A novel maximum-likelihood hard- and 8-level soft-decision scarce-state-transition (SST) type syndrome-former error-trellis decoding system of (n,n-l) convolutional codes with coherent BPSK signals for additive white Gaussian noise channels is proposed. The proposed system retains the same number of binary comparisons as the syndrome-former trellis decoding method of Yamada et al. (1983). Like the original SST-type register-exchange Viterbi decoding system, the proposed system also has the same advantage of drawing less power when implemented on CMOS LSI chips. A combination of the two techniques results in a less complex and low power consumption decoding system.

Novel scarce-state-transition syndrome-former error-trellis decoding of (n, n-1) convolutional codes

A computer simulation comparison of the tolerances to additive white Gaussian noise of two maximum-likelihood trellis decoding systems for use on discrete memoryless channels is presented. The first decoding system applies the Viterbi algorithm to the encoder trellis of a convolutional code; this is the well known standard Viterbi decoding system. The second decoding system, proposed by Yamada, uses the same algorithm but applies it to the syndrome-former trellis of the code. High-rate (n, n-1) systematic and nonsystematic convolutional codes, with rates 1/2, 3/4, 5/6 and 7/8 are used throughout the tests. Simulation results are presented for hard- and soft-decision decoding with BPSK modulation and coherent detection. Results show that the Viterbi and Yamada decoding systems give identical error performance for the same code. The implementation complexity of the systems is also examined; a useful reduction in the number of binary comparisons required by the Yamada system can be achieved.

L.H.C. Lee

Papers

Error performance of maximum-likelihood trellis decoding of (n, n-1) convolutional codes: a simulation study