the upstream transmission of the PS PON – for single links, single ONUs or the entire PON. Single direction WDM upgrade (downstream) as overlay without affecting the operation of the PS PON – for single ptp links or as a broadcast channel for the entire PON. Bi-directional WDM upgrade as overlay without affecting the operation of the PS PON– for single links, single ONUs or the entire PON. Reuse only of the fibre infrastructure and substitute the PS PON by a new WDM PON. 2000 EURESCOM Participants in P917-GI page 37 (73) WDM technologies for access networks Deliverable 10 4.3.3 Upgrade constraints, restrictions and limitations 4.3.3.1 System constraints The starting situation is an existing FSAN PON. Therefore the ATM PON ITU-T G.983.1 specification must take into account, especially the power budget, the wavelength ranges, possible architectures, transmission methods and the bit rates. Spite of WDM being a good solution to upgrade the downstream direction in some ONUs, there are various system-level difficulties involved with the introduction of WDM dedicated channels in a pre-existent PON system, the most important of which regards the control of the TDMA upstream channel, requiring attention. In the case of a dedicated channel in only the downstream direction the ONU still needs the TDMA signalling information in order to properly transmit to the OLT. In this case there are 2 alternatives: either the control information is also present on the dedicated channel, or the ONU must continue to receive the original broadcast channel solely for the purposes of control. Of course, in the case of bi-directional transmission there is no need for the control information, the link being effectively point-to-point. The first of these solutions poses difficulties as an upgrade scenario since normally OLT equipment does not output the control information on an electrical interface to be used by a second transmitter. That is, there is no intrinsic limitation to the addition of downstream-only dedicated channels but without access to control data, normally not available outside the OLT, the TDMA channel cannot be implemented Another potential problem is the fact that the data destined for the dedicated link must be output from the ATM switch on a separate interface from the OLT. This should normally be a simple question of switch configuration, since there are usually several OLTs connected to the same switch. If however there are not enough free ports on the switch then there is a bottleneck which can only be overcome by increasing the number of ports. 4.3.3.2 Technological constraints The use of DWDM in the upgrade of a PON system can only be achieved if a number of technological conditions are met, since the original system must continue to function after the upgrade. The authors identified possible constraints in the following parameters, that should be visited before a DWDM upgrade option: 4.3.3.3 Increased losses due to the new components installation. Crosstalk between channels, present because wavelengths are not perfectly separated in WDM components. Back-reflections from components, the figures given for the return loss of optical components are generally very small, and it is unlikely that difficulties will arise in this area. DWDM devices are temperature sensitive passive, requiring special attention. Economic Constraints To upgrade an existing FSAN PON with WDM overlay techniques naturally contain the installation of new components or equipment in the system. In this connection some economical aspects have to be taken into account that cause any kind of expenses or expenditures to be avoided or at least to be minimised: page 38 (73) 2000 EURESCOM Participants in P917-GI Deliverable 10 4.3.3.4 DWDM technologies for access networks Advance payment (in the planning and introduction phase) to be minimised with a decision for a system concept that allows a low-priced introduction for the first customers or services and that can nevertheless be e.g. modularly extended with little working and installation expenditures and interruption free service provision. The cost for components and equipment is the most important part to note. So the aim is to use as few as possible separate components. The decision about the kind of new or reuse of existing components at the three relevant locations OLT, Remote node and ONU/ONT has to be made in comparison of the component cost, the installation expenditures, the technical properties and quality advantages. The scope of changes and new equipment at the ONU has to be minimised above all because of the comparatively large number of these locations per PON system. Personnel costs for the installation of the equipment and for putting the system in operation and for Operation and Maintenance have to be planned. To limit these personnel payments the system once put in operation has to run automatically without constant or permanent need for maintenance or servicing. Service life of the technique/equipment at least for the time until investment depreciation and the future capability in view of the compatibility, functionality and further upgrade possibilities of the technology. Financing of investment for the WDM upgrade to be balanced with adequate secure income of new offered applications and services. Operator Constraints The necessary extensions of the system features like increase of capacity (with chance to offer more services) and better exploitation of fibre infrastructure are desirable improvements wanted by the operator. But the upgrade of a running system cause interruptions for installation of the new components. In the starting time system problems can often cause service works and reduction of functioning. So customers connected to the FSAN PON who usually enjoy the smooth work are annoyed. Therefore an everlasting change of the network like installing discrete components step by step according to the present effort is not the idea of the operator. More expensive but avoiding the described disadvantages is to include a single integrated component in one advance step. Other constraints concerning the operator are related to economical aspects like service security, service life, simple OAM and to build a flexible and future secure platform. With view of the whole operator network it is desired to come closer to an uniform technology over all hierarchical steps to an all optical network. 4.4 WDM Upgrade solutions 4.4.1 Solutions for business customers The solutions with the aim to increase the bit-rate of certain existing ONUs for business customers fall into two main categories: downstream only, and bi-directional dedicated links. There is also an additional driver to provide a dedicated channel between two ONUs present on the same PON. 2000 EURESCOM Participants in P917-GI page 39 (73) WDM technologies for access networks Deliverable 10 Since it is likely that the first users of WDM channels on a PON will be the business customers it is necessary to envisage solutions which provide very few (even just one) dedicated channel – these will be the entry points of WDM in the PON, leading eventually to widespread use among all sectors. Here we will present only one example. 4.4.1.1 Downstream dedicated links in a single-fibre PON The addition of a small number of downstream channels to a single-fibre PON is complicated by the fact that the upstream signals at 1.3µm use the same fibre as the downstream signals, both broadcast and dedicated, around 1.55 µm. Hence the behaviour of wavelength selective components must also be specified at 1.3µm, or extra components must be used to bypass critical parts of the network. The first solution is not an option at present since no commercial WDM components specified for the 1.55 µm window have a defined behaviour at 1.3 µm. The solution, shown in Figure 15, introduces extra losses, but can nonetheless be viable in cases where the initial power margin is sufficient. In fact the total extra loss on the broadcast channel is 6.45 dB, that on the upstream channel is 3 dB, and the worst case total internal loss on the WDM channel is (7.8 + M 0.95) dB where M is the total number of WDM channels. FP Tx 1.3 Rx 1.5 Rx 1.3 Tx 1 DFB 1:N FP Tx 1.3 Rx 1.5 ... ... 0 Tx 2 DFB ... FP Tx 1.3 Tx M DFB 1 Tx 0 DFB 2 M Rx 1.5 ... Figure 15 - Addition of WDM channels ( -fibre PSPON It is also possible to use the AWG to multiplex the broadcast channel, thus saving an add/drop component. However this entails that the broadcast channel suffers additional losses due to two AWGs in series, and it is unlikely that the power margin of a typical PSPON will allow such an architecture Figure 16. page 40 (73) 2000 EURESCOM Participants in P917-GI Deliverable 10 DWDM technologies for access networks FP Tx 1.3 Rx 1.5 0 ... FP Tx 1.3 Rx 1.5 Rx 1.3 1:N 0 0 Tx 0 DFB FP Tx 1.3 Rx 1.5 Tx 1 DFB ... DE MUX M ... ... MUX FP Tx 1.3 Tx M DFB Rx 1.5 1 Figure 16 - Architecture for the addition of WDM channels to single-fibre PSPON using AWGs. 4.4.1.2 Bi-directional dedicated links 4.4.1.2.1 Bi-directional upgrade based on discrete components As has been pointed out in the chapter system constraints, unless the necessary MAC protocol is also present on the dedicated downstream channel it is not possible to control the upstream TDMA channel of the ONU. One solution is the addition of a dedicated upstream link each time a downstream link is installed. For the two-fibre PON case this is simple – it is enough to implement the same architecture as for the downstream direction. For the single fibre PON a bi-directional link can be achieved using the architecture of Figure 17. FP Tx 1.3 Rx 1.5 Rx 1.3 1:N 0 DFB Tx 1.5 Tx 1 DFB Rx 1.5 Rx 2 DFB 1 2 Tx 0 DFB Figure 17 - Upstream and downstream channels using cascaded discrete components in a single fibre PON In Figure 17 the grey OADMs could be red/blue band WDMs to allow the use of multiple broadcast channels in the downstream direction. The architecture allows the addition of as many pairs of channels as the system power margin can support, with the same number of components as for the downstream-only case. 4.4.1.2.2 Bi-directional upgrade based on AWG wavelength routers The same approach can be used here as for the discrete component case. The 1.3µm/1.55µm WDM components on the ONU side of the splitter are replaced by OADMs and each ONU uses 2 channels of each AWG. The additional losses here are the same as in the discrete component case, approximately 1.2dB Figure 18. 