Abstract
Summary form only given, as follows. The authors consider the problem of optimum block-structured communication system design for a bandlimited additive white Gaussian noise channel. The transmitter consists of a block encoder, a signal selection unit, and a modulator. The receiver consists of a demodulator and a source decoder. The objective is to design the source encoder, the modulation signal set, and the decoder so as to minimize the mean-squared error subject to constraints on the average transmitted power and the channel bandwidth. Necessary conditions for optimality are derived; however, it is difficult to solve these necessary conditions in the most general case. For the special case in which the modulation signal set belongs to the QAM family and for a maximum-likelihood (ML) demodulator, an algorithm that iteratively solves the necessary conditions for optimality for the encoder and the decoder subject to the above-mentioned constraints has been developed. Numerical results are obtained for Gauss-Markov sources and different choices of system parameters. For comparison purposes, the performance of a conventional vector quantization system combined with QAM modulation and ML demodulation has also been determined. These results indicate that significant performance improvements over the conventional vector quantization system can be obtained. Comparisons are made against a system with a linear decoder in which the encoder, signal set, and the linear decoder are optimized. Various channel mismatch issues are studied, and the relative robustness of the above systems is investigated.
Original language | English |
---|---|
Pages | 58-59 |
Number of pages | 2 |
State | Published - 1990 |
Event | 1990 IEEE International Symposium on Information Theory - San Diego, CA, USA Duration: 14 Jan 1990 → 19 Jan 1990 |
Conference
Conference | 1990 IEEE International Symposium on Information Theory |
---|---|
City | San Diego, CA, USA |
Period | 14/01/90 → 19/01/90 |