A modified Viterbi decoder based upon cross correlation for use in bandwidth efficient systems

A novel convolutional decoding technique is described which improves error performance of synchronous data links operating in a high radio interference environment. The decoder has as its basis the Viterbi algorithm but decoding decisions are based upon maximum correlation rather than maximum log likelihood ratio. The radio interference originates from data transmitted on overlapping QPSK carriers. First the theoretical aspects of the carriers orthogonality were investigated. It was found that the raised cosine shaping of the data symbols removed the orthogonality which had been established between the carriers carrying symbols of rectangular shape. The theoretical aspects of the decoding process were examined resulting in a precise definition of the technique together with clarification as to how reduction in transmission errors can be achieved. The most relevant parameter for the analysis of this decoder is defined as a decoding variable associated with each state of the conventional decoder. This variable describes the contents of the radio interference and has 12 distinct values for two overlapping QPSK signals. The performance of the decoding process is compared with that obtained from 8 levels of quantized soft decoding using the Viterbi algorithm. Computer simulations are presented that show that when both types of decoder are operating in extreme radio interference conditions such as in CDMA systems, the new decoder BER performance is superior to that of the conventional Viterbi decoder. But when the decoding signal is corrupted by Gaussian noise only then both decoders exhibit exactly the same BER