Backhauling in TV White Space Narayan B Mandayam
Backhauling in TV White Space Narayan B. Mandayam (joint work with Cyrus Gerami, Larry Greenstein, Ivan Seskar) WINLAB, Rutgers University IEEE Distinguished Lecture 1 WINLAB
What is White Space? p TV Band Devices: Fixed or Portable n p Max. Fixed antenna height = 30 m, Portable < 3 m Permissible channels (6 MHz each) X p Additional Ruling on Sep 23 2010 Transmit Restrictions n n n Protected region around primary TV transmitters Sense and avoid protected devices TX power: p Fixed: 30 d. Bm (6 d. Bi antenna gain) = 4 W EIRP, p Portable: 20 d. Bm (no antenna gain) = 100 m. W, § Co- and Adjacent-channel not allowed § Co-channel not allowed, Adjacent = 16 d. Bm 2 WINLAB
What is “Really” White Space? p Economist n p ≈ $, votes New Technology, Cognitive Radios ≈¢ Free speech, Bill of Rights priceless Folks who are “out there” n p Social Good Engineer n p ≈$ Regulator/Politician n p Markets, Property Communication/Information Theorist n W 3 WINLAB
How much TV White Space is there in NJ? p TV Towers around NY City and Philadelphia n Lots of white space spectrum available in NJ! # of channels (fixed) vs. # of 5 X 5 sq. mi. grids 7 – 31 channels available per cell (42 – 186 MHz) 4 WINLAB
Radio Coverage p Prime spectrum with a wide range of applications n n ~50 -200 MHz available depending on TV transmitter density Power constraints result in achievable bit-rate profile for fixed, fixed-mobile, and mobile-mobile ~5 Mbps @ 2 Km range for LOS fixed-mobile ~3 -5 x Wi. Fi range for non-LOS services, e. g. ~50 Mbps @ 250 m 5 WINLAB
White Space Networks n Range of possible usage scenarios, with sweet spot in outdoor networks with medium range and speed Bit-Rate 6 100 m WINLAB
Sample Applications: Cellular Data Boost n “Cellular data boost” network can be used to offload fast-growing cellular traffic using dual-mode radio Mesh network of outdoor white space hot spots; backhaul data to existing BTS ¨ Intended for transport of non-real time data such as mail, content, facebook … ¨ Potential for ~2 -5 x capacity boost depending on % coverage & service mix ¨ 7 WINLAB
Sample Applications: Distribution/Backhaul DISTRIBUTION AND BACKHAUL USING WHITE SPACE 8 WINLAB
Sample Applications: n Long range V 2 V/Emergency Network Long-range V 2 V useful for traffic control/warnings, geographic apps, p 2 p content, etc. Supplements short-range 802. 11 p/DSRC n n V 2 V links (from mandated car radios? ) can be used to form a high capacity emergency backup network using ad hoc mesh between cars and fixed AP’s Application requirements well matched with WS range/bit-rate properties 9 WINLAB
Sample Applications: Cognitive Digital Home Service Provision Device GENIE NODE Provides end-user service Central spectrum manager Relay and Wireless Access Devices Provides relay/connectivity support 10 WINLAB
Design Implications for White Space Networks n. White n space radio systems require the following building blocks: Flexible BW PHY, preferably operating in non-contiguous spectrum Spectrum sensing for TV primary and other incumbents, coordination with data base n Opportunistic link layer access with distributed congestion control procedures for fair sharing among secondary users n n Discovery and bootstrap protocols for ad hoc network formation Common coordination channels and/or spectrum servers for improved coordination among multiple types of secondary users, e. g. Databases n n … and of course, cognitive SDR platforms (wideband, flexible, low-cost) 11 WINLAB
