on the shamai laroia approximation for the information
play

On the Shamai-Laroia Approximation for the Information Rate of the - PowerPoint PPT Presentation

On the Shamai-Laroia Approximation for the Information Rate of the ISI Channel Yair Carmon and Shlomo Shamai (Shitz) Department of Electrical Engineering, Technion - Israel Institute of Technology 2014 Information Theory and Applications


  1. On the Shamai-Laroia Approximation for the Information Rate of the ISI Channel Yair Carmon and Shlomo Shamai (Shitz) Department of Electrical Engineering, Technion - Israel Institute of Technology 2014 Information Theory and Applications Workshop San Diego, USA February 2014 Acknowledgment: Prof. Tsachy Weissman, FP7 Network of Excellence in Wireless COMmunications NEWCOM#, Israel Science Foundation (ISF). Y. Carmon and S. Shamai On the Shamai-Laroia Approximation ITA 2014 1 / 31

  2. Outline Introduction 1 ISI Channel and I.I.D. Information Rate Analytical Lower Bounds on the Information Rate The Shamai-Laroia Conjecture (SLC) Low SNR Analysis 2 Counterexamples 3 High SNR Analysis 4 Conclusion 5 Y. Carmon and S. Shamai On the Shamai-Laroia Approximation ITA 2014 2 / 31

  3. Introduction ISI Channel and I.I.D. Information Rate Inter-Symbol Interference Channel Model Input-output relationship: L − 1 � y k = h l x k − l + n k l =0 Input Sequence x ∞ −∞ is assumed i.i.d , with P x = Ex 2 0 Noise Sequence n ∞ −∞ is Gaussian and i.i.d., with N 0 = En 2 0 Inter-symbol interference (ISI) coefficients h L − 1 0 Channel frequency response H ( θ ) = � L − 1 k =0 h k e − jkθ The simplest (non-discrete) model for a channel with memory Ubiquitous in wireless and wireline communications Results presented here extend straightforwardly to a complex setting Y. Carmon and S. Shamai On the Shamai-Laroia Approximation ITA 2014 3 / 31

  4. Introduction ISI Channel and I.I.D. Information Rate Inter-Symbol Interference Channel Model Input-output relationship: L − 1 � y k = h l x k − l + n k l =0 Input Sequence x ∞ −∞ is assumed i.i.d , with P x = Ex 2 0 Noise Sequence n ∞ −∞ is Gaussian and i.i.d., with N 0 = En 2 0 Inter-symbol interference (ISI) coefficients h L − 1 0 Channel frequency response H ( θ ) = � L − 1 k =0 h k e − jkθ The simplest (non-discrete) model for a channel with memory Ubiquitous in wireless and wireline communications Results presented here extend straightforwardly to a complex setting Y. Carmon and S. Shamai On the Shamai-Laroia Approximation ITA 2014 3 / 31

  5. Introduction ISI Channel and I.I.D. Information Rate Mutual Information Rate Given by 1 2 K + 1 I ( y K − K ; x K − K ) = I ( y ∞ −∞ ; x 0 | x − 1 I = lim −∞ ) K →∞ Is the rate of reliable communications achievable by a random code with codewords distributed as x ∞ −∞ For Gaussian input I has a simple expression � π � � I Gaussian = 1 1 + P x | H ( θ ) | 2 log dθ 2 π N 0 − π When the input is distributed on a finite set (constellation), no closed form expression is known Y. Carmon and S. Shamai On the Shamai-Laroia Approximation ITA 2014 4 / 31

  6. Introduction ISI Channel and I.I.D. Information Rate Approximating the I.I.D. Information Rate Two main ways to investigate I : 1 Approximations and bounds based on Monte-Carlo simulations Provide the best accuracy But little theoretic insight High computational complexity, that grows quickly with the number of dominant ISI taps c.f. [Arnold-Loeliger-Vontobel-Kavcic-Zeng’06] 2 Analytical lower bounds Not as tight as their simulation-based counterparts But much easier to compute Useful in benchmarking communication schemes and other bounds May provide theoretical insight Y. Carmon and S. Shamai On the Shamai-Laroia Approximation ITA 2014 5 / 31

  7. Introduction ISI Channel and I.I.D. Information Rate Approximating the I.I.D. Information Rate Two main ways to investigate I : 1 Approximations and bounds based on Monte-Carlo simulations Provide the best accuracy But little theoretic insight High computational complexity, that grows quickly with the number of dominant ISI taps c.f. [Arnold-Loeliger-Vontobel-Kavcic-Zeng’06] 2 Analytical lower bounds Not as tight as their simulation-based counterparts But much easier to compute Useful in benchmarking communication schemes and other bounds May provide theoretical insight Y. Carmon and S. Shamai On the Shamai-Laroia Approximation ITA 2014 5 / 31

  8. Introduction Analytical Lower Bounds on the Information Rate Data Processing Lower Bounds For any sequence of coefficients a ∞ −∞ , −∞ ) ≥ I ( � −∞ ; x 0 | x − 1 k a k y − k ; x 0 | x − 1 I = I ( y ∞ −∞ ) Can be simplified to, I ≥ I ( x 0 ; x 0 + � k ≥ 1 α k x k + m ) � �� � additive noise term with α k = � l a l h − l − k / � l a l h − l � l / ( � l a l h − l ) 2 ) independent of x 0 l a 2 and m ∼ N (0 , N 0 Different choices of a ∞ −∞ yield different bounds Y. Carmon and S. Shamai On the Shamai-Laroia Approximation ITA 2014 6 / 31