2000 EURESCOM Participants in P917-GI page 41 (73) WDM technologies for access networks Deliverable 10 FP Tx 1.3 Rx 1.5 Rx 1.3 1:N 0 DFB Tx 1.5 Tx 1 DFB Rx 1.5 Rx 2 DFB Tx 3 DFB DFB Tx 1.5 Rx 1.5 Rx 4 DFB Tx 0 DFB Figure 18 - Upstream and downstream channels using AWGs in a single fibre PON Another approach involves use of the cyclic properties of the AWG. the upstream channels can use a wavelength 1 free spectral range away from the downstream wavelength. The upstream and downstream signals use exactly the same path, and are separated only at the terminals. The OADM on the ONU side of the AWG in the remote node is therefore relocated to the OLT Figure 19 so the losses are the same, but twice the number of channels can be offered. FP Tx 1.3 Rx 1.5 Rx 1.3 1:N 0 Tx 1 DFB DFB Tx 1.5 Rx 1.5 Rx 2 DFB Tx 3 DFB DFB Tx 1.5 Rx 1.5 Rx 4 DFB Tx 0 DFB Figure 19 - Upstream and downstream channels using the cyclic properties of AWGs in a single fibre PON 4.4.1.2 Conclusions From the calculation of the impact on single- and two-fibre PSPONs of the addition of downstream dedicated WDM channels it appears that it is possible to add channels gradually with losses comparable to, if not better than, those currently available using AWG routers. In the case of the single fibre solution the power margins of the original broadcast and upstream channels need to be fairly high. This is true for channel numbers up to about 16, after which it will probably become attractive to remove the splitter altogether and provide each ONU with a dedicated channel. The advantage of this approach is that initially, when perhaps there is some doubt about the rate of growth of WDM applications, one can provide connections as requests occur without having to predict the total number of connections to be provided in the future. Making all WDM overlay channels bi-directional, achieved by adding one or two components to each optical path, avoids the difficulties encountered in TDMA control when only downstream WDM channels are implemented. The extra loss amounts to about 1.2dB per channel, but the number of ONUs with dedicated channels is halved, for a given initial power margin, with respect to the unidirectional case. Another possible solution is the use of the cyclic properties of AWGs which allow the same page 42 (73) 2000 EURESCOM Participants in P917-GI Deliverable 10 DWDM technologies for access networks optical path to be shared by two different wavelengths. The losses are the same as for the previous cases, but the number of ONUs that may be connected is not affected. The current costs of the components required for the various upgrade scenarios are such that the solution based on discrete approach is more economical, though it is probable that the costs of routers will fall more quickly than those of OADMs, for example. 4.4.2 Solutions for increased demands at single ONUs In previous section “Solutions for business customers” it was discussed how to upgrade downstream or bi-directional capacity of single dedicated links to business customers. Another driver described in Section “Drivers for the WDM upgrade” is the increased bit rate demand at single ONUs. From technical point of view there is no difference concerning the possible solutions. So this section refers to the results of investigations made there. 4.4.3 Solutions for increased gross Downstream bit rate In the upgrade of downstream capacity at the PON system it is thinkable to assign additional dedicated wavelengths to every ONU (wavelength per ONU) or to provide a new service using one wavelength to all ONUs (wavelength per service). It is possible to make use of the broadcast and select architecture, routed wavelengths architecture or a combination of both. 4.4.4 Wavelength per Service: It is possible to use this technique both for single or two fibre PONs. Because in twofibre PONs the transmission in upstream and downstream direction is managed on two separate fibres in the downstream fibre only the 1.3µm is occupied. So the free 1.55µm window can be used for transmission of upgrade wavelength(s). If one wants to provide a broadcast service to all ONUs an additional wavelength or a WDM channel can be inserted in the downstream fibre using a 1.3/1.5 WDM. This wavelength will be transmitted to all ONUs connected to the passive power splitter. A coarse WDM at the ONUs detours it to a new receiver. Upgrading a single fibre PON is of course more complicated. The original downstream TDM channel and the upgrade channel both must use the 1.55µm window. So additional to the wide band 1.3µm/1.55µm WDMs add/drop WDMs at 1.55µm have to be added at OLT and ONUs. This will make the upgrade much more expensive then the two-fibre solution. The ONU has to select multiple wavelength, this factor increase the complexity and cost of the ONU. 4.4.4.1 Wavelength per ONU: By wavelength per ONU upgrade there have to be wavelength-based point to point links between the OLT and each ONU. So each ONU has one dedicated wavelength channel on the same fibre infrastructure without affecting the installed fibre cable. These allow different applications by assigning the wavelength to specific issues like increasing of up- and downstream bit rates, multiplexing (Adding or Segregation) of services (with special format), segregation of different operators or segregation of different QoS (e.g. IP traffic with QoS classes and different services level agreement). This upgrade could be done with two possible solutions: 2000 EURESCOM Participants in P917-GI page 43 (73) WDM technologies for access networks Broadcast and Select concept Wavelength routing concept Deliverable 10 The broadcast and select architecture to provide separate wavelength channels to all ONUs, upgrade wavelengths are multiplexed by AWG and broadcast to all ONUs. Discrete filters or add/drop WDMs must be inserted at the receivers to let only the destined wavelength pass to the ONUs. In the Wavelength routing concept, the PON architecture uses power splitter as combiner for upstream transmission and wavelength router for downstream capacity upgrade. 4.4.5 Solution for increased bi-directional capacity demand A way to increase the bi-directional capacity of a two-fibre PON is to add an WDM overlay of permanent wavelength paths in the region of 1.55µm. This system concept gives a capacity increase without changing the fibre network structure. The existing TDM/TDMA PON system will not be infected. 1,3 µm 1,3 µm 3 db 1,3 µm 1/N 3 db ..... 1,5 µm 1 3 db 1p 1 3 db 1p ..... i A W G 1/ N 3 db ip 1,3 µm 1,5 µm 1,5 µm A W..... G 1/ N 1,5 µm 3 db i 3 db ip Figure 20 bi-directional upgrade with WDM PON overlay concept [20] To separate the original TDM/TDMA PON from the new overlay at the central office, at the ONUs and to detour the Power splitter at the Remote Node coarse 1.3/1.55 µm WDMs have to be inserted. The key element used for multiplexing, demultiplexing and routing functions of different wavelength signals is the arrayed waveguide grating (AWG). At the OLT as well as at the Remote Node wavelength routers AWGs) have to be inserted. Because of the frequency selective routing function of the AWG cost effective receivers can be used. This also considerably improves the protection against interference and eavesdropping. The individual ONU source wavelength must match with its related wavelength of the wavelength router port, so technical problems like temperature dependency of both components (ONU-laser, wavelength router) and controlling and stabilisation of the ONU Laser sources have to be solved. 4.4.6 Future networks In future various upgrade possibilities are thinkable that only use the fibre infrastructure of the former PON system. So the power splitter at the remote node can be replaced by a router (e.g. AWG or active device) and to all ONUs point-to-point wavelength connections can be realised. More then one wavelength per ONU for much higher demands are thinkable too. The most freedom and flexibility will be provided by dynamic wavelength access. The access to wavelengths for transmitting and receiving between the individual ONUs and the OLT are allocated in dynamic and flexible way. In this case a procedure and a related Medium Access Control (MAC) protocol must be implemented to provide a page 44 (73) 2000 EURESCOM Participants in P917-GI Deliverable 10 DWDM technologies for access networks resource management and handle the wavelength allocation between OLT and ONU. Further on, tuneable components such as tuneable lasers and filters are required. These will lead to more complex and expensive network. These solutions are further steps of not upgrading then more exchange of PON for much more powerful systems and depend also on development of future services and demands as well as availability of commercial and mature components and trade off between cost and performance. 4.4.7 Component issues All the components necessary for the architectures presented before are currently available. Table 3 shows the prices, as of June 1999, of DWDM components necessary for the addition of dedicated downstream channels in the downstream direction only. Table 3 - Prices (June 1999) of WDM components WDM device & Quantity Channel spacing 1–9 10 - 24 100 Wide band WDM 1.3- ~200 EUR 1.55µm OADM 200GHz 936 EUR 831 EUR 719 EUR OADM 400GHz 811 EUR 718 EUR 619 EUR 8-channel AWG 200GHz 8044 EUR 7306 EUR 6382 EUR 8-channel AWG 400GHz 7047 EUR 6398 EUR 5588 EUR Based on this information, it is possible to extrapolate the cost per port of the wavelength routers and hence to estimate the total component cost, for a given number of added channels, for the discrete component and AWG based solutions for downstream-only dedicated channels. However, the OADM devices have been available for some time and their cost seems to have bottomed out, whereas the router technologies (AWGs, for example) are relatively new and it is foreseeable that their cost will continue to decrease for some time. Clearly the results of Figure 21, although calculated for the downstream-only case, are valid for the bi-directional case since the additional components necessary are the same. Figure 21 - Cost per channels added for 1 fibre (left) and 2 fibre PONs with downstream-only dedicated channels The composite component necessary for the reception of the broadcast channel even by ONUs with dedicated links could perhaps be integrated into a single device, and in this case the approach would be more attractive. Otherwise this particular architecture performs badly from the point of view of the power budget and it would probably be more sensible to go directly to the bi-directional solution. 2000 EURESCOM Participants in P917-GI page 45 (73) WDM technologies for access networks Deliverable 10 The figures quoted by manufacturers suggest that crosstalk will not be a great problem in these upgrade scenarios, at least in the cases anticipated for the solutions for business customers in which there are relatively few dedicated channels present on the PON. However, in more evolved networks in which there is a large number of dedicated channels serving both business and residential customers it is likely that crosstalk will be more of an issue. 