GAC-decomposition based soft decoding of block codes for multimedia data

Performance of a new nonalgebraic maximum-likelihood decoding method of block codes based on generalized array code (GAC) structure is studied. Conventionally, algebraic method is widely used in decoding of block codes, which emits soft-decision information in demodulation, thus degrading the system performance greatly. If soft-decision decoding is adapted, the complexity is intractable, especially for long block codes. We propose to use GAC decomposition to decrease the complexity of block codes. Because GAC has a regular structure, the soft-decision trellis decoding becomes feasible and easy. Simulation results show that by using decomposition method, code (7, 4, 3) and code (15, 5, 7) can obtain a coding gain of up to 24 dB and 34 dB, respectively in comparison with uncoded BPSK at BER around 10^-5. Furthermore, the method is proven to be effective in compressed image transmission system under fading environment, where it makes image transmission quality significantly improved.