Method of parametric adaptation of parallel and sequential turbo codes using neural networks

• Date: 2026
• Main Author: Zaitseva, Liliia
• Format: Article
• Language: Ukrainian
• Published: Kyiv National University of Construction and Architecture 2026
• Subjects: 5G; 6G; wireless technologies; turbo codes; neural networks; multi-layer perceptron; decoding uncertainty
• Online Access: https://es-journal.in.ua/article/view/358189
• Journal Title: Environmental safety and natural resources

The work is devoted to the study of increasing the efficiency of functioning of modern wireless technologies 5G and 6G. The article presents a method of parametric adaptation of parallel and sequential turbo codes using neural networks of the multilevel perceptron type and the decoding uncertainty indicator.The use of neural networks of the multilevel perceptron type for adjusting the external logarithmic ratios of the likelihood functions of probabilistic algorithms for decoding parallel and sequential turbo codes is considered.Turbo codes are decoded using the maximum a posteriori probabilities (MAP) decoding algorithm, which calculates the posterior probability of each decoded symbol, minimizing the probability of an information symbol (bit) error.The aim of the work is to develop a method for parametric adaptation of parallel and sequential turbo codes using neural networks of the multilevel perceptron type and a decoding uncertainty indicator.The use of the decoding uncertainty indicator for parallel and sequential turbo codes at the training stage when determining the weight coefficients of the weight matrix and when functioning of neural networks is proposed.