Iterative decoding of the product codes can be performed by soft-in/soft-out decoder, which is based on the Chase algorithm and produces the extrinsic information for the next decoding step. This decoding scheme, called block turbo code(BTC), is quite similar to turbo codes, but makes it possible the implementation of decoder with lower complexity and comparable performance.
In thins thesis, we perform the iterative decoding of product codes composed of extended BCH codes. The decoding complexity of the product codes can be reduced by removing unnecessary test pattern in the Chase algorithm using the adaptive error pattern. The proposed decoder with adaptive error pattern computes the simple and efficient threshold value at each iterations. If the threshold value is satisfied the predefined criterion, the decoder will decrease the error pattern for the next iterative decoding step by 1. With the help of adaptive error pattern scheme, we extend the decoder to terminate the decoding with stopping criterion. Therefore we can achieve the low complexity BTC decoder.