1. 1 Spectrum Project Overview Roy Yates (~Chris Rose) NSF PI Meeting January 13, 2004 Mandayam Raychaudhuri Rose Spasojevic Yates Rutgers Bauer Wildman Michigan State Friedman Cornell

2. 2 Project Motivation  What everyone agrees on:  Spectrum use is inefficient  FCC licensing has yielded false scarcity

3. 3 Proposed New Methods  Open Access (Commons)  [Shepard, Reed, Benkler, Noam …]  Agile wideband radios will dynamically share a commons  Spectrum Property Rights  [Coase … Hazlett, Faulhaber+Farber]  Owners can buy/sell/trade spectrum  Flexible use, flexible technology, flexible divisibility, transferability

4. 4 Spectrum Property Rights (The triumph of economics)  Current allocations are inefficient.  A spectrum market will (by the force of economics) yield an efficient solution  Q: What should the rights be?

5. 5 Property Rights Proposals [Faulhaber & Farber]  Fee-simple ownership  People can buy/sell/lease specific frequencies in specific locations with the right to exclusive use  Fee-simple ownership with non- interference easement  People can buy/sell/lease specific frequencies in specific locations with the right to emit anytime without interference  Others (non owners) can transmit as long as they don’t meaningfully interfere  Examples: UWB, agile radios that can vacate fast

6. 6 Spectrum Clearinghouse/Spot Market [Noam]  Packets are sent with access tokens  Token specifies channel  Token prices?  would depend on congestion  determined by an automated clearinghouse  Assured access available in futures mkt

7. 7 Open Access (the triumph of technology)  Technology Panacea  Spread spectrum, UWB  Agile adaptive transmitters  Short range communications  Ad hoc multi-hop mesh networks  Minor technical rules for transceivers  power, spreading  The optimistic view: Licensed spectrum will be unnecessary  Q: What are limits of technology?

8. 8 Open Access Technical arguments against  Partially developed theory  ad hoc network capacity,  without mobility [Gupta, Kumar]  with mobility [Grossglauser,Tse]  IT relay and interference channel  Infant technology  UWB, antenna arrays  Transmitter agility  Technology not separable from user assumptions  Capabilities of technology vary with cooperation

9. 9 MAC Channel N transmitters, One Receiver,

10. 10 Capacity Allocation 1 Receiver, 8 Transmitters  Above policies have same capacity  Spreading makes no difference  Agility makes no difference  Strong (unreasonable?) assumptions on technology  May depend on cooperation  Multiple (2) Receivers?  See Rose & Popescu, Globecom 03  Punchline: Iterative waterfilling is often very bad. Signal Space Dimensions Rec’d Power

11. 11 Property Rights vs Commons  Transaction Costs  Direct costs of buyers and sellers  High for property rights. Zero for commons.  Indirect costs of dispute resolution  Low for property rights? High for commons?  Scarcity  If spectrum is scarce:  commons  tragedy of the commons  Property rights will efficiently allocate resources  If spectrum is not scarce,  Commons OK without costs of property rights mgmt

12. 12 Is Spectrum Scarce? Some UWB Observations  UWB Mask: -41 dBm/Mhz (3-10 GHz)  10 -4 mW/MHz  2  W for 20 MHz

13. 13 UWB Spatial Interference Poisson field of UWB transmitters

14. 14 WiFi Coverage

15. 15 WiFi Coverage (UWB Inteference)  UWB Interference  Effects are cumulative, but does it matter?  Random Sum of RV’s [Sousa+Silvester, 1990]

16. 16 WiFi Coverage  N UWB sources: E[N]=  R 2  Interference (per user): E[I]=K/R 2 r 0 2  Total Interference: I T =E[N]E[I}=  r 0 2  Traditional user at dist R  20 MHz BW,  20 mW xmit Power  Channel  P R (r) = K/r 4 r0r0 R

17. 17 WiFi Coverage  SIR dB = 35 +20log(r 0 )-10log( )-40log(R)  R=10, r 0 =1,  m 2 : SIR=5 dB (Yuck)  R=10, r 0 =1,  m 2 : SIR= 25 dB (OK!)  Densities depend on future applications r0r0 R  Traditional user at dist R  20 MHz BW,  20 mW xmit Power  P 0 = 20 mW/2  W =10 4

18. 18 Spectrum Scarcity?  Usable NLOS spectrum = 10 GHz  Advances in Moore’s Law   devices will use all 10 GHz  Capacity increases with infrastructure  Reducing R increases capacity (frequency reuse)  Capacity increases with technology?  Analogy of the day:  Oil has been scarce for 100 years, but  Technology advances keep yielding new supplies

19. 19 Spectrum Project Technical Approaches I  Approaches/methods/solutions depend strongly on radio technology  Comm Theory approaches:  Chris Rose  Interference Adaptive Transmitters, Multibase iterative waterfilling  Eric Friedman  Fair Share Rate allocation for Gaussian MAC channel  Predrag Spasojevic  Achievable Rates via cooperative coded transmissions

20. 20 Spectrum Project Technical Approaches II  Network Protocol Approaches:  Narayan Mandayam  Cooperation via pricing for wideband, energy constrained wireless network nodes  Utility maximization, rewards for packet forwarding  D. Raychaudhuri  Network protocols/etiquettes for sharing/cooperation  Roy Yates  Coarse packet based measurements

21. 21 Conclusion  Project Goal:  Connect technical approaches to regulatory policy proposals (Models for scarcity?)  FCC ?  Personal View: FCC actions correctly reflect uncertainty of theory/technology  FCC 02-289 Nov 28, 2003  Notice of Inquiry: “Establishment of an Interference Temperature Metric to Quantify and Manage Inteference and to Expand Available Unlicensed Operation in Certain Fixed, Mobile and Satellite Frequency Bands”