Research Challenges in Wireless Communications & Networking D. Raychaudhuri WINLAB, Rutgers University Piscataway, NJ 08854 ray@winlab.rutgers.edu 1 Introduction 2 Wireless Research: Strategic Themes (near-future) Several fundamental problems need to be solved before the “mobile Internet” can take off: Developing PHY/MAC for broadband radios ~Kbps Mbps Gbps, adaptive, robust, QoS,... Scaling wireless system capacity widespread service implies ~Gbps/Sq-Km Designing wireless system-on-chip (SOC) low-cost/low-power, integrated CMOS Unifying wireless network architectures (WLAN/IP, 2.5G, 3G cellular) & protocols multiple radio technologies, faster/simpler standards process Creating “useful” mobile information services ...beyond web browsing on hand-held devices 3 Wireless Research: Strategic Themes (long-term) Pervasive computing via large-scale sensor networks (connecting people with their physical environment) viable in 5-10 yrs Technical challenges: self-organizing (ad-hoc) networks low-power/low-cost/multipurpose wireless sensors scalable network routing and content distribution distributed information processing in the network end-user interfaces & applications Above topics involve wireless, but are also inherently cross-layer or interdisciplinary... 4 Wireless Product Trends Wireless local loop (WLL) MIMO/OFDM, ATM/IP, Broadband Wireless Access (BWA) 3G+ or BWA+?? WCDMA, 3G.PP, etc. Integrated Cellular (3G) Digital Cellular (2/2.5G) Wireless LAN (802.11b) OFDM, mob IP, security, QoS,.. 2001 802.15.3 WPAN, etc. driver technologies OFDM/CDMA, MIMO, diversity, RRM,.. Public WLAN 3G/WLAN IWF, 4G: self-org 802.11 Wireless LAN (802.11x) potentially disruptive technology areas Short-range radio (Bluetooth) BWA/3G combo (local access providers) Home LAN WPAN (802.15.3.x) 2002-03 WLAN/3G/2G (cellular operators) low- tier 802.11 Home network sensor nets, etc. (consumer & verticals) UWB, ad-hoc nets >2005 convergence opportunities??5 Wireless Research Challenges: Major Areas Wireless research topics can be organized into following major categories radio modems: signal processing and hardware wireless systems: design and optimization mobile networks & protocols Many wireless problems of current importance are cross-layer in nature, so that a holistic approach is essential .... 6 Radio Technology 7 Radio Technology: Research Topics Selected research topics in the radio/modem area include: putting radio modems on “Moore’s Law” signal processing innovations (MIMO, adaptive antennas) flexible software-defined radios (SDR) ultra wideband (UWB) integrated wireless system-on-chip (sensors, etc.) 8 Radio Technology: Moore’s Law applies to wireless! • As computing and communications converge, network BW must follow CPU & memory size…. 802.11a, UWB,.., 1000 1000 1000 1000 1000 Cable Modem Gbps Router 802.11bWLAN, DSL ATM 10 10 100 Memory Size 10 100 CPU Speed 100 LAN/WAN Switching 10 100 Local Access Wireless Access 100 CPU Kbps Kbps Mbps 1 1 1 1 3G Mobile 56K modem Wireless Sw Ethernet 10 Mhz Local Access LAN/WAN CDPD Memory MB short-range radio speeds outpacing Moore’s law over last ~5 yrs! 1 1990 Year 1995 2000 9 Radio Technology: Modem Evolution Time/Frequency processing Time/Frequency + spatial processing Multicarrier Modulation (OFDM, etc.) DVB, 802.11a, etc.. QPSK/GMSK Equalized QPSK/QAM/ GMSK,.. Multiple antenna spatial processing (MIMO, etc.) 4G and next-gen WLL ~10-100 Mbps depending on cell size & mobility ~5-10 bps/Hz achievable with QAM Pulsed communication IS-136, etc. US HDTV, WLL, 802.11b UWB Spread Spectrum (CDMA) IS-95 Wideband CDMA (w/ interference canc. & multiuser det) WPAN and WLAN ~100-500 Mbps no allocated spectrum no RF carrier short-range, high-data rate UMTS/IMT-2000 ~2 Mbps depending on cell size ~0.5 bps/Hz typical for proposed systems (works at vehicular mobility speeds) 10 Short-range radio channels Example opportunistic transmission scenario: : vehicular user passes by an “Infostation” z Offset w W d 11 Short-range radio channel Initial results show that channel is well-behaved for distance ~5m 100’s of Mbps readily achieved with various modem techniques Data from Domazetovic & Greenstein [2001] 12 Source: J. Foerster, Intel