5G Ultra-High Capacity Network Design With Rates 10x LTE-A Mischa Dohler CTTC, Spain IEEE ComSoc Distinguished Lectureship Tour Texas/Arizona USA November 2012 © 2012 Mischa Dohler Slide template is copyright of BeFEMTO. All rights reserved. 1 IEEE ComSoc Austin/Texas Chapter DLT Drivers & Vision © 2012 Mischa Dohler 2 IEEE ComSoc Austin/Texas Chapter DLT Smartphones Cause Capacity Crunch home office home 80% Indoors 20% Outdoors © Thierry Lestable, Sagemcom © 2012 Mischa Dohler 3 IEEE ComSoc Austin/Texas Chapter DLT The Resulting Capacity Demand Prediction Total worldwide mobile traffic by 2020: more than 127 exa‐bytes (10^18) more than 30 times increase compared to today Total mobile traffic (EB per year) 140.00 Yearly traffic in EB 120.00 100.00 Europe Americas 80.00 Asia 60.00 Rest of the world W orld 40.00 20.00 2010 2015 2020 Source: IDATE © Thierry Lestable, Sagemcom © 2012 Mischa Dohler 4 IEEE ComSoc Austin/Texas Chapter DLT Are we prepared to meet this capacity demand? IMT‐A 4G capacity targets [ITU‐R M.2133]: 2.2b/s/Hz for downlink and 1.4b/s/Hz for uplink in urban deployment supported rate is thus maximal 100Mbps/km2 (500m cell size; 40MHz) Capacity needs (reality check): peak density of 8,000 people/Km2 of which only 10% subscribe to the broadband service of which only 20% require access at the same time each requiring 5Mbps 8,000 X 10% X 20% X 5Mbps = 800 Mbps/Km2 4G requirements … … were short by a factor of 10 yesterday … are short by a factor of 50 today … will be short by a factor of 100 tomorrow © 2012 Mischa Dohler 5 City Average People/Km2 Athens 5,400 Madrid 5,200 London 5,100 Barcelona 4,850 Warsaw 4,300 Naples 4,100 Berlin 3,750 Paris 3,550 IEEE ComSoc Austin/Texas Chapter DLT Addressing Capacity Needs in the Past Increase in capacity over past decades: Martin Cooper: overall: doubled every 30 months over past 100 years million‐fold increase in capacity since 1957 Breakdown of these gains: 5 x PHY; 25 x spectrum; 1600 x reduced cells, 5 x rest Reduced Cells MHz Ratio Gain/Cost Most Important! Breakdown of (estimated) cost: Reduced Cells MHz PHY [HOT ‐ TOP 3 IEEE DOWNLOAD IN APRIL 2011] Mischa Dohler, R.W. Heath Jr., A. Lozano, C. Papadias, R.A. Valenzuela, "Is the PHY Layer Dead?," IEEE Communications Magazine, vol 49, issue 4, April 2011, pp 159‐165. © 2012 Mischa Dohler 6 IEEE ComSoc Austin/Texas Chapter DLT Trend confirms: Link‐Layer won’t help © Avneesh Agrawal, Qualcomm © 2012 Mischa Dohler 7 IEEE ComSoc Austin/Texas Chapter DLT Time for Major Design Changes To facilitate the exponential capacity increase, dramatic changes to how we design cellular systems are needed! … oh, and, yes, spectrum is not a problem and power efficiency neither … © 2012 Mischa Dohler 8 IEEE ComSoc Austin/Texas Chapter DLT Going 5G: Change in Design Paradigms 1. Differentiate clearly between outdoors and indoors design 80% of time 20% of time Similar data experience outdoors & indoors important! © 2012 Mischa Dohler 9 IEEE ComSoc Austin/Texas Chapter DLT Going 5G: Change in Design Paradigms 2. Put most efforts on architecture and management thereof HBS ABS ABS Polarization 1 Polarization 2 ABS © 2012 Mischa Dohler 10 IEEE ComSoc Austin/Texas Chapter DLT Going 5G: Change in Design Paradigms 3. Acknowledge heterogeneous nature of wireless arena © 2012 Mischa Dohler 11 IEEE ComSoc Austin/Texas Chapter DLT Going 5G: Change in Design Paradigms 4. Use absolute area capacity rather than spectral efficiencies users care about rates spectrum is heterogeneous © 2012 Mischa Dohler 12 IEEE ComSoc Austin/Texas Chapter DLT Going 5G: Change in Design Paradigms 5. Cost considerations must be taken into account at design phase © 2012 Mischa Dohler 13 IEEE ComSoc Austin/Texas Chapter DLT Outline of Talk (all very high‐level!) A. Outdoors 5G Design B. Indoors Femtocell Design C. Management Through SON © 2012 Mischa Dohler 14 IEEE ComSoc Austin/Texas Chapter DLT A. Outdoors 5G Design © 2012 Mischa Dohler 15 IEEE ComSoc Austin/Texas Chapter DLT 5G Architecture Design Principles 1. Water-filing applied to equipment placement and its choice. 