Set Level Design for N-Dimensions Using Genetic Algorithms

The transmission of an analog signal using a digital transmission network involves converting the analog signal to a digital signal using a quantizer. In general, uniform level quantizations are used for their simplicity and accuracy. Nonuniform levels, however, provide better signal to noise ratio at the same data transmission rate. Additionally, vector quantization can be used to exploit the correlation of speech signals. Genetic algorithms is an optimization technique that has been used to search for optimal values in a variety of situations. In this paper, genetic algorithms will be used to design optimal quantization levels for both scalar and vector cases.