WS Building Blocks: NC OFDMA PHY p NC OFDMA approach used to opportunistically fill spectrum White Space Center freq Primary freq Min. tones needed for freq. synchronization p Allows for flexible spectrum sharing for secondary coexistence WINLAB
Case for Noncontiguous OFDMA - I C • Three available channels 1 23 • Node A transmits to node C via node B. B • Node B relays node A’s data and transmits own data to node C. • Node X, an external and uncontrollable interferer, transmits in channel 2. X A q If we use max-min rate objective and allocate channels, node B requires two channels; node A requires one channel q Scheduling options for Node A and Node B? 13 2
Case for Noncontiguous OFDMA - II #1: Contiguous OFDM #2: Multiple RF front ends C C 1 3 B 2 X 3 Nulled Subcarrier C 12 B #3: Non-Contiguous OFDM (NC-OFDMA) A • Transmission in link BC suffers interference in channel 2 B 2 X 2 A A NC-OFDM accesses multiple • Spectrum fragmentation limited by number of radio fragmented spectrum chunks with single radio front ends 14 14
NC-OFDM Operation Non-Contiguous OFDM Nulled Subcarrier X[2] = 0 X[1] X[3] X[1] Serial to Parallel IFFT x[1] x[2] x[3] X[3] 1 3 Parallel to Serial D/A • Node B places zero in channel 2 and avoids interference • Node A, far from the interferer node X, uses channel 2. • AP Both nodes use better channels. Modulation B 2 X 2 A NC-OFDM accesses multiple fragmented spectrum chunks with single radio front end • Node B spans three channels, instead of two. • Sampling rate increases. 15
Resource Allocation in NC-OFDMA Benefits: q Avoids interference, incumbent users q Uses better channels q Each front end can use multiple fragmented spectrum chunks Challenges: q Increases sampling rate § Increases ADC & DAC power § Increases amplifier power q Increases peak-to-average-power-ratio (PAPR) Develop centralized, distributed and hybrid algorithms for carrier and forwarder selection, power control 16
WS Building Blocks: NC OFDMA MAC p p p NC OFDMA offers the possibility of a simple FDMA MAC instead of CSMA or TDMA (. . CSMA may still be used for enduser access) Simplifies ad hoc network operation and avoid classical mesh self interference and exposed node problems Requires a cooperative access policy (i. e. not greedy, and with some form of congestion backpressure) f 2 f 3 f 1 rate r 2 rate r 3 freq LINK 1 LINK 2 LINK 3 Rates r 1, r 2, r 3 periodically adjusted via cooperative procedures WINLAB
Architectures for Secondary Coexistence p p Secondary co-existence an important requirement for WS Various schemes possible depending on system model n n n Completely autonomous, using performance feedback only Common coordination channel Common Internet based spectrum server Spectrum Server (optional) freq Secondary B Spectrum Secondary A Spectrum Internet WS AP w/ backhaul Control information WS Mobile Access Protocol Secondary System A Common Coordination Channel (optional) Secondary System B WINLAB
DISTRIBUTION AND BACKHAUL USING WHITE SPACE 3 G WHITE SPACES WIFI FIBER BACKHAUL NETWORK 19 WINLAB
OUTLINE • • The Proposed System First order Methodology Achievable Capacity Traffic Demand How many cells would need fiber? Aggregating Flows Conclusion and Future Directions White Space: Where are we? Where do we go? 20 WINLAB
HOW WILL IT LOOK? • NJ as case study • Proximity to NY & Philly • Highest population density • WINLAB in NJ! • Cells of 5 mi X 5 mi • total 307 • Antenna (base-station) in each • FCC’s max allowed height=30 m • FCC’s max TX power=4 W • Based on fixed devices rules of FCC 21 WINLAB
WHAT WILL IT DO? Antenna coverage Use of Sector antennas for more concentrated transmission Fiber Internet user Wifi Wireless Distribution and Backhaul 4 sector antennas WINLAB