  9. Introduction Analytical Lower Bounds on the Information Rate Data Processing with a Sample Whitened Matched Filter a ∞ −∞ are chosen so that α k = 0 for every k > 0 (non-causal ISI eliminated) In this case the noise term is purely Gaussian The resulting bound was first proposed in [Shamai-Ozarow-Wyner’91]: I ≥ I ( x 0 ; x 0 + m ) = I x ( SNR ZF-DFE ) with I x ( γ ) the MI of a scalar Gaussian channel at SNR γ with input x 0 SNR ZF-DFE the output SNR of the zero-forcing decision feedback equalizer (DFE): � 1 � π � SNR ZF-DFE = P x � | H ( θ ) | 2 � exp log dθ N 0 2 π − π Very simple, but quite loose in medium and low SNR’s Y. Carmon and S. Shamai On the Shamai-Laroia Approximation ITA 2014 7 / 31

  10. Introduction Analytical Lower Bounds on the Information Rate Data Processing with a Mean Square WMF Choose a ∞ −∞ so that the noise term has minimum variance I ≥ I ( x 0 ; x 0 + � ) � I MMSE k ≥ 1 ˆ α k x k + ˆ m � �� � min variance A tight bound in many cases Still difficult to compute and analyze Some techniques for further bounding were proposed Using probability-of-error bounds and Fano’s inequality [Shamai-Laroia’96] Using a mismatched mutual information approach [Jeong-Moon’12] However, none of the resulting bounds is both simple and tight Y. Carmon and S. Shamai On the Shamai-Laroia Approximation ITA 2014 8 / 31

  11. Introduction Analytical Lower Bounds on the Information Rate Data Processing with a Mean Square WMF Choose a ∞ −∞ so that the noise term has minimum variance I ≥ I ( x 0 ; x 0 + � ) � I MMSE k ≥ 1 ˆ α k x k + ˆ m � �� � min variance A tight bound in many cases Still difficult to compute and analyze Some techniques for further bounding were proposed Using probability-of-error bounds and Fano’s inequality [Shamai-Laroia’96] Using a mismatched mutual information approach [Jeong-Moon’12] However, none of the resulting bounds is both simple and tight Y. Carmon and S. Shamai On the Shamai-Laroia Approximation ITA 2014 8 / 31

  12. Introduction The Shamai-Laroia Conjecture (SLC) The Shamai-Laroia Conjecture [Shamai-Laroia’96] conjectured that I MMSE is lower bounded by replacing x ∞ 1 with g ∞ 1 , i.i.d. Gaussian of equal variance: I MMSE = I ( x 0 ; x 0 + � k ≥ 1 ˆ α k x k + ˆ m ) ≥ I ( x 0 ; x 0 + � m ) = I x ( SNR MMSE-DFE-U ) � I SL k ≥ 1 ˆ α k g k + ˆ SNR MMSE-DFE-U is the output SNR of the unbiased MMSE DFE: � 1 � π � � � 1 + P x | H ( θ ) | 2 SNR MMSE-DFE-U = exp log dθ − 1 2 π N 0 − π I SL — a simple, tight and useful approximation for I MMSE , I Y. Carmon and S. Shamai On the Shamai-Laroia Approximation ITA 2014 9 / 31

  13. Introduction The Shamai-Laroia Conjecture (SLC) The Shamai-Laroia Conjecture — Example BPSK input, h = [0 . 408 , 0 . 817 , 0 . 408] (moderate ISI severity) 1.2 I MMSE I SLC Gaussian upper bound Shamai-Ozarow-Wyner lower bound 1 0.8 Information [bits] 0.6 0.4 0.2 0 −10 −5 0 5 10 15 20 P x /N 0 [dB] Y. Carmon and S. Shamai On the Shamai-Laroia Approximation ITA 2014 10 / 31

  14. Introduction The Shamai-Laroia Conjecture (SLC) The Shamai-Laroia Conjecture — Example, cont’ BPSK input, h = [0 . 408 , 0 . 817 , 0 . 408] (moderate ISI severity) 0.82 1 0.99 0.8 0.98 0.78 Information [bits] Information [bits] 0.97 0.76 0.96 0.95 0.74 0.94 0.72 0.93 10 11 12 13 14 15 16 4.6 4.8 5 5.2 5.4 5.6 5.8 6 6.2 P x /N 0 [dB] P x /N 0 [dB] I MMSE I SLC Gaussian upper bound Shamai-Ozarow-Wyner lower bound Y. Carmon and S. Shamai On the Shamai-Laroia Approximation ITA 2014 11 / 31

  15. Introduction The Shamai-Laroia Conjecture (SLC) The “Strong SLC” and its Refutation A stronger version of the SLC reads I ( x 0 ; x 0 + � k ≥ 1 α k x k + m ) ≥ I ( x 0 ; x 0 + � k ≥ 1 α k g k + m ) for every α ∞ α ∞ −∞ and m (not just ˆ −∞ and ˆ m ) [Abbe-Zheng’12] gave a counterexample Based on a geometrical tool using the Hermite polynomials Cannot straightforwardly refute the original SLC Uses continuous-valued input distributions (finite alphabet is more interesting) Y. Carmon and S. Shamai On the Shamai-Laroia Approximation ITA 2014 12 / 31

  16. Introduction The Shamai-Laroia Conjecture (SLC) The “Strong SLC” and its Refutation A stronger version of the SLC reads I ( x 0 ; x 0 + � k ≥ 1 α k x k + m ) ≥ I ( x 0 ; x 0 + � k ≥ 1 α k g k + m ) for every α ∞ α ∞ −∞ and m (not just ˆ −∞ and ˆ m ) [Abbe-Zheng’12] gave a counterexample Based on a geometrical tool using the Hermite polynomials Cannot straightforwardly refute the original SLC Uses continuous-valued input distributions (finite alphabet is more interesting) Y. Carmon and S. Shamai On the Shamai-Laroia Approximation ITA 2014 12 / 31

Recommend


More recommend