Online bit error rate estimation-based stopping criterion for turbo decoding

Iterative turbo decoding is crucial for achieving superior bit error rate (BER)performance. Nevertheless, each subsequent decoding iteration suffers from a high complexity in decoding system latency. Thus, convergence and non-convergenceoutput(CNCO) stopping criteria (CNCOSC) were developed. These s...

Full description

Saved in:

Bibliographic Details
Main Author:	Mohamad, Roslina
Format:	Thesis
Language:	English
Published:	2016
Subjects:	Bit error rate Coding theory
Online Access:	http://psasir.upm.edu.my/id/eprint/70300/1/FK%202016%2046%20-%20IR.pdf
Tags:	Add Tag No Tags, Be the first to tag this record!

id	my-upm-ir.70300
record_format	uketd_dc
spelling	my-upm-ir.703002019-08-28T06:16:49Z Online bit error rate estimation-based stopping criterion for turbo decoding 2016-04 Mohamad, Roslina Iterative turbo decoding is crucial for achieving superior bit error rate (BER)performance. Nevertheless, each subsequent decoding iteration suffers from a high complexity in decoding system latency. Thus, convergence and non-convergenceoutput(CNCO) stopping criteria (CNCOSC) were developed. These stopping criteria can terminate with an optimal average iteration number (AIN) at various signal-tonoise ratios (SNRs). However, the threshold computation and termination rules in CNCOSC require an accurately estimated SNR, thereby increasing the complexity of the receiver. Thus, the aim of this thesis is to develop a low complexity and robust stopping criterion, referred to as the online BER estimation (OBE) stopping criterion(OBEsc), that works in a varying SNR environment and SNR mismatch, without requiring the knowledge of channel SNR. To achieve this particular target, the convergence and non-convergence behaviours of BER in iterative decoding are investigated. In addition, the enhancement of CNCO detection is formulated using the OBE algorithm. The study then develops BER thresholds calculation to determine the correct thresholds according to a given turbo code structure. Finally, termination rules based on the enhanced CNCO detection and BER thresholds are developed.The results show that the OBEsc is capable of detecting the correct CNCO by achieving a lower AIN performance (the lowest AIN = 1) at a varying SNR environment than the benchmark stopping criterion (Bsc) while maintaining the BER performance. OBEsc is also capable of coping with the SNR mismatch by saving approximately 85.71% AIN compared to Bsc, and maintaining the earliest termination at low SNRs compared to the well-known CNCOSC. Furthermore, OBEsc has a better BER performance and faces a smaller BER performance degradation (less than 0.5 dB)than CNCOSC. This shows that OBEsc is capable of operating as a robust stopping criterion without requiring SNR estimation in its stopping rule. In terms of time taken for the predefined threshold simulation, the OBEsc possesses the lowest execution time of 1.59x104 seconds. The computational complexity of the OBEsc is the second lowest complex, only requiring around 2N+2 to 2N+14 real operations compared to the lowest and highest complexity of CNCOSC, which are N+1 and 7N+28 real operations,respectively. In addition, the OBEsc does not require the SNR estimator. Thereby, it significantly reduces complexity at the receiver compared to CNCOSC. The robust performance of OBEsc indicates that it is better suited for use with a turbo decoder than CNCOSC in a varying SNR environment. At the same time, OBEsc can reduce the complexity in the receiver and decrease the delay in turbo iterative decoding. Bit error rate Coding theory 2016-04 Thesis http://psasir.upm.edu.my/id/eprint/70300/ http://psasir.upm.edu.my/id/eprint/70300/1/FK%202016%2046%20-%20IR.pdf text en public doctoral Universiti Putra Malaysia Bit error rate Coding theory
institution	Universiti Putra Malaysia
collection	PSAS Institutional Repository
language	English
topic	Bit error rate Coding theory
spellingShingle	Bit error rate Coding theory Mohamad, Roslina Online bit error rate estimation-based stopping criterion for turbo decoding
description	Iterative turbo decoding is crucial for achieving superior bit error rate (BER)performance. Nevertheless, each subsequent decoding iteration suffers from a high complexity in decoding system latency. Thus, convergence and non-convergenceoutput(CNCO) stopping criteria (CNCOSC) were developed. These stopping criteria can terminate with an optimal average iteration number (AIN) at various signal-tonoise ratios (SNRs). However, the threshold computation and termination rules in CNCOSC require an accurately estimated SNR, thereby increasing the complexity of the receiver. Thus, the aim of this thesis is to develop a low complexity and robust stopping criterion, referred to as the online BER estimation (OBE) stopping criterion(OBEsc), that works in a varying SNR environment and SNR mismatch, without requiring the knowledge of channel SNR. To achieve this particular target, the convergence and non-convergence behaviours of BER in iterative decoding are investigated. In addition, the enhancement of CNCO detection is formulated using the OBE algorithm. The study then develops BER thresholds calculation to determine the correct thresholds according to a given turbo code structure. Finally, termination rules based on the enhanced CNCO detection and BER thresholds are developed.The results show that the OBEsc is capable of detecting the correct CNCO by achieving a lower AIN performance (the lowest AIN = 1) at a varying SNR environment than the benchmark stopping criterion (Bsc) while maintaining the BER performance. OBEsc is also capable of coping with the SNR mismatch by saving approximately 85.71% AIN compared to Bsc, and maintaining the earliest termination at low SNRs compared to the well-known CNCOSC. Furthermore, OBEsc has a better BER performance and faces a smaller BER performance degradation (less than 0.5 dB)than CNCOSC. This shows that OBEsc is capable of operating as a robust stopping criterion without requiring SNR estimation in its stopping rule. In terms of time taken for the predefined threshold simulation, the OBEsc possesses the lowest execution time of 1.59x104 seconds. The computational complexity of the OBEsc is the second lowest complex, only requiring around 2N+2 to 2N+14 real operations compared to the lowest and highest complexity of CNCOSC, which are N+1 and 7N+28 real operations,respectively. In addition, the OBEsc does not require the SNR estimator. Thereby, it significantly reduces complexity at the receiver compared to CNCOSC. The robust performance of OBEsc indicates that it is better suited for use with a turbo decoder than CNCOSC in a varying SNR environment. At the same time, OBEsc can reduce the complexity in the receiver and decrease the delay in turbo iterative decoding.
format	Thesis
qualification_level	Doctorate
author	Mohamad, Roslina
author_facet	Mohamad, Roslina
author_sort	Mohamad, Roslina
title	Online bit error rate estimation-based stopping criterion for turbo decoding
title_short	Online bit error rate estimation-based stopping criterion for turbo decoding
title_full	Online bit error rate estimation-based stopping criterion for turbo decoding
title_fullStr	Online bit error rate estimation-based stopping criterion for turbo decoding
title_full_unstemmed	Online bit error rate estimation-based stopping criterion for turbo decoding
title_sort	online bit error rate estimation-based stopping criterion for turbo decoding
granting_institution	Universiti Putra Malaysia
publishDate	2016
url	http://psasir.upm.edu.my/id/eprint/70300/1/FK%202016%2046%20-%20IR.pdf
_version_	1747812806897434624

Online bit error rate estimation-based stopping criterion for turbo decoding

Similar Items