CAN WHITE SPACES BE USED? >< FIRST ORDER CRITERION Use Radio ACHIEVABLE CAPACITY (PER CELL) Use Fiber AVAILABLE BANDWIDTH SPECTRAL EFFICIENCY Resources used: FCC rules DEMAND RECEIVED SNR NOISE TX POWER Propagation models 4 W PATH Ltraffic OSS Internet survey Internet usage statistics ITU Terrain NJ pop statistics Thermal and Noise Census 2000 Figure 23 WINLAB
NJ TOWERS AT A GLANCE • Towers in NJ, NY, DE & PA • Coverage can be 100 km (r) AVAILABLE BANDWIDTH 24 WINLAB
AVAILABLE BANDWIDTH FCC’S PROTECTION RULE Primary TV tower Protected radius Secondary White Space radio Additional separation ring 25 WINLAB
AVAILABLE BANDWIDTH CHANNEL AVAILABILITY AVAILABLE INCLUDING SPACE PER ADJACENT CHANNEL TV tower coverage Additional separation ring Available for possible use Available as White Spaces 24 25 CHANNEL EFFECT 25 26 26 WINLAB
AVAILABLE BANDWIDTH DATABASE 25 • Repeat this for each cell and you get bandwidth database • Each channel is 6 MHz • 7 – 31 channels available per cell (42 – 186 MHz) X • No islands • Similar channels available in neighboring cells INTERFERENCE! WINLAB 27
AVAILABLE BANDWIDTH FREQUENCY REUSE PLANNING reuse factor (r) of 2 : • SNR at cell-2 = 19 d. B • SNR at cell-4 = 5 d. B – Interference • 14 d. B isolation for r=2 • Median path loss: ITU terrain model for LOS • • Obstruction height: 15 m for sparse population and 30 m for dense population 1% outage with 8 d. B shadowing variance 28 WINLAB
>< ACHIEVABLE CAPACITY DEMAND AVAILABLE BANDWIDTH SPECTRAL EFFICIENCY BANDWIDTH DATABASE RECEIVED SNR NOISE TX POWER 4 W (6 d. BW) PATH LOSS Thermal (-136 d. BW) + Noise Figure (10 d. B) 29 WINLAB ITU Terrain
LET’S CONSIDER ONE CELL • 54 MHz (9 channels) available • 27 MHz usable (reuse) • Spectral Efficiency: 6. 23 bps/Hz (path loss and population and building heights) • Max Achievable Capacity: ~168 Mbps ~75. 7 GB/hour 30 WINLAB
>< ACHIEVABLE CAPACITY US Census 2000 Our Approximation SIMULTANEOUS ACTIVE USERS (α) USAGE PER HOUSEHOLD • Pop/sq mi pop/cell • 3 people per household • 74. 2% have internet clients/cell • 18 MB/hr (Cisco Survey) 90 MB/hr (5 times more) 126 MB/hr (7 times more) 180 MB/hr (10 times more) α = 10% α = 30% α = 50% 31 WINLAB DEMAND
LET’S CONSIDER ONE CELL • • Cell pop: 8750 Cell households: 2917 Cell internet connections: 2164 Cell traffic using α = 30% : 18 MB/hr/link: 11. 7 GB/hr 90 MB/hr/link: 58. 4 GB/hr 126 MB/hr/link: 81. 8 GB/hr 180 MB/hr/link: 116. 9 GB/hr 32 WINLAB
LET’S CONSIDER ONE CELL >< ACHIEVABLE CAPACITY α = 30% & 18 MB/hr/link : α = 30% & 90 MB/hr/link : α = 30% & 126 MB/hr/link : α = 30% & 180 MB/hr/link : 33 75. 7 > 11. 7 75. 7 > 58. 4 75. 7 F I<B 81. 8 ER 75. 7 F I<B 116. 9 ER WINLAB DEMAND
HOW MANY CELLS NEED FIBER? (OUT OF 307) 18 MB/hr 90 MB/hr 126 MB/hr Cells requiring fiber connection 34 WINLAB 180 MB/hr
AGGREGATING MULTIPLE FLOWS ER T S LU C Proposed Solution: CLUSTER HEAD • Use Excess Capacity for aggregation R E B I F • Excess Capacity = Achievable Capacity Demand • Clustering • Plant more fiber at cluster heads • Plant cluster heads in high capacity cells to route traffic through • Detailed routing study 35 WINLAB
EXAMPLE OF AGGREGATION • Group cells into clusters (illustrated in figure) • Have 1 fiber connected cell in each cluster • If in each cluster: Excess Capacity > Total Demand X (2 or 3) Then: 1 fiber per cluster is sufficient! Else: Add 1 fiber to cluster After calculations for α = 30% & 126 MB/h: CELLS WORST CASE REQUIRES 10 MORE FIBER 36 WINLAB
