[Paper Review] Discovery of good double and triple circulant codes using multiple impulse method

Mohamed Askali, Saïd Nouh+2 more|arXiv (Cornell University)|Jun 22, 2015

Advanced Control Systems Design4 references4 citations

TL;DR

This paper presents an optimized multiple impulse method (MIM) enhanced with genetic algorithms to discover new optimal double and triple circulant codes (DCC & TCC) with improved minimum distances. By using MIM as a minimum distance evaluator within a genetic search framework, the authors identify several new codes with the highest known parameters, significantly advancing the state of the art in structured linear error-correcting codes.

ABSTRACT

The construction of optimal linear block error-correcting codes is not an easy problem, for this, many studies describe methods for generating good error correcting codes in terms of minimum distance. In a previous work, we have presented the multiple impulse method (MIM) to evaluate the minimum distance of linear codes. In this paper we will present an optimization of the MIM method by genetic algorithms, and we found many new optimal Double and Triple Circulant Codes (DCC & TCC) with the highest known parameters using the MIM method as an evaluator of the minimum distance. Two approaches are used in the exploration of the space of generators; the first is based on genetic algorithms, however the second is on the random search algorithm.

Motivation & Objective

• To develop an efficient method for discovering optimal linear block codes with high minimum distance.
• To improve the multiple impulse method (MIM) for accurate minimum distance evaluation in structured codes.
• To apply genetic algorithms to explore the generator space of double and triple circulant codes more effectively.
• To identify new codes with the highest known parameters in terms of minimum distance and rate.
• To compare the performance of genetic algorithms versus random search in the discovery of good codes.

Proposed method

• The multiple impulse method (MIM) is used as a core evaluator for the minimum distance of linear codes.
• Genetic algorithms are applied to explore the space of generator matrices for double and triple circulant codes.
• The MIM method is optimized to improve computational efficiency and accuracy in minimum distance estimation.
• Two search strategies are compared: genetic algorithm-based optimization and random search in the generator space.
• The method evaluates candidate codes by estimating their minimum distance using MIM before selecting high-performing candidates.
• The approach focuses on structured codes—specifically double and triple circulant codes—due to their favorable encoding properties and known performance potential.

Experimental results

Research questions

• RQ1 Can the multiple impulse method be effectively optimized to enhance minimum distance evaluation for structured linear codes?
• RQ2 To what extent can genetic algorithms outperform random search in discovering high-performance double and triple circulant codes?
• RQ3 What is the maximum minimum distance achievable for double and triple circulant codes of given length and dimension?
• RQ4 Are there new optimal codes with previously unknown parameters that can be discovered using this hybrid approach?
• RQ5 How does the integration of MIM with metaheuristic search improve the efficiency of code discovery?

Key findings

• The authors discovered multiple new double and triple circulant codes with the highest known minimum distance parameters for their respective lengths and dimensions.
• The optimized MIM method demonstrated high accuracy and efficiency in estimating minimum distances, enabling reliable code evaluation.
• Genetic algorithms significantly outperformed random search in identifying high-quality codes, especially in larger search spaces.
• Several new optimal codes were identified, extending the known bounds for DCC and TCC in the literature.
• The hybrid approach combining MIM with genetic algorithms proved effective in exploring structured code spaces systematically.
• The study confirms that the integration of advanced search techniques with precise distance evaluation can lead to the discovery of previously unknown optimal codes.