4.4.8 Management and protection aspects The management and protection in the PON procedures and technique are fully covered by the FSAN G.983 specification. In fact from the system point of view the protection is implemented by a parallel PON systems over the same infrastructure. The upgrade scenarios using WDM to provide dedicated point to point links in the downstream direction suffer from the problem of the distribution of OAM information in the PON. While in new WDM PONs it is possible, but very difficult, that the problem be overcome at the hardware level in the OLT, in non-green field situations the two solutions are either to always install bi-directional links, or else to maintain the original downstream broadcast channel even in the ONUs with a dedicated link. The second option further reduces power margins in the system, particularly in onefibre PONs, and necessitates the use of several additional components per ONU. The first option, on the other hand, is the most attractive because it is likely that the first users of dedicated channels be business customers interested in high bandwidth twoway traffic. The independence of the FSAN-PON and the WDM systems simplifies the OAM function, only need to certify that both systems are compliant to the specification as far as survivability is concerned. This case could be more complex in a WDM scenario where the wavelength channel is routed from the core network without any conversion in the edge near the OLT. In this case special attention should be taken to the optical element that makes the interface with feeder section in the access network in order to certify its protection capabilities. The scenario of dynamic wavelength allocation in the access network is the final step in the upgrading scenario, more than an upgrade option it is a target architecture. In this case we are talking about an advanced system that needs a complex wavelength MAC protocol to assign the different wavelengths. The complexity of such a MAC protocol requires the existence of OAM information flows between the network elements that should integrate all management and survivability mechanisms embedded on it. 4.5 Evaluation of the upgrade solutions and migration path Corresponding to each upgrade driver there is a set of upgrade solutions. One has to evaluate each of the upgrade solutions with respect to the set of the following criteria: Cost, Technological Feasibility, Upgradability, Flexibility and identified constraints. Each evaluation should provide the “best” solutions. It is also quite likely that a “trade-off” between different criteria might be required to be able to determine the best solutions. The process is repeated for each driver. The next step remains in page 46 (73) 2000 EURESCOM Participants in P917-GI Deliverable 10 DWDM technologies for access networks determining the time variation of upgrade drivers. A good deal of uncertainty is involved in this task and therefore involves some guessing and “speculation”. When done, this corresponds to arranging different drivers in a chronological order, and might lead to several different scenarios. Arranging the best solutions along the same scenario should provide one/several migration paths. An evaluation might again need to be performed to determine how good the “best solution” for a given upgrading step is and how smooth the migration provides it, when one takes into account the migration from an earlier upgrade step. This quantitative evaluation requires a detailed analysis with large set of parameters. So for instance it must take into account a fixed FSAN reference model, assumptions on expected services and bit rates and large database with technical parameters and cost of the required components. Only in this way the different upgrade solution are comparable in cost and performance. Of course the evaluations is “trade off” between cost and performance. Beside the technical and economical aspects, the strategy and philosophy of network operator will have strong impact on the migration path. In the following table (Table 4) a qualitative evaluation is given. It takes into account the identified drivers and upgrade concepts as well as the WDM PON Networks discussion before. The large “X” indicates a promising solution, the squares marked slight grey indicates a possible solution. One may identify different migration paths. One is based on the broadcast and select approach the other on the wavelength routing approach. For the first demands (it is assumed this will be business customer demands) an upgrade can be started with discrete components. But with growing demands integrated components will be required, which promise an easier handling and installation, compactness, higher assurance, easier fault diagnostic and at least lower costs (of curse only these components will be commercially available in a sufficient quantity). 2000 EURESCOM Participants in P917-GI page 47 (73) Deliverable 10 WDM technologies for access networks D e d ic a t e d some O NUs a ll O N U s e n t ire P O N L in k s Inc re se d b a n d w id th D r iv e r s de m a nds by b u s in e s s c u s to m e r s Inc re a se d Inc re a se d b id ir e c tio n I m p le m e n ta tio n d o w n s tr e a m b it a l b it r a te o f a d d itio n a l d o w n s tr e a m b it b id ir e c tio n a l b it r a te d e m a n d s a t de m a nds B roa dc a st r a te d e m a n d s o n Inc re a s e d gros s r a te d e m a n d s o n Inc re a se d gros s s in g le O N U s a t s in g le c ha nnne l th e P O N th e P O N D if f e r e n tia tio n o f s e r v ic e s a n d p r o v id e r s Inc re a s e d n e tw o r k f le x ib ility H a n d lin g o f d if f e r e n t Q o S a s p e c ts ONUs Upgrade concepts E x p e te d c h a n g e s o f d ir v e r s w ith th e tim e B r o a d c a s t & S e le c t W D M upg ra de o nly o f the X w a v e le n g th p e r O N U do wns tre a m tra ns m is s io n a nd m a inta in the ups tre a m d is c r e te c o m p o n e n ts tra ns m is s io n o f the e x is ting X X W a v e le g th r o u tin g PS PO N in te g r a te d X c o m p o n e n ts B r o a d c a s t & S e le c t S ing le dire c tio n W D M upg ra de X w a v e le n g th to a ll X ONUs (do wns tre a m ) a s o v e rla y witho ut d is c r e te c o m p o n e n ts a ffe c ting the o pe ra tio n o f the ( a s p tp lin k ) W a v e le g th r o u tin g PS PO N in te g r a te d ( w ith s p e c tr a l (a s p t p lin k ) c o m p o n e n ts d is c r e te c o m p o n e n ts W a v e le n g th r o u tin g B i-dire c tio na l W D M upg ra de a s O v e r la y o v e rla y witho ut a ffe c ting the in te g r a te d c o m p o n e n ts o pe ra tio n o f the P S P O N B r o a d c a s t & S e le c t O v e r la y w a v e le n g th p e r ONU s lic in g ) X X X X X X X X X d o w n s tr e a m C o m p o s ite P O N X d e d ic a te d p tp lin k s lin k s c o m p o s ite w ith R e us e o nly o f the P S P O N fibre infra s truc ture a nd s ubs titute the W D M P O N ba se d on w a v e le n g th r o u tin g P S P O N by a n ne w W D M P O N (future ne two rk s ) W D M P O N ba se d on b r o a d c a s t & s e le c t d e d ic a te d b id ir e c tio n X p tp lin k s w ith f ix e d w a v e le n g th a llo c a tio n b id ir e c tio n a l W D M b r a o d c a s t w ith f ix e d w a v e le n g th s e le c tio n w a v e le n g th b a s e d F u tu r e n e tw o r k s X d y n a m ic X X r e c o n f ig u r a b le Table 4 Solutions to the identified drivers (slightgrey – possible solution, darkgrey/X– promising solution) 2000 EURESCOM Participants in P917-GI page 48 (73) Deliverable 10 4.6 WDM technologies for access networks Guidelines for Network planners In the competitive telecommunications environment, network planners face a common problem; how to upgrade networks in a profitable way and minimising risk, due to the uncertainty in service demand predictions and huge needed investments. There are several other factors in addition which will influence the specification for the new network. The planner can then make a shortlist of competing architecture solutions, each of which meet the basic specifications. The economics of these solutions may then be compared. At this stage variations in the introduction scenarios may be discussed to further improve the profitability of a particular solution. An effort is made to group all the factors and requirements which a planner has to keep in mind while introducing the WDM technology in upgrading the existing PON systems. In the first stage a planner has to make some serious evaluation whether he wishes to upgrade the existing network using WDM. The main factors influencing this decision are shown in the flow chart (Figure 22) and are divided in four broad categories: Existing equipment - The extent to which the existing equipment has to be integrated in the upgrade solution with a new technology like PON. Traffic - A comprehensive forecast has to be made to predict the expected traffic and traffic types in the access network. A corresponding revenue earning forecast may also be needed. The new system - What properties are expected from the new system. Future scalability and upgradability are one of the major issues. How manageable are the network management systems? Is backward compatibility important, etc. The other influences - As the name suggests, these involve the decisions made by the others outside the planner’s reach. These include the decisions made through internal/external politics and business strategy actions taken by the competitors. Once that is done then he has to consider the most compelling drivers for the required upgrade. In previous chapters a detailed description of different drivers has been performed. In the Flow chart Figure 22 such factors are grouped into nine broader categories to be able to give an overview. Some of these categories therefore might be overlapping: Before we presented the possible solutions for various stages in upgrading of PON networks with WDM, and several possibilities for corresponding migration paths are described. Every driver may be fulfilled through several different technological concepts as shown in the Figure 22. For every concept there may also be several technological implementations described as solutions in the Figure 22. E.g. a WDM upgrade of the downstream transmission may be realised both through broadcast & select and wavelength routing. Further each of these two routing mechanisms can be achieved through either discrete or integrated components etc. The choice of a particular solution has its pros and cons as described in earlier chapters. At this stage an evaluation has to be performed by the planner. The evaluation criteria can include both quantitative parameters such as Payback period and Net Present Value. For details of these methods see f.ex. [9]) or/and qualitative parameters such as upgradability and flexibility of a network solution. The inputs to this evaluation may include the costs (equipments and new installations f.ex.of extra fibers) and expected revenues from these new