2. Aggressive spatial re-use due to ultranarrow beams at BS 3. Aggressive joint design of wireless access and backhaul 4. Aggressive use of licensed and licenseexempt technologies © 2011 Mischa Dohler 16/18 1Gbps/km2 Architecture [1/2] Above-rooftop HBS serving many below-rooftop ABSs: HBS ABS ABS Polarization 1 Polarization 2 ABS © 2011 Mischa Dohler 17/18 1Gbps/km2 Architecture [2/2] Cost-efficient mixture of L/LE/60GHz wireless technologies: © 2011 Mischa Dohler 18/18 Deployment Possibilities Various planning and deployment possibilities: ABS a a a a ABS a a a a ABS a a b a a b ABS b b1 a ABS b b b1 b HBS-LE street ABS b1 b1 ABS ABS a b1 b b a b b1 ABS b b b HBS-L Roof- a a top ABS a b a b b ABS ABS a a a a ABS ABS a © 2011 Mischa Dohler a a a 19/18 1Gbps/km2 Simulation Scenario Grid Architecture: Aggressive Frequency Reuse: mimic Barcelona layout per sector 4 ABSs, 1 user each simulating 1 sector only access link and self-backhaul communicate simultaneously access link and self-backhaul use the same band Self-Organizing Network: © 2011 Mischa Dohler adaptive and distributed power and interference control at the ABS 20/18 1Gbps/km2 Simulation Results 40 MHz bandwidth & 4 beams achieve 1Gbps/km2 capacity density: 8 7 6 5 4 © 2011 Mischa Dohler 21/18 From Theory to Standards ETSI TC BRAN, TR 101 534: © 2011 Mischa Dohler Alvarion, CTTC, Polska Telefonia Cyfrowa, Siklu, Thales "Very high capacity density BWA networks; System architecture, economic model and technical requirements“ 22/18 From Standards to Prototyping [1/3] COBHAM’s high-capacity multi-beam Hub BS: © 2011 Mischa Dohler 23/18 From Standards to Prototyping [2/3] ALVARION’s SISO & CTTC’s MIMO Access BSs: © 2011 Mischa Dohler 24/18 From Standards to Prototyping [3/3] SIKLU’s cost-efficient 60GHz backhaul technology: © 2011 Mischa Dohler 25/18 From Prototyping to Practice [1/5] Tel Aviv 1Gbps/km2 live test trial, April 2012: © 2011 Mischa Dohler 26/18 From Prototyping to Practice © 2011 Mischa Dohler [2/5] 27/18 From Prototyping to Practice © 2011 Mischa Dohler [3/5] 28/18 From Prototyping to Practice © 2011 Mischa Dohler [4/5] 29/18 From Prototyping to Practice © 2011 Mischa Dohler [5/5] 30/18 B. Indoors Femtocell Design © 2012 Mischa Dohler 31 IEEE ComSoc Austin/Texas Chapter DLT The Small‐Cell Solution Space Only viable solution is via change in architecture & smaller cells: Remote Radio Head: Relay: Pico Cell: Femto Cell: “dumb” radio extension of BS; often RF/IF/BB‐over‐fiber wireless radio extension of BS; often limited intelligence intelligent BS; owned, planned and placed by operator intelligent BS; mostly owned and placed by consumer; no planning! © Josep Vidal, UPC © 2012 Mischa Dohler 32 IEEE ComSoc Austin/Texas Chapter DLT Definition & Differentiation Of Femtocell Femtocell Picocell Wifi Site Rental customer operator customer Installation customer operator customer Electricity Bill customer operator customer Radio Planning no (local) yes (prior & global) no Backhaul Connection via customer dedicated via customer Macrocell Interaction not (yet) yes not applicable Transmission