CONCLUSIONS AND FUTURE WORK • Feasibility study of a distribution plan in NJ – First order study promising in spite of conservative assumptions on traffic and propagation – system more cost effective than a fiber layout – Most effective in rural areas (where it’s needed) • No prior high speed internet connectivity • No fiber infrastructure • More bandwidth available and better propagation • Same methodology for other states/regions • Further issues that need to be studied: • Detailed routing strategies • Cost/benefit analysis 37 WINLAB
• WHITE SPACE: WHERE ARE WE TODAY? Database Testing and Trials – Google, Microsoft, Spectrum Bridge, Telcordia, etc. – No TVBDs and services rolled out yet! • Wireless Service Providers and TV Broadcasters still continue to resist • Service providers want more licensed spectrum • Broadcasters worry about interference • FCC working on next round of spectrum auctions • Reverse Auctions, Repackaging and Incentive Auctions 38 WINLAB
WHITE SPACE: WHERE WILL WE GO? • “Green” trumps “White”? 39 WINLAB
PILOT PROJECT: “BROADBAND TO BIVALVE” 40 WINLAB
PILOT PROJECT: “BROADBAND TO BIVALVE” • Set up Wi. Fi Hotspots in Bivalve, NJ • Backhaul to Bridgeton, NJ where Internet (T 1) connectivity exists • Use “Fixed Towers” and available TV White Space to provide backhaul as shown in exemplary figure – Could reuse water towers or weather towers as feasible for installing radios – Towers requires power supply • The set-up will also serve as a “research testbed” for protocol and application development to benefit rural areas • If an ISP partner is available, mobile hotspot service could be provided along the way to farms, etc. 41 WINLAB
PILOT PROJECT: “BROADBAND TO BIVALVE” • Hardware: Radio Router Node based on currently available second generation ORBIT platform – Multiple radio interfaces: 802. 11 (wifi), 802. 16 (wimax), LTE, Zig. Bee, Bluetooth, CRKit (whitespace capable) • Software: Whitespace Routing Protocol optimized for throughput – Local (hotspot) support – Caching capabilities 42 WINLAB
References • • C. Gerami, N. B. Mandayam, and L. J. Greenstein, “Backhauling in TV white spaces, ” Proceedings of IEEE GLOBECOM 2010, December 2010 O. Ileri and N. B. Mandayam. Dynamic spectrum access models: Toward an engineering perspective in the spectrum debate. IEEE Communications Magazine, 46(1): 153 -160, January 2008. D. Zhang, R. Shinkuma, N. B. Mandayam, “Bandwidth Exchange: An Energy Conserving Incentive Mechanism for Cooperation” in IEEE Transactions on Wireless Communications, vol. 9, No. 6, pp. 2055 -2065, June 2010 D. Zhang and N. B. Mandayam, “Bandwidth Exchange for Fair Secondary Coexistence in TV White Space, ” in Proceedings of International ICST Conference on Game Theory for Networks (Game. Nets), Shanghai, April 2011 M. N. Islam, N. B. Mandayam, and S. Kompella. Optimal resource allocation in a bandwidth exchange enabled relay network. In Proc. IEEE MILCOM’ 2011, pages 242– 247, November 2011 C. Raman, R. Yates, N. B. Mandayam, ”Scheduling Variable Rate Links via a Spectrum Server” in Proceedings of IEEE Dy. Span 2005, November 2005, Baltimore, MD D. Raychaudhuri, N. B. Mandayam, J. B. Evans, B. J. Ewy, S. Seshan, and P. Steenkiste. Cognet: an architectural foundation for experimental cognitive radio networks within the future internet. In Proc. ACM Mobi. Arch’ 2006 N. Krishnan, R. D. Yates, N. B. Mandayam, J. S. Panchal, “Bandwidth Sharing for Relaying in Cellular Systems” in IEEE Transactions on Wireless Communications, vol. 11, No. 1, pp. 11743 129, January 2012
Acknowledgments • U. S. National Science Foundation • Office of Naval Research • IEEE COMSOC • Debi Siering • WINLAB Collaborators: Cyrus Gerami, Larry Greenstein, Nazmul Islam, Ivan Seskar, Dipankar Raychaudhuri 44
- Slides: 44