investments, which in turn may require 2000 EURESCOM Participants in P917-GI page 49 (73) WDM technologies for access networks Deliverable 10 a more accurate traffic forecast. A negative outcome of this calculation/evaluation would imply that another concept or another implementation should be chosen. Driver Concept Alt. Solutions Increased bandwidth demands by business customers •Existing topology/network •Existing cables/fibres etc. Traffic •Present vendor relationship •Competitors decision •Traffic forecasts •Future revenues •Future traffic mix/types Choice of WDM upgrade concept Increased downastream bit rate demands at single ONUs WDM upgrade only of the downstream transmission and maintain the upstream trenamission of the existing PSPON Broadcast & select Other influences Existing Equipment The new system •Compatibility with existing systems •Equipment availability •Scalability •Future upgradability •Protection ? •Maintenance costs Increased bidirectional bit rate demands at single ONUs Implementation of additional broadcast channel Single direction WDM upgrade (downstream) as overlay without affecting the operation of the PSPON Wavelength routing Wavelength routing overlay Increased gross downstream bit rate demands on the PON Increased gross bidirectional bit rate demands on the PON Differentiation of services and providers Bidirectional WDM upgrade as overlay without affecting the operation of the PSPON Broadcast & select overlay Composite PON Increased network flexibility Handling of different QoS aspects Reuse only of the PSPON fiber infrastructure and substitute the PSPON by a new WDM PON WDM PON based on wavelength routing Future networks Evaluation Criteria Rank solutions f.ex. in terms of shortest PP and highest NPV Payback period Evaluate each solution Traffic and revenue estimates Net Present Value Upgradability etc. Equipment costs Figure 22 - Simplified methodology to provide guidelines for technology introduction 4.7 Conclusions The limitations observed in the FSAN Power splitting Passive Optical Networks, relating to increasing bit rate demands caused by foreseen drivers, have been shown. The high flexibility and the capability to provide very high bandwidth and possibility of using the existing infrastructure makes WDM a very attractive technology for upgrading existing PON systems. Various technical and architectural concepts for WDM upgrading of existing PONs have been introduced, the architectures have been described and analysed, advantages and disadvantages have been discussed. The following technical concepts that meet the drivers are considered to be the possible WDM upgrades of FSAN PON systems: Downstream WDM upgrade as overlay without affecting the operation of the PS PON– for single ptp links or as a broadcast channel for the entire PON. Upgrading only the downstream capacity and maintain the upstream transmission of the existing PS PON. Bi-directional WDM overlay upgrading without affecting the operation of the PS PON. The drivers and solutions have been joined together derived from the technical and architectural concepts. As the beginning step it is foreseen to upgrade a few business customer ONUs with dedicated links. So the solutions for this driver have been page 50 (73) 1998 EURESCOM Participants in P917-BOBAN Deliverable 10 WDM technologies for access networks discussed most detailed and the physically practicability of WDM for this start scenario was demonstrated The solutions have been evaluated resulting in possible migration paths depending on development of drivers and customer demands. Aspects of management and protection have been investigated. Various circumstances are influencing the practical realisation of WDM upgrade of PONs in the access. An important aspect is the commercial and cost effective availability of components like wavelength routers (AWGs) and transmitters (for instance LASERs, Super LEDs). Another constraint is the suitability of wavelength routers for use in the outside plant (at the remote node). Consequently an essential property of wavelength routers is temperature independence. It is mentioned that today no perfected, commercial products and components are available. To introduce the solutions in the network the costs of mentioned components must decrease. It is expected that through the practical operation of WDM in the Metro network component prices will fall and so in future the possibility of introduction in the access part will become more economically. As described in “Guidelines for network planners” a lot of concrete complex conditions for the special situations have to be took into account and a detailed analysis and a trade off between costs and performance have to be made. 2000 EURESCOM Participants in P917-BOBAN page 51 (73) WDM technologies for access networks Deliverable 10 5 Convergence of DWDM Networks 5.1 Reference Optical Metropolitan Access Architecture The network convergence through the DWDM technology towards all optical networking shall be examined through a reference General network considering the three domains, core, metro and access as presented in the first chapter. This model is partly based on the highlights of the documented studies [15] – [18]. In this model emphasis is given on the network functional description and technology requirements rather, than on detailed description of architecture. It is important that the reference architecture is simplified as much as possible while keeping at the same time a functional structure which is capable of describing both the future DWDM access networks and migration to those from the current networks. The legacy access and metropolitan area networks can be identified as parts of this reference network. Figure 23 shows a high level network architectural view consisting of three segments: core network, metropolitan network and access network. For the rest of this document we use this topology for illustrative purposes, however the principles are fundamental to a range of possible topologies. The metro network has a ring topology interconnecting Access Nodes (AN) and Hub Nodes (HN). An access node typically hosts switching facilities (LEX, ATM, IP) and is designed to be highly configurable. Hub nodes serve as the interface between the metropolitan access network and the backbone network. The access network interfaces directly the customer premises and is responsible for delivering and collecting traffic. Some traffic aggregation may take place in the distribution network as well. The topology of the distribution network can be tree, bus, ring or a combination of them. Legacy active rings, e.g. SDH rings, can also be accommodated. IP Backbone Network ATM Backbone Network HN HN HN AN AN ADM HN/AN METROPOLITAN NETWORK AN ADM ACCESS NETWORK ADM SDH ring CORE NETWORK Passive tree Passive bus AN = Access Node HN = Hub Node = Passive Splitter or Add/Drop = Customer Premises / Aggregation Point Figure 23 - A reference metropolitan access network architecture. The reference metropolitan access network covers an area of approximately 250 to 1250 km2, serving about 500 to 2000 high-end customers. Note that a head-end that serves thousands of residential customers could be a “customer” of this network. The page 52 (73) 1998 EURESCOM Participants in P917-BOBAN Deliverable 10 WDM technologies for access networks feeder ring circumference (metro network) is expected to be in the 40-150 km range, with customers located no more than a few km from an access node. In densely populated areas, it may be necessary to deploy a number of smaller rings. However the optimal size of the reference network and the optimal size and the placement of the feeder ring within that region will depend on factors such as location of existing fibre, geographic constraints, existing serving areas, customer location etc. 5.2 Functional description of DWDM access node A schematic view of a DWDM Access Node (AN) is shown in Figure 24. The AN supports various functionalities with respect to wavelength by-pass and wavelength add-drop between the metropolitan (feeder) and access (distribution) network segments. The possible functional scenarios are: Fibre by-pass - In this case a feeder fibre carrying transit traffic simply bypasses the AN without interfacing any equipment in the node. No DWDM functions occur in the AN. Wavelength by-pass - This refers to the case where some wavelengths of a feeder fibre bypass the AN while one or more other wavelengths of the same feeder fibre are dropped in the AN. This implies the use of wavelength multiplexers / demultiplexers and OADM. ACCESS NETWORK DMX DMX MUX METROPOLITAN NETWORK `Fixed W avelength Connection Wavelength Converter DMX Distribution Fibers MUX Configurable Opt. Add/Drop ATM switch or IP router etc. Fiber Bypass DMX MUX MUX Wavelength Bypass Feeder Fiber Ring Figure 24 - Multiple feeder fibres enter an access node. Both fibre and wavelength bypass, and configurable add/drop are supported. Some wavelengths are terminated on an electronic switch, while others are routed transparently to the distribution network. For the added-dropped wavelengths the following possibilities exist: Fixed wavelength interconnection - A fixed wavelength from a feeder fibre is dropped in the AN and directly routed to a distribution fibre, and vice-versa. Configurable wavelength interconnection - This scenario uses configurable OADM to select the wavelength which is dropped from a feeder fibre in the AN.The selected wavelength is directly routed to a distribution fibre. Wavelength conversion - This solution overpasses the problem of using the same wavelength in both network segments by making use of wavelength converters. A selected wavelength is still dropped from a feeder fibre in the AN using an 2000 EURESCOM Participants in P917-BOBAN page 53 (73) WDM technologies for access networks Deliverable 10 OADM, however it is also converted to an appropriate wavelength which is vacant in the distribution fibre. 5.2.1 Use of OXC instead of OADM All the above functions could also be realised using an Optical Cross-Connect (OXC) in the place of OADMs. This choice would depend on techno-economic reasons based on the number of wavelengths used. 5.2.1.1 Optical-electrical-optical conversion Finally, in case the AN hosts a switch (ATM, IP router, etc) o-e-o conversion could not be avoided for wavelengths carrying relevant services. In this case DWDM convergence has the meaning that the e-o interfaces of the switch should be able to transmit at selected wavelengths in order to match with the available wavelength slots on either side of the AN. 5.2.2 Network level interfaces The ring topology has gain substantial importance in networking architectures, regardless of the type of signal (electrical/optical), service or even network segmentation. This is basically due to its inherent characteristics related to the topology robustness towards network failures which is based on simple protection schemes and to the simplicity of node conception (only one direction for information flow – only one input side and one output side) In the following sections we will address the problem of feeder ring interconnection with either ring or PON access architectures, , regarding traffic flow and protection issues under the agreed scenario for optical DWDM networking migration towards the access network segment. 