Power < 23dBm 23‐30dBm 20dBm mainly closed public closed possible yes vertical Access Rights Handover Photo ©Sagem, TID © 2012 Mischa Dohler © 3g.co.uk 33 © bandaancha.eu IEEE ComSoc Austin/Texas Chapter DLT The Femtocell Opportunity 1. Smart Digital Home: 3. Allows Traffic Offload: © Femto Forum, Thierry Lestable, Sagemcom © Thierry Lestable, Sagemcom 2. Reduces Customer Churn: 4. Energy Saver For All: © Mobile VCE, D. Laurenson © Femto Forum, Thierry Lestable, Sagemcom © 2012 Mischa Dohler 34 IEEE ComSoc Austin/Texas Chapter DLT Finding Femtocells Very Appealing © Femto Forum © 2012 Mischa Dohler 35 IEEE ComSoc Austin/Texas Chapter DLT Femtocell Technology Providers The ecosystem is now mature enough © Thierry Lestable, Sagemcom © 2012 Mischa Dohler 36 IEEE ComSoc Austin/Texas Chapter DLT Femtocell Rollouts & Deployments To‐Date 36 commercial deployments in 23 countries 15 roll‐out commitments in 2012 © 2012 Mischa Dohler 37 IEEE ComSoc Austin/Texas Chapter DLT Femtocells Business Model This image cannot currently be display ed. 50£ one off +5£/month 5$/month 180$ one off +1.5$/month This image cannot currently be display ed. 10$/month 250$ one off This image cannot currently be display ed. free of charge 159$ one off This image cannot currently be display ed. © BeFEMTO 100€ one off +8€ /month 15€/month 10€/month 199€ one off 32$/month 255$ one off +42$/month In BeFEMTO, four business cases are investigated: ‐ One off fee: the user pays 50€ once, and the monthly payment does not change (35€/month) ‐ No fee: the user does not pay for the femtocell, and the monthly payment does not change (35€/month) ‐ Decrease: the user does not pay for the femtocell, and the monthly payment decreases by 5€ to 30€/month ‐ Increase: the user does not pay for the femtocell, and the monthly payment increases by 5€ to 40€/month © 2012 Mischa Dohler 38 IEEE ComSoc Austin/Texas Chapter DLT Femtocell Market Growth Forecast © Informa Telecoms & Media © Thierry Lestable, Sagemcom © 2012 Mischa Dohler 39 IEEE ComSoc Austin/Texas Chapter DLT Related Fora & Standards Bodies Small Cell Forum (before Femto Forum) not‐for‐profit, founded 2007, rebranded 2012 manufacturer‐driven main driver behind femtocell uptake 3GPP IWLAN & EPC standards (for Non‐3GPP Access & Mobility) H(e)NB standards (for Basic FemtoCells) LIPA & SIPTO standards (for Local Access & Offloading) ANDSF Standards (for Discovery, Selection & Policy) LIMONET (for LIPA Mobility), SaMOG(for Trusted WiFi), BBAI (for Broadband Network interworking) IEEE/WFA/WBA Hot Spot 2.0 (Discovery & Seamless Access Control) GSMA WLAN Task Force (2003?) WiFi Offload White Paper, April 2010 GSMA‐WBA Joint Task Force –WiFi Roaming White Paper, Jan 2012 © Prabhakar Chitrapu, InterDigital © 2012 Mischa Dohler 40 IEEE ComSoc Austin/Texas Chapter DLT Femtocell Design Challenges © BeFEMTO 1. Low Device Cost: efficient, low‐cost power amplifiers, highly sensitive receivers, flexible channel bandwidth, reliable RF filters; low cost and low power implementation; etc. 2. System Interference Management: minimize interference to macro (and vice versa); minimize interference to adjacent femtos; coping with unplanned rollouts; coverage estimation, interference cancellation; etc. 3. Femtocell Capacity Maximization: link and access management (handover, admission control, resource management, load balancing and flow control); dynamic bandwidth allocation and sharing; etc. 4. Backhaul Issues: wired or wireless backhaul, reducing signaling load, QoS provisioning and traffic priorization, joint access and backhaul design; etc. 5. Viable