Abstract
It is already known that trellis coded modulation (TCM), constructed by using the encoder of a convolutional code C with short constraint length followed by a convolutional processor and a signal apper, can achieve large squared free distances, where the convolutional processor is a rate-one convolutional code with large constraint length. The previously used decoding results in large error coefficients and hence the obtained coding gain is distant from the asymptotic coding gain.In this paper, we propose to use an expanded trellis of C with some recovered information to decode the TCM. With the proposed decoding, additional coding gain can be obtained as compared to the previously used decoding.