5.2.2.1 Ring-Ring interconnection Ring-ring interface can assume several forms, depending on the desired functionality and perspective of evolution. The interface nodes within AN are “principal” nodes in interfacing with the above ring (feeder). All traffic incoming/outgoing the distribution ring passes these nodes. Two strategies are possible with different implications in traffic routing and network protection following a failure (Figure 25): In single hubed rings all traffic incoming/outgoing the ring passes the same node; there is therefore only one common node to both rings (feeder and distribution). Two options regarding traffic routing can be taken: Pure hubbed traffic distribution or mixed hubbed traffic distribution. page 54 (73) 1998 EURESCOM Participants in P917-BOBAN Deliverable 10 WDM technologies for access networks WDM Ring Feeder AN OADM OADM Distribution In/Out traffic Inner Ring traffic Mixed traffic Dist Hubbed traffic Dist OADM a) Single Hubbed Ring Feeder AN HUB 1 HUB 2 OADM OADM Distribution OADM WDM Ring Normal state Protected state Out traffic In traffic b) Double Hubbed Ring Figure 25 - Protection in a) single hubbed rings; b) double hubbed rings Several options exist for the traffic routing in double hubbed rings The straight forward approach consists on defining one of the hubs as a primary hub and the other one as backup, being traffic routed as in a single hubbed ring only by one hub at a time. Another option consists on independently routing of incoming/outcoming traffic using one particular hub for each traffic direction. Again here, in a pure hubbed solution the inner ring traffic should be routed via a hub node and this should be done using the least occupied hub, which can be allocated dynamically or pre-established according to the existing or estimated knowledge on traffic pattern distribution. Also other mixed traffic distribution strategies can be used, being the hub nodes almost exclusively used for ring interconnection. 5.2.2.1.1 Protection strategies Protection strategies depend on the already established strategies for each ring individually. It is assumed here that protection is carried out in the optical domain as protection can also be implemented by the client, in the electrical domain such as in the case of SDH clients. In the latter case the optical transport should be considered unprotected, as mixed multilayer protection strategies involving optical and client electrical protection are not currently being implemented and add an extra level of complexity to distribution-feeder interfaces. For the feeder, which is a multi fibre ring, optical sections can be protected with a n:m fibre shared strategy, or then a n+n dedicated fibre strategy, on the exact option depending on the actual traffic distribution and priority level in the feeder. Protection against hub failures is only possible for double hubbed rings in which case regardless of existing routing policy, in case of failure of one of the hubs the interconnection becomes singled hubbed. 2000 EURESCOM Participants in P917-BOBAN page 55 (73) WDM technologies for access networks Deliverable 10 5.2.3 Possible functional/physical implementation of ring interconnections 5.2.3.1 OXC/Configurable OADM functionality The use of OXC and configurable OADMs in the interconnection nodes is the final step towards the full optical networks. Both elements are identical in terms of functionality, as they allow several or one input/output trunk ports (fibre DWDM carrying signal ports), and one or more single channel input /output ports, being individual channels routed by means of optical switches (switching matrix) to the desired output port. OADM is better suited to distribution-feeder ring interconnection. In this case the dropped /inserted optical channels (whose number can range from one optical channel to a few), are used for interconnection of rings, supporting traffic from/to feeder ring. OXC for the ring interconnection is recommended in the case of high traffic flowing in the ring interface and the requirement of high flexible assignment of channels and high demand for direct wavelength routing. From our study the use of configurable OADMs and OXCs are far from real deployment because only fixed OADMs are available in the market with limited functionality and monitoring capabilities, which are necessary to implement a transparent network. 5.2.3.2 OADM/EADM(EXC)/OADM This implementation can be useful in a transition phase as a fast straight forward solution for interconnecting pre-existing rings with OADM nodes. Since this is not a pure optical solution, the traffic can be further manipulated in the electrical domain which may present some advantages like routing or aggregating traffic to legacy rings (eg SDH rings) with a co-located node. Fj Feeder Fj Feeder Fj Feeder k k D E M U X Fj Feeder OADM Control M U X Fi Dist. O/E/O k EADM k O/E/O Fi Dist. OADM l Add/Drop Fi Dist. a) Fj Feeder Fj Feeder k k k k Fi Dist. Filter b) Legend: Control Fi access Fi Dist. Coupler c) k - WL of optical Ch in WDM signal l - WL of optical add/drop Ch Fi - fiber in the distrbution ring Fj -fiber in the feeder ring i,j,k,l- order number; from 1 to I,J,K I - total nº of fibers in the distribution ring J- total nº of fibers in the feeder ring K - total nº of optical chs in WDM signal L - Max nº of add/drop optical chs Figure 26 - Possible implementations of Access Nodes: a) reconfigurable optical node; b) Reconfigurable OEO node: c) single channel optical add/drop; OADM -optical add drop multiplexer; EADM - electrical add drop multiplexer; page 56 (73) 1998 EURESCOM Participants in P917-BOBAN Deliverable 10 5.2.4 WDM technologies for access networks Feeder/SDH access ring interconnection Interconnection with a SDH distribution ring is a particular case because the single wavelength allocated per fibre in a SDH ring cannot be directly mapped on a DWDM channel, i.e. one cannot simply connect the optical output of a SDH card on a DWDM MUX/DEMUX/OADM. This is because the wavelength on the SDH side does not normally coincide with a wavelength of the WDM grid, and moreover the power, linewidth and wavelength stability are not consistent with those of an optical channel within a DWDM comb. Therefore adaptation of the optical SDH card signal to the optical transport layer is needed. In the case of low rate SDH distribution rings (STM-4) in which traffic aggregation is desirable previous to feeder ring interconnection, an intermediate electrical stage EADM – might be needed and the multiplexed signal will then make use of an interface to feeder as mentioned before. 5.2.5 Feeder/FSAN PON interconnection As presented before the DWDM introduction on the PON allows service discrimination, user discrimination and superposed PONS on different wavelengths but on the same infrastructure. Here we will present some alternatives to root wavelengths from the access to the metro network. The interface of the PON to the feeder ring depends on the type of PON and the type of the services being offered. A distribution PON for CATV for example would require a very simple interface, while an interactive service providing PON, with protection requirement for some users or services, would require a more complex interface. In addition to the required functionality, optical amplification would be most likely required. 5.2.5.1 Distributive service PON For distributive service provisioning, e.g. CATV distribution, the functionality required in the AN is the filtering of the correct wavelength carrying the service, which must take place before the PON power splitter which is either co-located in the AN or connected to it by a PON feeder cable, Figure 27 (a). The filter in the AN is tuned (it can be a fixed filter also) to the wavelength (j) carrying the distributive service and extracts it from the feeder, while leaving the remaining channels undisturbed. This channel is then distributed by the PON infrastructure to all ONUs in the PON. 2000 EURESCOM Participants in P917-BOBAN page 57 (73) WDM technologies for access networks Feeder Deliverable 10 Feeder Feeder AN Fk feeder 1…N filter =j AN Fk feeder 1… N AN Fk feeder 1…N BBFilter BBFilter 1… j 1…j AWG 1:M 1:M M 1 onu onu onu j M 1 onu onu onu onu onu onu onu onu a) j 1 b) c) Figure 27 - Distributive DWDM PONs: a) shared; b) c) discriminate users/services 5.2.5.2 Interactive services PON This case requires upstream traffic handling. The upstream traffic depends much on the type of interactive service, but for most services it is considerably less than the downstream. This is particularly true for BB interactive services such as VOD and BB Internet access, as in the upstream one mostly has sparse, low bitrate, control information. In the case of equal distribution of downstream /upstream traffic, e.g. videoconferencing, or POTS, OADM functionality is required in the feeder ring interconnection. This is also true for asymmetric traffic, however in this case some aggregation of upstream traffic could be performed for traffic generated in the several virtual PONs superposed on the same physical infrastructure. The figure bellow illustrates three types of DWDM interactive PONs and their interconnection to feeder ring, for interactive service provisioning. In Figure 27 (a) one wavelength j is used for both the downstream and upstream traffic. In the AN the traffic is discriminated by a directional coupler and the interface with the feeder is realised using an OADM. Case (b) is similar but the interface with the feeder is now done via several wavelength add/drop channels (one for each coloured PON) and therefore DWDM DEMUX/MUX is required here. If the number of wavelength channels in the PON is high, then an OXC could be used and the upstream/downstream DWDM signals could be trunk inputs of the cross-connect. The same applies to the last case (c) of AWG PON. page 58 (73) 1998 EURESCOM Participants in P917-BOBAN Deliverable 10 WDM technologies for access networks Feeder Feeder Feeder AN Fkfeeder 1…N OADM j add/drop =j AN Fkfeeder 1…N OXC/ OADM 1…j j Directional coupler 1…j 1…j AWG 1:M 1:M onu j M onu onu onu onu onu onu onu onu a) j 1 M 1 1 OXC/ OADM 1…j BB Directional coupler j onu onu AN Fkfeeder 1…N b) c) Figure 28 - Interactive DWDM PONs: a) shared; b) c) discriminate users/services 5.2.6 Wavelength routing policy The independent wavelength assignment in the different domains of the optical network (core, feeder and distribution) is an important functionality for the full optical network concept implementation. This can be based on several mechanisms. The more evident is the use of wavelength conversion functionality, allowing reusing of the same wavelength allocation grids and components for all the segments of the network. This, allied to allocation reconfigurability within rings, also permits direct wavelength routing (without changing the wavelength) from the core to the access, Figure 29, and vice-versa. However this is still not for near future implementation, unless it is based on O/E/O transponders. Another mechanism is the use of a pre-defined set of rules for wavelength allocation in the different network segments, associated with network planning and configuration. One of the possibilities is to use space division multiplexing in the feeder ring, Figure 30, reserving different fibre rings according to traffic flowing direction (to the access segment, to the core segment, inner ring traffic), therefore allowing the use of similar wavelength sets for interfacing with upper and lower rings, also allowing direct wavelength routing from/to core to/from access. In the same fibre rings traffic should refer to the same network segment and be routed for maximum wavelength reuse within the fibre ring. Distribution rings would use for inner ring routing only a subset of available wavelengths. 