System Architecture: control & data planes, access control, authentication, local breakout, efficient forwarding, seamless mobility, zero‐config, etc. © 2012 Mischa Dohler 41 IEEE ComSoc Austin/Texas Chapter DLT Problem Of Interference With Femtos Source: Zubin Bharucha, DOCOMO Euro‐Labs © 2012 Mischa Dohler 42 IEEE ComSoc Austin/Texas Chapter DLT Rel‐10 ICIC In Heterogeneous Networks To support femtocell deployment effectively, inter‐cell interference coordination (ICIC) is necessary Different from homogeneous network (macrocell deployments), o Low power nodes (femto eNBs) must mute (or reduce transmission power) Named as “Protected resources” here o High power nodes (macro eNBs) need not mute o Named as “Non‐protected resources” here Protected/Non‐protected resources are multiplexed in frequency or time‐ domain Both ICIC techniques are effectively supported in Rel‐10 Frequency-domain ICIC Time-domain ICIC Frequency Carrier #1 Carrier Carrier #2 #1 Frequency Time Cell layer Time Cell layer Macro layer Femto layer Source: Zubin Bharucha, DOCOMO Euro‐Labs © 2012 Mischa Dohler 43 IEEE ComSoc Austin/Texas Chapter DLT LTE‐A Control & Data Frame Structure • • data region interference mitigation (a lot of work with viable solutions) control region interference mitigation (surprisingly little work given it is the actual problem since no control means no data) Source: Zubin Bharucha, DOCOMO Euro‐Labs © 2012 Mischa Dohler 44 IEEE ComSoc Austin/Texas Chapter DLT LTE‐A Detailed Control Frame Structure • The control region contains 3 control channels: – PCFICH: control frame indicator; occurs only on first OFDM symbol; scattered in frequency domain; indicates size of control region – PDCCH: downlink control channel; spread in time and frequency; carries scheduling information – PHICH: HARQ indicator channel; spread in time and frequency; contains HARQ information • Focus on the performance of the first two because of differences in their distribution patterns – the PCFICH has restricted positions in the time domain, whereas the PDCCH is dispersed in the time and frequency domains Source: Zubin Bharucha, DOCOMO Euro‐Labs © 2012 Mischa Dohler 45 IEEE ComSoc Austin/Texas Chapter DLT LTE‐A Control Interference ‐ Current Approaches (a) •No coordination Heavy interference on 2 OFDM symbols Source: Zubin Bharucha, DOCOMO Euro‐Labs © 2012 Mischa Dohler (b) • Femto control channel sparseness Interference to first OFDM symbol is lowered 46 (c) •Almost blank subframe Only interference from reference symbol Femto data transmission is not allowed IEEE ComSoc Austin/Texas Chapter DLT Novel Femto‐Macro ICIC Approach [1/4] The proposal from DoCoMo Euro‐Labs advocates carefully selecting the Physical Cell Identifier (PCI) of HeNBs at start‐up, such that any interference caused by their control channels to the PCFICH of any trapped macro UEs is avoided. o In order for this to be possible, the HeNB needs to identify the eNB that it is closest to. Identifying the eNB means that the HeNB must be aware of the PCI of the eNB (decoded using synchronization procedure). Source: Zubin Bharucha, DOCOMO Euro‐Labs © 2012 Mischa Dohler Illustration only 47 IEEE ComSoc Austin/Texas Chapter DLT Novel Femto‐Macro ICIC Approach [4/4] Qualcomm Proposition in BeFEMTO for Resource Partitioning in Space for Femto‐Macro Interference Mitigation: Problem Statement: proper resource partitioning is a must to handle femto‐macro interference beam selection in MIMO‐enabled femto BS equipment could be made use of Innovative Proposition: design of suitable coordinated beam selection algorithm