[Paper Review] Two-Dimensional Tail-Biting Convolutional Codes

Liam Alfandary, Dan Raphaeli|arXiv (Cornell University)|Sep 18, 2011

Error Correcting Code Techniques6 references3 citations

TL;DR

This paper introduces two-dimensional tail-biting convolutional codes (2D TBCCs), deriving their algebraic structure, minimum distance properties, and designing iterative belief propagation-based decoding algorithms. It demonstrates that 2D TBCCs outperform comparable 1D TBCCs and other codes in word-error rate, achieving performance within 0.2 dB of optimal maximum-likelihood decoding using suboptimal decoding techniques.

ABSTRACT

The multidimensional convolutional codes are an extension of the notion of convolutional codes (CCs) to several dimensions of time. This paper explores the class of two-dimensional convolutional codes (2D CCs) and 2D tail-biting convolutional codes (2D TBCCs), in particular, from several aspects. First, we derive several basic algebraic properties of these codes, applying algebraic methods in order to find bijective encoders, create parity check matrices and to inverse encoders. Next, we discuss the minimum distance and weight distribution properties of these codes. Extending an existing tree-search algorithm to two dimensions, we apply it to find codes with high minimum distance. Word-error probability asymptotes for sample codes are given and compared with other codes. The results of this approach suggest that 2D TBCCs can perform better than comparable 1D TBCCs or other codes. We then present several novel iterative suboptimal algorithms for soft decoding 2D CCs, which are based on belief propagation. Two main approaches to decoding are considered. We first focus on a decoder which extends the concept of trellis decoding to two dimensions. Second, we investigate algorithms which use the code's parity check matrices. We apply conventional BP in the parity domain, but improve it with a novel modification. Next, we test the generalized belief propagation (GBP) algorithm. Performance results are presented and compared with optimum decoding techniques and bounds. The results show that our suboptimal algorithms achieve respectable results, in some cases coming as close as 0.2dB from optimal (maximum-likelihood) decoding. However for some of the codes there is still a large gap from the optimal decoder.

Motivation & Objective

• To develop a theoretical framework for two-dimensional tail-biting convolutional codes (2D TBCCs) using algebraic methods.
• To analyze the minimum distance and weight distribution of 2D TBCCs for improved error correction.
• To design iterative soft-decision decoding algorithms based on belief propagation for 2D CCs.
• To evaluate performance of proposed decoders against optimal decoding benchmarks and identify gaps.

Proposed method

• Derives bijective encoders and parity-check matrices for 2D TBCCs using algebraic techniques in polynomial rings.
• Adapts a tree-search algorithm to two dimensions to identify codes with high minimum distance.
• Proposes two decoding approaches: a 2D trellis-based belief propagation decoder and a parity-domain BP decoder with a novel modification.
• Implements generalized belief propagation (GBP) to improve decoding convergence and performance.
• Uses conventional belief propagation in the parity domain and enhances it with a new iterative update rule.
• Compares performance of suboptimal decoders with maximum-likelihood decoding bounds and other code types.

Experimental results

Research questions

• RQ1 How can the algebraic structure of 2D tail-biting convolutional codes be formalized using polynomial ring theory?
• RQ2 What are the minimum distance and weight distribution properties of 2D TBCCs, and how do they compare to 1D counterparts?
• RQ3 Can iterative belief propagation decoding be effectively extended to two-dimensional convolutional codes?
• RQ4 How close can suboptimal decoding algorithms come to achieving maximum-likelihood performance in 2D CCs?
• RQ5 What improvements can be achieved by modifying belief propagation in the parity domain for 2D codes?

Key findings

• The proposed 2D TBCCs achieve better word-error rate performance than comparable 1D tail-biting convolutional codes and other standard codes.
• A modified belief propagation algorithm in the parity domain improves decoding convergence and performance over conventional BP.
• Generalized belief propagation (GBP) yields better results than standard BP, especially for codes with complex structure.
• The best suboptimal decoding algorithms achieve performance within 0.2 dB of maximum-likelihood decoding for certain code constructions.
• Tree-search optimization successfully identifies 2D TBCCs with high minimum distance, supporting improved error correction capability.
• Performance gaps remain between suboptimal decoders and optimal ML decoding for some code configurations, indicating room for improvement.