2000 EURESCOM Participants in P917-BOBAN page 59 (73) WDM technologies for access networks Deliverable 10 1 CORE 1 C/F interface Inner ring WL transposition to free WL channel FEEDER Access Node 1 DISTRIBUTION 1 Inner ring WL transposition to free WL channel Figure 29 - Direct wavelength routing from core to distribution, with the same ITU grid, by means of transposition within incoming ring of existing wavelengths in order to free the required wavelength and therefore transparently route all way through destination, ie, without converting it. This situation is shown in Figure 30 where core ring RcoreK directly interfaces the feeder ring Rfeeder1 in the Core/Feeder interface node. Within feeder ring, Rfeeder 2 to I are reserved to inner ring traffic while RfeederI interfaces with Rdist of distribution. Within the distribution ring a part only of the wavelength channels is used. Rcore 1 Rcore k CORE Rcore K C/F interface IC Inner Rfeeder 1 Rfeeder 2 Rfeeder i Rfeeder I FEEDER Access Node Inner ID DISTRIBUTION Rdist. Figure 30 - Wavelength allocation policy with by space division multiplexing, with different fibre Rings within each network segment interfacing only particular fibre rings on the neighbour segments. (Rcore, Rfeeder, Rdist) If less performing technology is to be used in the different network segments, one possibility is to use different grid spacing but anchored to the same reference, the grid being narrower in the core and becoming wider towards the access (e.g. 50GHz, 100GHz and 200GHz), Figure 31. For the direction from the core towards the access it would be possible to route sets of wavelengths grouped for the same network segment but for different same level rings (the C/F filter extracts a set of wavelengths, with the red ones directing to ring B and the violet ones to ring A - these are interleaved.) For routing to a particular ring, wavelength allocation has to respect this level ring grid spacing. In the inverse direction, at the interconnection of rings, high performance page 60 (73) 1998 EURESCOM Participants in P917-BOBAN Deliverable 10 WDM technologies for access networks components namely in terms of crosstalk rejection should be required in order to narrow the channel spectrum. This would also allow the upgrade of network domains and reuse of components on domains of the same network segment. Direct wavelength routing from core to access network is also possible in this scheme. 1 CORE To Rings A&B 1 C/F interface A FEEDER B Access Node 1 DISTRIBUTION 1 Figure 31 - Direct wavelength routing, different ITU Grid 5.2.7 Aggregation of traffic in the distribution network for WDM applications In this section we analyse the solutions based on DWDM technology for concentrating traffic in the optical network. The analysis uses some of the arguments and solutions presented in a detailed published study [15]. Then a comparison of these solutions is made and it is shown that some solutions have very good performance in concentrating traffic and reducing the number of switches in an all-optical network. The access network is by default a section carrying different services and low bit rates according to customers’ demand. To increase network efficiency, a metro section appears with a main function of traffic aggregation. As optical switching is far from the deployment phase, some kind of traffic aggregation is required in the access network to justify DWDM application. 5.2.7.1 Non-transparent solution for traffic aggregation in the distribution network, Example : Wavelength per customer In this solution one wavelength is dedicated to each customer/ONU. The various electronic services requested by a customer are multiplexed in an SDH multiplexer at the customer premises, as shown in Figure 32. For example, services from customer 1 are multiplexed on a single wavelength and eventually delivered to a port on a SDH switch within the access node. The same type of traffic (e.g. ATM traffic from customer 1 and 2) from different customers is multiplexed together in this switch and directed to the appropriate switch. Statistical multiplexing of traffic typically occurs in the switch or router and the traffic is sent out to the feeder network to the corresponding hub node. As an example, FR traffic is delivered to the hub node that interfaces with the FR backbone network. The disadvantage of this architecture is that it requires multiplexing equipment at customer premises and several network switches in each access node. 2000 EURESCOM Participants in P917-BOBAN page 61 (73) WDM technologies for access networks Deliverable 10 Figure 32 - Wavelength per customer architecture. From [15] 5.2.7.2 Transparent solution for aggregating traffic in the distribution network, Example: Wavelength per service Figure 33 - Wavelength per service architecture. From [15] The second strategy to aggregate traffic involves allocating resources in “simple” units, such as full wavelengths. For example, an STM-1 ATM connection and an STM-1 IP connection may be each allocated a full wavelength rather than being more efficiently multiplexed. Furthermore a wavelength is dedicated per electronic service type requested by the customer. The solution is shown in Figure 33, where customer 1 would require two wavelengths while customer 2 three wavelengths. The types of services per customer may be changed by tying the corresponding wavelengths to a switch in the access node. The great advantage of this architecture is that it eliminates the need for multiplexing equipment at the customer premises. However, this solution increases the complexity on the ONU level and makes inefficient use of wavelengths for relatively small data rates such as STM-1 or below. In such cases it might be useful to aggregate the traffic from multiple customers before feeding the traffic to the distribution network. 5.3 Use of DWDM for the reduction and concentration of network switches The aim of the all-optical network concept is the possibility of routing a wavelength transparently from the access network to the core network. The architecture for “wavelength per service” in such a network will help to reduce further the number of page 62 (73) 1998 EURESCOM Participants in P917-BOBAN Deliverable 10 WDM technologies for access networks network service switches concentrating the ANs, which is an objective for all operators. This can be illustrated as follows: One option is to place an ATM, IP and FR switch in each AN as shown in Figure 34. A better option however is to have the wavelengths carrying different types of traffic not terminated all in the same AN, but have for example IP wavelengths transparently routed to the next AN where an IP router is located. Similarly, wavelengths carrying FR traffic are routed to the AN where an FR switch is placed. Thus the ATM switch in the first node, the IP router in the second node and the FR switch in the third node can be shared by the customers of all three nodes, Figure 35. This configuration results in the significant reduction of network switch equipment and correspondingly leads to lower maintenance and upgrade costs as well as less power and space needs. However, as mentioned earlier, more wavelengths are needed because the traffic is carried in the feeder network without first being multiplexed in a switch. Figure 34 - Several switches are placed in each access node. Traffic is multiplexed as soon as it enters the access node, goes through O-E-O conversion and is further multiplexed/converted at the hub node Figure 35 - Switches are placed only in the hub nodes. Traffic is routed transparently (specified by double lines) to the feeder ring through the access nodes. Much fewer switches are needed at the expense of more wavelengths. However the same number of wavelengths exits the hub node switches in either architecture. 2000 EURESCOM Participants in P917-BOBAN page 63 (73) WDM technologies for access networks 5.4 Deliverable 10 Network migration concepts towards all optical networks Although the immediate advantages of introducing DWDM networks is evident, care must be taken in the course of planning to ensure a smooth transition and reuse of existing infrastructure and technology, and a sufficient degree of flexibility at the same time that allow future upgrades. Currently existing networks in the core, metropolitan and regional network segments are based on SDH technology, either in ring-ring or in meshed-ring architectures, supported by optical fibre infrastructure. As traffic demand growths resulting on increased capacity demand, upgrade via new SDH systems of the same capacity will soon lead to fibre exhaustion. Going up in capacity of SDH systems implies substituting all the node equipment to the new hierarchical SDH level (STM-4 toSTM-16 and to STM-64). One possibility to immediate upgrade an existing network and also free occupied fibres or spare existing ones, is to directly map existing SDH rings on DWDM by colouring the rings, thus creating space for immediate upgrade of capacity on the same infrastructure. 