Message Passing Algorithm for Iterative Decoding of Channel Codes

Packet data broadcast systems require error free delivery of packet data without retransmission requests. Turbo codes can be used to achieve very low packet error rates. To achieve this low error rate with a very small gap from the system capacity, several design considerations need to be made. We need to use higher constraintlength codes (increasing complexity exponentially) with larger block sizes. Efficient interleavers need to be used and a larger number of iterations needs to be performed. To reduce the complexity and delay at the decoder, smaller constraint-length codes with smaller block sizes can be used with random interleavers and the number of iterations can be fixed. This, however, results in larger gaps from the capacity. This paper proposes a simple yet powerful two-tier channel coding scheme that has the potential to reduce the gap from capacity (from 5 dB to 4.1 dB, for target packet error 1× 10−4), while using smaller sized packets and smaller constraint length codes with random interleavers. The use of multiple packet detectors identified with the proposed scheme, reduces the packet error floor effect and even very low packet error rates like 1 × 10−6 can be targeted maintaining a modest gap of 4 dB where other schemes exhibit significantly large gap values due to packet error rate floors. The improvement in the gap is nearly 2-3 dB in this region.