Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs)

1. Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs) Submission Title: [DAA for Multi-band OFDM UWB] Date Submitted: [ 22 Sept 2005] Source: [Jim Lansford] Company [Alereon] Address [7600C N. Capital of Texas Hwy., Suite 200, Austin, TX 78731] Voice:[ +1 512 228 9786], FAX: [+1 206 337 1703], E-Mail:[jim.lansford@alereon.com] Re: [To be considered in context of TG3a regulatory] Abstract: [Detect and Avoid (DAA) is under discussion by both Japanese and European regulatory agencies as a non-collaborative coexistence technique. DAA requires the receiver to detect the presence of interference and suppress energy in that portion of the band. MB-OFDM is particularly well suited to implement DAA, because the FFT can be used as a channelized radiometer, and the IFFT can be used to sculpt the transmit spectrum.] Purpose: [To inform] Notice: This document has been prepared to assist the IEEE P802.15. It is offered as a basis for discussion and is not binding on the contributing individual(s) or organization(s). The material in this document is subject to change in form and content after further study. The contributor(s) reserve(s) the right to add, amend or withdraw material contained herein. Release: The contributor acknowledges and accepts that this contribution becomes the property of IEEE and may be made publicly available by P802.15. Jim Lansford, Alereon

2. Background • UWB was developed as an “underlay” • FCC bought into the concept • Most of the rest of the world has been skeptical • It’s a coexistence problem • What are the issues? • Detecting the presence of an “incumbent” signal (FWA) • Dropping emissions in the shared band so that interference on the “incumbent” is minimized • This is a form of cognitive radio • Similar to “waterfilling” • TBD: Detection level/confidence and suppression level Jim Lansford, Alereon

3. How does MB-OFDM implement DAA? • Detect: Channelized radiometer • 128 channel FFT inherent in the design • Integrate spectra to achieve desired Pd • Avoid: Bandstop filter (frequency domain) • 128 point IFFT • Additional signal processing techniques can increase notch depth, subject to RF linearity Detect Detect Mixer & Filter Filter/ Filter/ Despread Despread ADC ADC Demap Demap Decimate Decimate Viterbi Viterbi Deinterleave Deinterleave Depuncture Depuncture 90° FFT/ FFT/ Chan Chan Est Est MAC/PHY Interface MAC/PHY Interface CFO CFO IFFT IFFT Equalize Equalize AGC AGC Spreading Spreading Convo Convo - - Mapping Mapping Mixer & Filter lutional lutional Interp Interp . . Interleave Interleave DAC DAC Coder Coder Puncture Puncture /Filter /Filter Jim Lansford, Alereon

4. Detect (1) • Channelized radiometers have been used for decades • Narrowband detection in a wideband channel • Used in Radar and communication intercept receivers • Theory well developed • If FFT bins contain noise alone, distribution is Rayleigh • If narrowband signal + noise, distribution is Rician (●)2 ∫ < > FFT P(S+N) P(N) Jim Lansford, Alereon

5. Detect (2) • Averaging spectra reduces the variance of the distribution • For ensemble average of N spectra, variance within a bin decreases by 1/N, standard deviation by 1/sqrt(N) • Thus, detection probability can be made arbitrarily close to 1 but integration time can become large • There is a considerable body of research for fluctuating signals (such as pulsed beacons) Jim Lansford, Alereon

6. Detect (3) • For MB-OFDM system, FFT resolution is 528/128=4.125MHz • kTB for a single tone is -174+66= -108dBm • Assume 6.6dB NF in RF => N0= -101dBm • FWA characteristics • 5 or 10MHz BW • 4W uplink, Downlink power <-90dBm at UWB system • For FWA downlink, detection probability at -90dBm should be more than adequate – unless DL is in a fade • Possible for devices in piconet to collaborate • Uplink can always be easily detected • Uplink confirms local activity after downlink detected Jim Lansford, Alereon

7. Detect (4) • Detection test setup • This is a demo • Fading channel tests will be done later UWB transmitter UWB Receiver + FWA signal Generator Rhode & Schwartz Spectrum Analyzer Jim Lansford, Alereon

8. Detect (5) Spectrum analyzer Noise floor MB-OFDM signal (zoomed: 200MHz) Jim Lansford, Alereon

9. Detect (6) “FWA” Signal @ -90dBm “FWA” Signal + OFDM Signal (10ms averaging, MB-OFDM @ -81dBm) Jim Lansford, Alereon

10. Avoid (1) • Spectral “notches” have been discussed widely in 802.15.3a in reference to MB-OFDM • Generally, can be considered frequency domain bandstop filters (FIR) • Like any FIR filter, intentional ISI can be introduced to control notch depth and width • EVM of RF chain must be consistent with desired depth • Typically, these notches can be 15-20dB in depth • More advanced techniques are under evaluation to achieve deeper notches Jim Lansford, Alereon

11. Avoid (2) 20MHz MB-OFDM signal with 20MHz notch (5 tones) MB-OFDM signal plus “FWA” inside 20MHz notch Jim Lansford, Alereon

12. Summary • These measurements were made on an actual MB-OFDM system • BER/PER increase from removal of five tones is insignificant • Erasure performance has been well documented • Detection algorithm works, but needs further study • Fluctuating FWA signal • Strategies for downlink+uplink detection • Collaborative coexistence (in a laptop, for example) would work even better • Spectral sculpting really works… Jim Lansford, Alereon

13. References • M.Skolnik, Introduction to Radar Systems, McGraw Hill, 1980 • Johanna Vartiainen, “Concentrated Signal Extraction using Consecutive Mean Excision Algorithms,”Proceedings of the 2005 Finnish Signal Processing Symposium - FINSIG'05, August, 2005, Kuopio, Finland, pp. 87-90. • S. M. Kay, Fundamentals of Statistical Signal Processing: Detection Theory., Prentice Hall, 1998. Jim Lansford, Alereon