5.4.1 Migration Phases The following phases are proposed for a smooth network migration, assuming an existing hierarchical network which may be of a ring-ring or meshed-ring architecture: A two level ring-ring SDH topology is assumed. Introduction of transponders and OADMs in both network levels and straightforward mapping of existing SDH rings on wavelength rings. Introduction of OXC on the top level ring. Capacity upgrade by stacking of rings. New ring deployment for network expansion. Installation on new generation DWDM equipment in distinct fibres while there is fibre available. Grouping of same generation equipment in same physical infrastructure rings, reusing older generation equipment in areas with less capacity demand. Single- W P ADM T T W P ADM 1 ADM OADM OADM OADM 1 To SDH WDM 1- 16 1 Figure 36 - Direct mapping of MS-Spring into DWDM rings page 64 (73) 1998 EURESCOM Participants in P917-BOBAN Deliverable 10 WDM technologies for access networks O O XC O XC XC Lower level rings OADM OADM OADM Lower level ring traffic Figure 37 - DWDM architecture -introduction of OADM and OXC in both levels of existing hierarchy. 5.4.1.1 Advantages The main advantages of this approach are: 5.4.1.2 Use and optimisation of already deployed optical fibre infrastructure by introducing DWDM. Use of an architectural concept quite similar to SDH, in architecture, requirements and functionality, therefore being familiar to the network operator and allowing better assimilation. Architectural inherent protection facilities of utmost importance in view of the high level of traffic aggregation carried in one single fibre. Allows gradual, follow-up technological upgrade of existing network as well as future network expansion and capacity upgrade. Commercial availability of the required optical functionality for the introduction phase (OADM and transponders). A reasonable number of used wavelengths, e.g.16, both in technical and economical terms. This is a moderately low value as DWDM systems are commercially available with a much higher wavelength density. In spite of this, not all the wavelengths will be initially used. Spare resources will remain available to account for future upgrades and optical protection allowance. Disadvantages The following disadvantages can be pointed out: Either coloured optical interfaces for SDH equipment or transponders will be necessary for the implementation of DWDM-SDH rings. Both solutions are expensive for the initial phase. The first option seems to be more suitable for new equipment installation, however unless tuneable lasers are used a great diversity of spare parts is necessary and the interface must stick to a fixed wavelength grid which may turn the installation of future generation WDM systems more difficult. The second option is in all phases expensive but is quite flexible and assures independence of terminal equipment. Necessity of DWDM optical amplifiers, wideband to cover the used DWDM signal and with flat gain within the covered band. Necessity of optimisation of first approach by straightforward mapping of SDH rings on DWDM rings in order to minimise the number of used OADMs. This 2000 EURESCOM Participants in P917-BOBAN page 65 (73) WDM technologies for access networks Deliverable 10 can be achieved by a different effective number of nodes for each of the several overlapping co-located rings. As a result of using SDH, MS-SPRing protection, only half of the total capacity is used, the other half being reserved for protection. This feature will initially be an advantage but as capacity needs increase and the number of available planned wavelengths diminishes, it will soon turn out to be a disadvantage. New multilayer protection schemes, including also protection in the optical layer, must be considered. In particular the diversity of optical routes available due both to the ring double node interconnection in the lower hierarchical level and to the two level hierarchy, allows the implementation of other survivability mechanisms, similar to pre-planned route restoration. Fast evolving DWDM components and systems market, adding up to reduced standardisation, results in foreseeable incompatibility between the purchased systems of today and their future upgrades, demanding for some caution on network planning trade-offs and initial investment. The 16 wavelength DWDM system may be more suited to the medium term future needs and far cheaper than higher density systems, which may became completely obsolete within the lifetime of the first option. OXC systems are not yet commercially available. Non existence of efficient optical network supervision and management, allowing either integration or interworking with existing supervision and management systems. 5.5 Time frame for the DWDM application 5.5.1 DWDM network element maturity The deployment of DWDM in the metropolitan and access segments, and especially their convergence towards a wavelength transparent network is critically dependent on the maturity of a number of key network elements. Based on the status of technology, a possible time frame for DWDM solutions could be derived. However, this timeframe must be considered as indicative since no precise estimation can be made on when a novel network element will really become available. Today, only solutions based on fixed OADMs are feasible. This first generation of OADMs can be used to realise the fixed wavelength interconnection scenario. All the other DWDM convergence scenarios require that specific DWDM network elements become available. Configurable OADM are expected to emerge first, enabling the scenario of configurable wavelength routing. A possible timeframe for this second generation of OADM is end of 2001. Tuneable transponders allowing O-E-O wavelength conversion are also expected in the same timeframe, relaxing the requirements for a common wavelength allocation scheme between the feeder and distribution network. First OXCs may be expected in year 2001, however based on transponders (O-E-O conversion). Finally, full interoperability using wavelength converters, optical switches or wavelength routers looks more into future, not foreseen within the next 2 years. page 66 (73) 1998 EURESCOM Participants in P917-BOBAN Deliverable 10 5.5.2 WDM technologies for access networks Network Management Systems maturity. Network management systems have been a limiting factor to the rapid growth of any lightwave system. Because hardware advances tend to lead software development by 1 to 2 years, new lightwave systems usually become available before complete NMS packages are written. Today, network management systems for DWDM systems tend to be element managers, monitoring the performance of the network elements rather than the wavelength channels themselves. Nevertheless, convergence of DWDM in metro and access networks would require an optical layer management system capable of providing functions such as fault management, including OTDR, configuration management, performance monitoring, and restoration. These functions are today supported by the SDH layer, therefore it seems inevitable for OAM to migrate from the SDH layer to the optical layer. For example, today performance monitoring systems consist of detection of optical power outside a specific range, which triggers an alarm or an automatic protection switching module. There is no way at this time for performance monitoring systems to actually estimate the quality of the client signals. However, to implement transparent all optical DWDM connections, there is a need to monitor the performance of lightwave signals without knowledge of what the signal is carrying. In other words, performance monitoring is required at the analogue optical layer in addition to the digital signal layer. Significant deployment of DWDM in the metropolitan access network should not be expected before suitable optical layer monitor and management systems are developed, which provide operators with full OAM capabilities. 2000 EURESCOM Participants in P917-BOBAN page 67 (73) WDM technologies for access networks 6 Deliverable 10 Conclusions In our study we considered the introduction of DWDM technology into the access networks, to fulfil the increasing bandwidth demand. We focused on two different application areas, where we expect that DWDM will first emerge: Connecting large business customers, and DWDM upgrade of PONs (because we foresee a bandwidth demand that can only be met by using DWDM) Furthermore, we investigated the aspect of network convergence, because we believe that the introduction of DWDM in the access will bring us closer to realise the full potential of the all optical networks. When introducing DWDM in the access incumbent operators should consider existing infrastructures and devise a migration path. Our study indicates that European operators adopted similar fibre solutions to serve BCs, and also to connect residential customers. Nearly all of them make use of optical access networks based on a physical ring topology and SDH transmission. Business customers are connected through point to point or logical ring connections. Some operators are using PONs to serve residential users and SMEs. This creates a rather common “present situation” or starting point from which future networks, such as DWDM access networks will evolve. In spite of the commonly claimed “booming” increase of BC capacity demand, most BC connections still appear to be in the STM-1 range, or below. Since these demands can be cost effectively served by proven legacy technology a considerable further increase in demand is needed to boost DWDM deployment in the access network. For the time being DWDM deployment in the access is solely motivated by bandwidth upgrade both for the customer drop and for links higher up in the network. Single provider NE systems as well as homogenous transmission formats are being employed. Wavelength per service type application of WDM in the access networks emerges with the competition between service providers. Solutions with fibre sharing schemes in either ring topologies or tree branch PON structures are also likely to emerge. In the first case, passive or tuneable OADMs will be installed all the way along the fibre path, while in the second topology, a single location may be retained for the passive DWDM. Presently, however, only fixed wavelength implementations are used with passive ADM. The responses to our RFI and the assessment of the state of the art in DWDM systems show, that DWDM isn’t yet mature nor cost competitive for general introduction in the access, not even for large business users. Nevertheless, we see two trends that are major drivers of the deployment of DWDM in the access: The necessity to rapidly link all LEXs with a broadband service access node, such as a DSLAM, and provide further bandwidth upgrading capacity. The emerging demand for VPNs linking all the premises of a business, (Nowadays provided through a separate fibre or through a dedicated PVC in an ATM network). DWDM technology allows the creation of dedicated wavelength channel VPNs giving complete independence from user to user. page 68 (73) 1998 EURESCOM Participants in P917-BOBAN Deliverable 10 WDM technologies for access networks At present no ‘telecom quality grade’ DWDM access systems are available. Nevertheless, DWDM systems