Research, 2001 Throughput (Mbps) Radio Technology: UWB 500 450 400 IEEE802.11a UWB IEEE802.11g 350 300 250 200 150 100 50 0 0.00 “sweet spot” for use as nx100 Mbps WPAN 10.00 20.00 30.00 40.00 50.00 60.00 70.00 Distance (m) UWB appropriate for energy-efficient radio links, typically short-range Also has potential hardware complexity advantages... Pragmatic bit-rate comparison between UWB and 802.11x options 13 Radio Technology: Hardware Innovations As wireless modems become faster and more ubiquitous, key hardware innovations urgently needed: compact RF components, including MEMS mixed signal design & testing silicon integration and packaging UWB radio architecture software-defined radio @ 10-100 Mbps integrated wireless sensors (low-power) 14 Wireless Systems 15 Wireless Systems: Research Topics Designing and optimizing wireless systems via radio resource management (power control, interference avoidance, scheduling, etc.) Selected research topics in the wireless systems area include: scaling cellular system capacity scaling ad-hoc network capacity & throughput per user radio resource management for 3G and ad-hoc nets interference avoidance spectrum sharing in unlicensed bands 16 Wireless Systems: Increasing the scale of networks • Rapidly increasing use of untethered data devices implies that wireless access network capacity (bps/sqKm) will soon have to scale to “gigabit” levels... Growing proportion of all computing devices --> 50% +? Telecom Network Internet Mobile Comm Devices Wireless Access Networks Sensors/ low-tier data Mobile PDA/PIA Semi-mobile Laptop, etc. Fixed PC/WS Example: ~10,000 devices/sq-Km @1 Mbps peak and 0.1 Mbps avg implies system capacity ~Gbps/sq-Km 17 Wireless Systems: Increasing the scale of networks • Consider first the scaling limits of existing and emerging wireless network standards... • 2G cellular/PCS: cell size ~ 3-5 Km, avail BW ~ 5 Mhz, spectral eff ~ 0.2-0.3 bps/Hz max capacity ~ 100 Kbps avg, 1 Mbps peak (with packet MAC) per sq-Km off by 3 orders-of-magnitude! • 3G Cellular/PCS: cell size ~ 3-5 Km, avail BW ~ 25 Mhz, spectral eff ~ 0.3-0.5 bps/Hz max capacity ~ 1 Mbps avg, 10 Mbps peak (with packet MAC) per sq-Km still off by 2 orders-of-magnitude! • Wireless LAN (802.11x, Hiperlan): cell size ~ 0.1-0.5 Km, avail BW ~ 100 Mhz, spectral eff ~ 0.2-0.3 bps/Hz max capacity ~ 100 Mbps avg, 1 Gbps peak per sq-Km correct order-of-magnitude, but too many access points & limited mobility 18 Wireless Systems: Architecture Evolution Standard IP, ATM, etc. Standard IP + M interface Dynamic provisioning/ QoS Mobile/Wired Network GW WAP services. etc. Static Regulated spectrum, static freq co-ord High-speed radio hot spot Radio macrocell Unregulated spectrum, dynamic freq coordination Radio Microcell (~0.5-1 Km radius) provisioning Custom wireless protocol Location-aware information services, mcast, cache, etc. Gigabit Metro Area Network (w/ integrated mobility support) BTS AP/ mini-BTS WPAN Mbps/Km2 Gbps/Km2 2G/3G end-users IP end-users 2G/2.5G/3G radio access (single standard) Cellular Macrocell (~5-10 Km radius) WLAN+ or “4G” or new radio access (multiple standards) Faster radio PHY’s with high interference rejection & bps/Hz efficiency WLAN Microcell (~100m radius) Current Wireless Network IP end-users Scalable Heterogeneous Pico/Micro/Macrocellular 19 Wireless Network Model Wireless Systems: RRM Model for Cellular systems hk 1 BS k hk 4 h14 hk 5 hk 2 hk 3 h15 BS 1 Source: Prof. R. Yates, Rutgers U h16 • Multiple cell scenario with desired and interfering signals • Algorithms for allocation of bit-rate, base station, channel, tx schedule, power • Common theme: reduce interference, transmit when the channel is “good” 20 Wireless Systems: RRM in 3G – adaptive incremental redundancy example Source: Dr. L. Razoumov, Rutgers U 21 Wireless Systems: Efficient Spectrum Use • Scaling of wireless services will need new spectrum (~Ghz) particularly for new highspeed data services • Need to rethink traditional approach to spectrum regulation More unlicensed spectrum (e.g. 5 Ghz U-NII) Market mechanisms other than one-time spectrum auctions? Spectrum etiquette procedures for coexistence of QoS-based wireless services (beyond “LBT”) Incentives for efficient utilization of spectrum resources? Relationship to property rights? 