Complexity-performance trade-offs in turbo codes for IMT-2000

Turbo codes have been found to achieve very low error rates at signal-to-noise ratios less than 1 dB above the Shannon bound. A drawback of turbo coding is that its superior performance is achieved at the cost of interleaving and iterative decoding, and this can translate to significant delay. In this paper we investigate complexity-performance trade-offs for turbo codes suitable for ITM-2000 systems through examination of parameters including the interleaver type and size, number of decoding iterations, polynomials for the constituent codes, and trellis termination techniques. We show that these parameters cannot be selected in isolation.