developed for LANs may fit in for limited application fields. Under these conditions, transmission oriented DWDM systems can be suitable for access links, but for temporary use, only, since there are major issues to be resolved before a large scale roll-out can take place. These are the following: Interworking of DWDM equipment from different manufacturers through the optical interface. Although by now most manufacturers are using wavelength grids that are compliant to the ITU standard, this isn’t sufficient to ensure full compatibility. Full remote supervision of user end NE: configuration, service provisioning, fault reporting, performance monitoring. The following conclusions can be drawn from the responses received to the RFI: Though DWDM offers huge transmission capacities and supports a wide variety of input signal formats (conditions that are essential for the migration of DWDM to the access part of the network), it suffers from limited configurability and lacks networking functionality: Available transponders emit at pre-selected wavelengths. No dynamic wavelength allocation is possible, yet. OADMs are fixed (not dynamically configurable) devices and as such they can not be used for the dynamic interconnection of DWDM branches and rings. Optical cross-connects (that would enable the implementation of all-optical fully configurable networks), are still in the stage of development and their deployment is believed to be a few years away. Available DWDM systems at most support point to point connections multiplexing several channels between a LEX and an end user on a single path or on a ring (thus providing a 1+1 redundancy with path diversity which is required for large business clients). Ultimately the migration of DWDM to the access network is going to be governed by cost. Cost is a much more complex issue in the access networks than in the core, and renders necessary the techno-economic comparison of DWDM and non-DWDM solutions (such as xDSL and SDH/FITL). The responses to the RFI we have circulated gave very little information regarding hardware/system costs. One aspect, however, that should be taken into account is that almost all vendors offer “metro-oriented” WDM solutions that should allow the development of WDM access network at a lower cost. Our investment study revealed that today the deployment of DWDM technology in the access cannot be justified purely on a cost basis. Another field of application of DWDM in the access that holds a great potential is PON upgrade. The FSAN Initiative OAN work group considers the upgrade of APON using DWDM. A wavelength allocation plan is already under discussion. We also see a high potential in using DWDM technology to upgrade existing PON systems. Flexibility can be increased and very high bandwidth can be provided by the re-use of the existing infrastructure and adopting DWDM. A number of technical and architectural concepts exist regarding how to do this. These are described in detail, analysed and compared, highlighting the advantages and disadvantages of each of them. 2000 EURESCOM Participants in P917-BOBAN page 69 (73) WDM technologies for access networks Deliverable 10 We consider the following DWDM upgrades feasible and meaningful for FSAN PON systems: downstream DWDM upgrade as overlay without affecting the operation of the PSPON – for single point-to-point links or as a broadcast channel for the entire PON. Upgrading the downstream capacity, only, and maintaining the upstream transmission of the existing PSPON. Bi-directional DWDM overlay upgrading without affecting the operation of the PSPON. We foresee that the first, most immediate demand will be to upgrade a few business customers' ONUs with dedicated links. We have demonstrated the feasibility of the solution to this scenario. The large-scale deployment of DWDM in the access is hindered by the lack of key components, such as wavelength routers (AWGs) and transmitters (for instance LASERs, Super LEDs), that are suitable for use in the access network. Application in the access network is especially demanding, because a controlled environment can not be provided, or only at a very high cost. Therefore components need to be thermally very stable. (Providing a controlled environment for components in the core network is not that critical.) Furthermore, the cost of the components is also critical, because unlike in the core networks, they are not shared by that many customers, and also a much larger number of them is required. Finally, we provide guidelines for network planners. However, each particular situation needs to be analysed in detail, and the decision should be made considering the trade off between cost and performance. As outlook for the following generation of DWDM access networks will have beyond this characteristics like reconfigurability and dynamic allocation of wavelengths depending on capacity demands per area and time period. A special attention was paid to the interesting aspect of the convergence of DWDM access and core networks. We have identified the access node that interconnects the metropolitan and access network domains as the most critical element of DWDM convergence. This is due to the fact that it interfaces various technologies and architectures, is cost sensitive, must be flexible to accommodate new services / demands and has a large impact on OAM. We looked at the ways wavelength routing can be realised in the access node. A number of options exist employing a variety of DWDM functionalities, ranging from fixed wavelength add/dropping to configurable optical cross-connection and optical conversion. These solutions exhibit vastly different flexibility features depending on the actual DWDM network elements employed, special attention was taken to the different degrees of wavelength protection and restoration. We have found that the immaturity of the technology poses a major barrier towards convergence of DWDM networks. Simple forms of transparent wavelength routing can be readily implemented using available fixed OADMs, nevertheless at the expense of limited flexibility. More flexible solutions allowing the selection of wavelengths to be added/dropped and effective use of wavelength grids among the different network segments require more sophisticated network elements such as configurable OADM, OXC and wavelength converters, which are not available for the time being. page 70 (73) 1998 EURESCOM Participants in P917-BOBAN Deliverable 10 WDM technologies for access networks DWDM convergence would also require the evolution of today’s DWDM products towards systems more focused on access/metro applications. They should justify their cost effectiveness against conventional solutions based on SDH/ATM and should also provide full OAM capabilities. Management systems should evolve from the presently available simple network element monitoring to wavelength channel monitoring, and realise in the optical layer all the management functionalities provided today by the SDH layer. From optical networking point of view, DWDM convergence is envisaged to have a large impact on network structure and planning because it: Introduces new networking concepts such as “wavelength per customer” and “wavelength per service”. Affects the planning of networks by introducing wavelength routing in the access nodes, allowing the concentration and reduction of electronic switches. Introduces new services and applications based on wavelength provisioning to customers. The major fields of application of DWDM convergence are: Wavelength VPNs capable of supporting customers’ native signal formats, Wavelength leasing, Wavelength on demand and Storage networks. A prerequisite for most of these applications is a large capacity demand exceeding the STM-4/16 level per customer or aggregation point. Such high capacity is still rarely demanded nowadays, however the liberalisation of the telecommunication market might help creating a significant amount of such demands. First users of these DWDM services are expected to be very large business customers and small operators. In summary, we consider DWDM convergence as a mid-future issue. Convergence should be expected when the technology becomes sufficiently mature and capacity demands justify the use of DWDM. We expect that DWDM will migrate gradually from the core to the metro and then to the access network. However, when the barriers will be overcome DWDM convergence will constitute a real paradigm shift in optical networking, changing the perception we presently have of optical networks. 2000 EURESCOM Participants in P917-BOBAN page 71 (73) WDM technologies for access networks 7 Deliverable 10 References 1. “Opportunities for WDM technologies in access network. An overview from EURESCOM P614”, Daniel Lecrosnier et all, NOC ’99 Netherlands Delft, Volume I, p134. 2. “Photonic Network Technologies and Related Components”, Hideo Kuwahara, NOC ’99 Netherlands Delph, Volume II, p1. 3. “Metropolitan Transport for Broadband Data Services”, Rafi Gidron, Doug Green, NOC ’99 Netherlands Delph, Volume II, p9. 4. “Resource Allocation in WDM OMS-SPRing Architectures with Arbitrary Demand Patterns”, Jason Spencer, A.Antonopoulos, L.Sacks, J.J.O’Reilly, NOC ’99 Netherlands Delph, Volume II, p41. 5. “The Optical Layer. What Strategy for the Future?”, S.Ferguson, P.Gigghito, Optical Networking, p26, 11th Tyrrhenian Workshop, St. Margherita Italy. 6. “Future Optical Metropolitan Area Networks”, S.M.Gemelos, D.Wonglumson, I.M.White, K.Shrikhande, L.G.Kazovsky, Optical Networking, p36, 11th Tyrrhenian Workshop, St. Margherita Italy. 7. “DWDM: In for the Short Haul?”, Andrew Cray, Data Communications, June 1999. 8. “Planning of Full Optical Network”, EURESCOM P709. 9. “Evolution Towards an Optical Network Layer”, EURESCOM P615. 10. “Building and Operating Broadband Access Network”, EURESCOM P917. 11. “Metro Optical Networks: Metro DWDM and the New Public Network”, Pioneer Consulting, July 1999. 12. “Local area WDM solutions: How fast, how far, how soon?”, Lightwave, Volume 16, Issue 9. 13. “Architectural Advantages of WDM Technology in Access Networks”, Jane Simmons, LEOS ’98 IEEE. 14. “On the value of wavelength add/drop in WDM rings”, Topics in Optics and Photonics, Vol. 20, June 1998. 15. “Architectural principles of Optical Regional and Metropolitan Access Networks”, A.M. Saleh and J. M. Simmons, J. Light. Techn., Vol. 17, No 12 (1999). 16. ACTS project “METON”. 17. Wagner R. et al., “MONET: Multiwavelength Optical Networking”, Journal of Lightwave Technology, Vol.14, No.6, June 1996, pp. 1349-1355. 18. Simmons, J.M.; Saleh, A.A.M.; Wasem, O.J.; Caridi, E.A.; Barry, R.A. Optical Fibre Communication Conference, 1999, and the International Conference on Integrated Optics and Optical Fibre Communication. OFC/IOOC '99. Technical Digest , 1999 , Page(s): 178 -180 vol.2 page 72 (73) 1998 EURESCOM Participants in P917-BOBAN