A Tree Search Method for Iterative Decoding of Underdetermined Multiuser Systems

摘要: Application of the turbo principle to multiuser decoding results in an exchange probability distributions between two sets constraints. Firstly, constraints imposed by multiple-access channel, and secondly, individual each users' error control code. A-posteriori computation for first set is prohibitively complex all but a small number users. Several lower complexity approaches have been proposed literature. One class methods based on linear filtering (e.g. LMMSE). A more recent approach compute approximations posterior probabilities marginalising over subset sequences (list detection). Most list detection are restricted non-singular systems. In this paper, we introduce transformation that permits application standard tree-search underdetermined We find resulting receiver outperforms existing methods.