Cochannel interference cancellation using optimized joint equalization

In this paper, single-carrier transmission impaired by cochannel interference is considered, which generally arises in a GSM/EDGE mobile radio network. Here, cochannel interference is created due to the reuse of the frequencies in nearby cells. In the case, that only a single antenna is available at the mobile receiver, the (multiple-input single-output) system is overloaded and equalization with low complexity is difficult. The optimum joint maximum-likelihood sequence estimation (JMLSE) algorithm would require a huge number of channel states to be considered in a corresponding Viterbi algorithm. The usage of a channel shortening filter in combination with a joint M-algorithm for detection was proposed in [1] in order to obtain good performance results at low complexity for single-antenna interference cancellation (SAIC) in the low-to-moderate carrier-to-interference power ratio (CIR) range, but unfortunately the M-algorithm has shown worse performance than joint delayed-decision feedback sequence estimation (JDDFSE) or joint reduced-state sequence estimation (JRSSE) for high CIRs. To solve this problem, we consider a combined scheme of JRSSE and the M-algorithm for joint detection of the desired user signal and one or more interferers, which corresponds to a special application of breadth-first sequence detection [2]. For the considered algorithm, each trellis state u is equipped with a list of M (best) paths to be stored and processed forward.