An Iterative, Turbo-like Approach to Convolutional Decoding

In this paper we present a new method to decode convolutional codes. The method is based on the Turbo principle which was originally proposed to decode parallel concatenated convolutional codes. It becomes possible because of a reinterpretation of the encoder structure. The newly interpreted code is decoded using an iterative scheme with information exchange between sub-decoders. In a second part of the paper, the new method is applied to the decoding of the GSM Full Rate Channel Code. The simulation results show that each iteration can improve the decoding performance while the original BCJR algorithm marks the upper limit.