Page 1
Noise calibration, delay coherence and SNR walls forsignal detection
Rahul Tandra Anant Sahai
Wireless Foundations CenterDepartment of Electrical Engineering and Computer Sciences
University of California, Berkeley
15th October, DySpAN 2008
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 2
Problem setup
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 3
Spectrum sensing: traditional picture
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 4
Spectrum sensing: traditional picture
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 5
Spectrum sensing: traditional picture
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 6
Spectrum sensing: noise uncertainty
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 7
Spectrum sensing: noise uncertainty
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 8
Spectrum sensing: noise uncertainty
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 9
Spectrum sensing: SNR walls
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 10
Impact of SNR walls — sensing overhead
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 11
Narrowband pilot signals
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 12
Narrowband pilot signals
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 13
Narrowband pilot signals
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 14
Cyclostationary signals
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 15
Cyclostationary feature detection
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 16
Cyclostationary feature detection
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 17
Cyclostationary feature detection
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 18
Feature detection: single-tap Gauss-Markov fading
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 19
Feature detection: single-tap Gauss-Markov fading
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 20
Feature detection: single-tap Gauss-Markov fading
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 21
Feature detection: multi-tap Gauss-Markov fading
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 22
Feature detection: multi-tap Gauss-Markov fading
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 23
Feature detection: multi-tap Gauss-Markov fading
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 24
Feature detection: finite-delay fading process
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 25
Feature detection: finite-delay fading process
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 26
Phase coherence vs Delay coherence
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 27
Phase coherence vs Delay coherence
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 28
Phase coherence vs Delay coherence
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 29
Phase coherence vs Delay coherence
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 30
Noise calibration: pilot case
fp
Pilot
f
fp fm
f
Band pass filters
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 31
Noise calibration: cyclostationary feature detection
AGC calibration:
Y [n] =1√
1N
∑Nn=1 |Y [n]|2
Y [n]
As N →∞
1√1N
∑Nn=1 |Y [n]|2
→
{1σa
Under H01√P+σ2
a
Under H1
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 32
Robustness gains from noise calibration
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 33
Robustness gains from noise calibration
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai
Page 34
Robustness gains from noise calibration
Colored noise model:
W [n] = M [n]− βM [n− 1],
M [n] is iid white Gaussian noise, β ∈ [0, βmax].
UC Berkeley Wireless Foundations Rahul Tandra, Anant Sahai