Modified Algorithm for Decoding Convolutional Turbo-Codes

Abstract

At the current stage of development of the theory of error-correcting codes, a turbo-code is a promising technique, which provides the necessary error-correcting characteristics for various data transmission systems. It represents a powerful class of error-correcting codes with a unique codec structure, capable of operating effectively at high speeds in channels with low useful energy, almost completely using channel capacity by the principles of K. Shannon. The process of turbo-decoding using specialized iterative probabilistic algorithms requires significant computational resources, which can become a limitation for real digital data transmission systems. The article proposes a suboptimal modification of the turbo-decoding algorithm based on the maximum a posteriori likelihood in an iterative form, which uses the approximation of the correction function during the calculation of the correlation metrics of the trellis-diagram transitions of a recursive systematic convolutional code. The proposed function outperforms analogues in the base interval of values by statistical indicators.

The algorithm's efficiency was demonstrated during computer simulation of the work process of a digital data transmission system on random frames and multimedia data. The experimental functional dependence of the bit error rate on the signal/noise ratio in the communication channel was found and analyzed. It was determined that the modified algorithm outperforms several analogues in the terms of energy gain in communication channels with a high level of noise component, its efficiency is as close as possible to the basic algorithm, but has lower computational complexity. The article can be useful for engineers and designers of data transmission systems since it allows more efficient analysis and synthesis of turbo-codes depending on the given tasks.