following LTE codebook designs comparative study between different restriction and codebook policies study outcomes: codebook restrictions yields gains to macro users at low femto capacity loss 50% improvement @ macro 5% deterioration @ femto Source: Qualcomm, BeFEMTO © 2012 Mischa Dohler 48 IEEE ComSoc Austin/Texas Chapter DLT C. Self‐Organizing Networking © 2012 Mischa Dohler 49 IEEE ComSoc Austin/Texas Chapter DLT Capacity‐Optimal Schedule? With femtocells, the degrees‐of‐freedom (DOF) increase significantly – and with it complexity! © 2012 Mischa Dohler [presented by Interdigital: Globecom’11 – IWM2M, Houston] 50 IEEE ComSoc Austin/Texas Chapter DLT System Degrees of Freedom High number of femtos/users/resources/etc + interference constraints ‐> Problem of DOF Degrees of Freedom (DOF) / Area: new system DOF = old system DOF x 20‐30 new system density = old system density x 4 new DOF/km2 = old DOF/km2 x 100 © 2012 Mischa Dohler 51 IEEE ComSoc Austin/Texas Chapter DLT SON Is The Only Solution SON Helps For Previously Manual Processes: reduce manual intervention for deployment savings automate repetitive processes examples: automatic planning & self‐configuration SON Helps For Too Fast/Complex Processes: improve run‐time operation based on real‐time data automate optimization of critical network elements example: self‐optimization & self‐healing Among the many possible approaches, we deal with cognitive/docitive RRM to facilitate SON. © 2012 Mischa Dohler 52 IEEE ComSoc Austin/Texas Chapter DLT SON In Cellular Networks © 2012 Mischa Dohler 53 IEEE ComSoc Austin/Texas Chapter DLT Important SON Tradeoffs © 2012 Mischa Dohler 54 IEEE ComSoc Austin/Texas Chapter DLT SON Through Cognition/Docition From cognitive to docitive networking paradigms: cognition: “system which is working under conditions it was not designed for” docition: “teaching” between expert nodes to accelerate learning/cognition result: truly autonomous SON with quick convergence © 2012 Mischa Dohler 55 IEEE ComSoc Austin/Texas Chapter DLT Cognitive Approach – Q‐Learning FOR each (s, a) DO Initialize table entry: Q ( s, a ) 0 ... a1 p1 al pl Observe current state s s1 Q( s1 , a1 ) WHILE (true) DO s2 Q( s2 , a1 ) ... Q( s 2 , a * ) ... Q( s2 , al ) v Q(v, a1 ) ... Q (v, a * ) ... Q (v, a l ) sk Q( sk , a1 ) Q( s1 , al ) Select action a and execute it Receive immediate cost c Observe new state v Update table entry for Q(s,a) as follows Q(s, a) Q(s, a) [c maxQ(v, a) Q(s, a)] a Q( sk , al ) State transition from s to v © 2012 Mischa Dohler 56 IEEE ComSoc Austin/Texas Chapter DLT Docition Approach – Exchange of Q‐Entries Nodes operate in a distributed and autonomous fashion but exchange suitable parts of their Q‐Table: Output Action pxy Input State Startup Docition. Docitive femto BSs teach their policies to newcomers joining the network. IQ‐Driven Docition. Docitive radios periodically share part of their policies with less expert nodes with similar gradient. © 2012 Mischa Dohler 57 IEEE ComSoc Austin/Texas Chapter DLT Performance – Superior Capacity Macrocell capacity as function of femtocell density: © 2012 Mischa Dohler 58 IEEE ComSoc Austin/Texas Chapter DLT Performance – Superior Convergence Precision CCDF of average SINR at macrouser for a 50 % occupation ratio: © 2012 Mischa Dohler 59 IEEE ComSoc Austin/Texas Chapter DLT Performance – Superior Convergence Speed Convergence speed improves by order of magnitude from 20,000 (cognitive) to 4,000 (docitive) iterations: © 2012 Mischa Dohler 60 IEEE ComSoc Austin/Texas Chapter DLT D. Future Challenges © 2012 Mischa Dohler 61 IEEE