22 Wireless Systems: Efficient Spectrum Use • Spectrum etiquette procedure a key issue for U-NII scenario • “CSCC” approach proposed as possible solution... Coordination channel using simple standard protocol at edge of band Semantics of higher layer coordination protocol TBD... Support arbitrary spectrum policies based on user priority, cost bids, etc. Channel: #1 #2 #3 #4 #5 #6 #N Common Spectrum Coord Channel (CSCC) Packet service ..... Streaming service A Streaming service B .... .... Periodic announcements incl..: Service type, User #, Channel #, service params, Priority, Cost/Price Bids, etc. 23 Wireless Systems: Efficient Spectrum Use Example of CSCC etiquette used for “dynamic pricing” based spectrum allocation: A fn B fn B contends for fn channel fn A CSCC User ID B Service Type Price Bid $.05/hr A raises bid on fn channel fn A Price Bid $.07/hr Price Bid $.09/hr A wins contention ( B records & reports transaction!) …e-cash exchange ? 24 Mobile Networks 25 Mobile Networks: Some Research Topics Selected research topics in the mobile networks area include: new MAC protocols: 802.11x, 803.15.x, sensor nets “4G” network architectures mobility protocols: beyond mobile IP new architectures (WLAN hot-spots, Infostations, ..) self-organizing wireless networks (sensors, etc.) ad-hoc network routing multicasting and mobile content delivery wireless network security 26 Mobile Networks: “4G” Protocol Evolution 2.5G/3G Services 4G Services uniform service API (Internet+) PSTN IP GSM/ GPRS 3G Access Network service feature modules 2.5G/3G Radio Security QoS VPN Content Delivery generic network API WLAN Services Low-tier services IP Mobile Service Middleware IP WPAN network layer (e.g. Bluetooth) Ethernet WPAN radio 802.11 Radio Radio-specific vertically integrated systems with complex intetworking gateways Today’s Wireless Systems Unified IP-based mobile network Generic Radio Access Network 3G/4G Radio WLAN radio incl support for multihop, mcast, etc, uniform radio API’s WPAN/lowtier radio Radio Independent modular system architecture for heterogeneous networks The Future 27 Mobile Networks: Protocols beyond mobile IP Global Internet Mobile IP overlay network access point radio bridge/ router (forwarding node) Radio Access Network 1 IP extensions or generalized L2 MAC?? Mobile IP provides a permanent IP address for users moving between wireless AP’s Desired RAN features for ad-hoc WLAN, sensor nets, 4G: - handoff support (micro-mobility) - discovery and self-organization - ad-hoc routing, integrated with MAC - peer-to-peer modes - multicast, QoS, security, etc. closer layer 2/3 coupling needed 28 Mobile Networks: 3G/WLAN interworking Cellular/2.5G,3G Bluetooth UWB, Bluetooth<-> 3G IWF Unified Mgmt Layer net link PHY Bluetooth<->WLAN IWF WLAN<->3G IWF WLAN, HiperLAN, UWB, 3G/WLAN interworking BT IWF1 WLAN IWF2 3G Protocol stacks Techniques for seamless service: - Authentication, global roaming - Security issues - Dynamic handoff - End-to-end QoS control Multiple devices with various radio interfaces - Network management - Service level agreements29 Mobile Networks: Hot-Spot MAC • Mobile user passes through hot-spot (Infostation) in sec during which ~MB files are downloaded/uploaded – Requires modifications to conventional WLAN MAC, incl fast synch, pre-authentication, etc. – Motivates 2-tier arch with ~10m service zone (for high-speed data transfer) and ~50m access control zone Infostations access point Data cache Low-speed control channel (for synch & service setup) Service Zone ~100 MB/s Fast transfer Access Control Zone Transit time ~sec Total transit time ~10sec 30 Mobile Networks: Hot-Spot MAC • 802.11a MAC can be used for opportunistic service – Pre-authenticate user in low-bit rate mode (~50m range) – Mobile terminal waits for modem to reach max 54 Mbps (~10m range) – High priority access mode used for Infostations access normal channel activity AP Beacon A1 IS Control packet Terminal enters WLAN coverage area Mobile requests advance authentication ........ A3 .. A2 Authentication message exchange PIFS PIFS Terminal enters max PHY speed zone IS transfer request* Priority Access initiated Infostations file transfer* ACK *RTS/CTS msgs not shown time 31 Mobile Networks: UWB Sensors • UWB potentially well-suited for sensor networks Bit-rate readily traded off against range Energy efficient modulation Robust to interference Multiple radio links supported by single UWB RF Low cost silicon for integrated sensor device UWB (R12, code 12) S1 UWB (R23, code 23) S2 UWB (R13, code 13) S3 32 Mobile Networks: UWB Sensor MAC • Potential MAC/link layer based on DS/CDMA UWB PHY: Continuous beacon for synchronization & sensor ID broadcast Low bit-rate, high-spreading gain common link establishment channel with a single code used in random access mode Handshake protocol for setting achievable link bit-rate with dedicated code Beacon S1 Beacon S2 S1 S2 Link establishment signal (S1,S2, C12) S1 Common code Control Code A Rate adaptation, ARQ Link ACK (S1,S2, C12) S2 Code B 33 Mobile Networks: Ad-hoc Networks • Ad-hoc network ideas proposed for tactical and sensor scenarios, with potential applications to WLAN/4G: flat network model with multi-hop routing radios on-demand routing protocols (DSR, AODV, etc.) designed for high node mobility (...fairly mature topic) enhancements via MAC clustering, energy-efficient routing, .. application-level data aggregation (diffusion routing, XML,..) geographically constrained routing SN MAC cluster (optional) radio links for multi-hop routing Active problem areas: -Scaling of capacity - Dynamic behavior - Energy efficiency - MAC/routing interactions - QoS routing - Geo routing - Security of ad-hoc nodes - Integration with WLAN, etc. 34 Mobile Networks: Hierarchical Ad-Hoc Net • Hierarchical, self-organizing network currently under consideration, based on: 3 service tiers (cellular, WLAN, personal area) BS’s, AP’s, FN’s (forwarding radio nodes), user devices automatic discovery and power mgmt protocols hierarchical, ad-hoc multihop routing and spatial MAC Internet BTS AP WLAN micro-cell Access Point Forwarding node FN 3G cell low-tier (e.g. sensor) user nodes personal-area pico-cell 35 Mobile Networks: Higher Layers • Research issues which arise in connection with information delivery over wireless nets: Qos with heterogeneous & time-varying radios transport layer problems (TCP timeouts, etc.) need for services such as reliable multicast information “pull” model vs. multicasting model opportunistic services (hot-spots, caching,..) delivery of the “right information” at the “right time and place” (location/content aware) media scaling to match radio and terminal capabilities sensor network & pervasive computing software models 36 Mobile Networks: Content Multicast • New real-time, context- and location-aware information delivery paradigms under consideration ... • Content multicasting based on XML investigated as possible option for delivering relevant info to mobiles. Interest profile User XML Descriptor SX SX Semantic Router A Semantic Router B content multicast Mobile interest profile contains: (user, location, terminal capability,..) Content Provider 37 Mobile Networks: Experimental Research • A flexible, open-architecture mobile/ad-hoc sensor network testbed recently established at WINLAB – open-source Linux routers and AP’s (commercial hardware) – Linux and embedded OS forwarding and sensor nodes (custom) – radio link and global network monitoring/visualization tools 802.11b PDA Management stations Radio Monitor AP Forwarding Node/AP (custom) 802.11b Linux PC Commercial Router network 802.11 with arbirtrary topology Compute & storage servers Sensor Node (custom) PC PC-based Linux router 38 Wireless Research: Multidisciplinary Research Topics In conclusion, we mention some wirelessrelated multidisciplinary research topics: spectrum regulation principles (...economics, policy) integrated wireless sensors (...materials, semiconductor) software models for pervasive computing (..CE, CS) dynamics of large-scale ad-hoc sensor nets (...math, control) security in ad-hoc sensor networks (...CS) new applications of sensors: environmental, medical, public safety, etc. (..CS, domain experts from various disciplines) robotics (..mechanical, controls) 39