Research On The Immune Algorithm-Based Blind Signal Separation Problems And Its Implementation On DSP

Blind Source Separation (BSS) is a new powerful signal and information processing method. In early 1980's, BSS technology was explored to separate the mixtures of source signals. Nowadays, however, it has attracted a great deal of attention from the signal processing community. Applications are related to a variety of areas such as speech signal recognition, image processing, biomedical signal processing and communication, etc. In general, researches on BSS may be roughly divided into two directions: one is the basic theory and algorithm research; the other is its application in information processing. In this paper, after an introduction to the development history, the current research status of BSS, basic theory was given, including the feasibility, mathematical model of BSS, and criterion function commonly used in BSS. Then a new method based on the immune algorithm is proposed, and implemented on C6713 DSK.The innovations and main works of the paper are listed as follow:1. A novel BSS algorithm based on the immune algorithm and kurtosis is proposed. Immune algorithm (IA) was an new technology of information processing, which extracted from the characteristics of biological immune system, such as self-adaptation, self-organization, diversity, immune memory, etc. It was a hotspot after article neural network(ANN) and evolve calculation (EC) in the field of calculated intelligence.This algorithm can improve the convergence velocity and avoid getting stuck at local optima in traditional algorithms. The results shows that the proposed method can separate independent sources from their linear mixtures of sub-Gaussian, super-Gaussian sources.2. To meet the requirement for data large dynamic range, high precision of IA_based BSS algorithm, TI high performance 32-bit floating-point C6713 DSK is selected. The software system design based on above hardware platform, including the DSP initialization program, the interrupt request service and speech separation. It also proposed some effective optimized methods according to the characteristics of the algorithm and DSP chip. The experiment shows, the algorithm can considerably reduce the calculation burden and the output speech signals are well separated.