ComSoc Austin/Texas Chapter DLT Conclusions on Outdoors 5G Operator’s capacity challenge today and for next years: capacity requirements are far off reality except for order‐of‐magnitude PHY, rather work on architecture infrastructure cost is major driver since ROI margins tighten management becomes major problem, SON is a must 1Gbps/Km2 architecture with the following properties... … LTE(‐A) & WiMAX‐agnostic architecture ... anytime and everywhere in urban environments … cost‐efficient to operators, service providers, users … spectrally efficient using both licensed and exempt bands ... autonomous operation facilitating deploy & forget experience © 2012 Mischa Dohler 62 IEEE ComSoc Austin/Texas Chapter DLT Conclusions on Femto Cells Heterogeneity Is Quickly Increasing: femtos, networked femtos, wifi, m2m, etc, etc future use outdoors, mobile, in relay form allow for viable QoS/QoE delivery Self‐Organizing Networking Will Be Key: distributed localized SON (important for signaling) standards compliant SON approaches for industry uptake take energy constraints into account Business Models subscriber models FON‐like approaches © 2012 Mischa Dohler 63 IEEE ComSoc Austin/Texas Chapter DLT Conclusions on Femto Cells Femtocell rollouts of the future: Data Traffic: 40% @Home 35% On the Move 25% Work © BeFEMTO & Thierry Lestable, Sagemcom © 2012 Mischa Dohler 64 IEEE ComSoc Austin/Texas Chapter DLT Rate Heterogeneous Access Technology Landscape high low WiFi & Femto (IEEE, IETF, 3GPP) Cellular xG (ITU, 3GPP, ETSI) ZigBee & LP Wifi (IEEE, IETF) M2M (ETSI, 3GPP) low © 2012 Mischa Dohler high 65 Range IEEE ComSoc Austin/Texas Chapter DLT ANNEX Subsequent slides are not part f endorsed by the IEEE nor ComSoc, and a private slide of M Dohler. © 2012 Mischa Dohler 66 IEEE ComSoc Austin/Texas Chapter DLT Disclaimer Besides my own, many third party copyrighted material – mainly from the BeFEMTO Femtocell Winterschool 2012 & BuNGee project – is reused within this tutorial under the 'fair use' approach, for sake of educational purpose only, and very limited edition. As a consequence, the current slide set presentation usage is restricted, and is falling under usual copyrights usage. Thank you for your understanding! © 2012 Mischa Dohler 67 IEEE ComSoc Austin/Texas Chapter DLT CTTC [www.cttc.es] Research, prototypes, commercialization: 6 cutting‐edge R&D areas: > 200 papers per year! Major branch in Hong Kong!!! © 2012 Mischa Dohler 68 IEEE ComSoc Austin/Texas Chapter DLT LENA NS3 Open‐Source Simulators (femtos too) CTTC working with Ubiquisys, the leading femtocell manufacturer: develop a common platform for LTE femto/macro cell vendors to evaluate their different solutions open source to foster adoption/contributions; based on NS‐3; http://iptechwiki.cttc.es/LTE‐EPC_Network_Simulator_(LENA) use case: LTE‐based Self Organized Networks need to test SONs algorithms before deployment Ubiquisys made extensive use of simulation to design its first generation of WCDMA intelligent femtocells Product–oriented: Real‐world interfaces for SON algorithms FemtoForum MAC Scheduler API specification © 2012 Mischa Dohler 69 IEEE ComSoc Austin/Texas Chapter DLT BeFEMTO Project [www.ict‐befemto.eu] > 30 Deliverables! © 2012 Mischa Dohler 70 IEEE ComSoc Austin/Texas Chapter DLT Femtocell Special Issues © 2012 Mischa Dohler 71 IEEE ComSoc Austin/Texas Chapter DLT Transactions on Emerging Technologies Year of Average Days to submission first decision Average Days to final decision 2009 158 243 2010 207 245 2011 61 83 © 2012 Mischa Dohler 72 IEEE ComSoc Austin/Texas Chapter DLT