Iterative decoding of turbo convolutional codes over the noisy two-user binary adder channel

Introduction: The two-user binary adder channel (2-BAC) is one of the simplest models for a multiple access channel [1]. The 2-BAC is a memoryless channel which, at each time interval, accepts two binary inputs, one from each user. Unless stated otherwise, we will assume that the binary alphabet {0, 1} of each user is mapped as 0 → 1 and 1 → −1 to feed the 2-BAC and that in the absence of noise the 2-BAC output is given by the arithmetic sum of its two inputs, i.e., the output symbols belong to the alphabet {−2, 0, 2}. In the noisy case the 2-BAC output is described by a conditional probability distribution [1]. Our purpose in this letter is to investigate the performance of a coding scheme for the 2-BAC employing just a pair of distinct turbo convolutional codes, not necessarily constituting a uniquely decodable code pair.