Биоинспирированный метод оптимизации кодов на основе преобразования Лаби

Abstract

RU: Обосновано применение кодов на основе преобразования Лаби для обеспечения заданного качества обслуживания в компьютерных сетях. Представлены особенности процессов кодирования и декодирования информации кодами Лаби. Показано, что эффективность кодов на основе преобразования Лаби значительным образом зависит от закона распределения вероятностей, на основе которого формируются кодовые символы. Предложен подход к оптимизации относительно длинных кодов Лаби с использованием обобщенных биоинспирированных процедур. Рассмотрены особенности и основные этапы предложенного метода оптимизации кодов основе преобразования Лаби.

EN: It has been shown that the packet loss problem considerably affects the quality of service users of computer networks. The limitations of classical approaches to the recovery of the lost packets are considered The principles of encoding and decoding fountain codes are given. It has been shown that the feature codes based on Luby transform is the using of a special law (density) of probability distribution for forming code symbols. The features soliton-like laws of probability distribution for Luby codes are considered. It has been shown that the efficiency of Luby codes with robust soliton probability distribution is significantly reduced with decreasing the number of information symbols. The optimization approaches for codes based on Luby transform different lengths are analyzed. The necessity of optimizing the laws of probability distribution for a relatively long Luby codes is justified. The approach to optimization of these codes with using generalized bioinspired procedures is proposed. The general scheme of population bioinspired procedures of search optimization is provided. The features and main stages of the proposed optimization method codes based on Luby transform are considered. During the first stage of the proposed approach to optimization Luby codes generates a set of vectors, which corresponding to different probability distributions. Further encoding information messages with Luby codes, modeling the information transmission over a communication channel and the recovery of the transmitted messages by decoding based on the belief propagation received code symbols are performed. Then search the best probability distribution for specified parameters Luby code with the help of generalized bioinspired procedures is carried out. The formal presentation of search "good" probability distribution for Luby codes in the form of non-linear programming problem is considered. It has been shown that as the objective function is advisable to take the amount of Luby codes redundancy. The restrictions for the formulated problem of minimizing objective function are provided.