A reduced complexity algorithm for iterative multiuser detection and decoding of asynchronous users

A reduced complexity algorithm is presented for iterative multiuser detection (MUD) and decoding of asynchronous users. For complexity reduction, multiple soft-input soft-output (SISO) modules are operated in parallel each executing the Verdu (see Multiuser Detection, Cambridge University Press, Cambridge, UK, 1998) optimum detection algorithm on a subset of active users. The novel approach introduced is the application of soft-output broadcasters (SOBC) to combine the soft values generated by parallel MUD-SISOs which model the same user. SOBC modules then exchange soft information with SISO decoders and after the first receiver iteration provide probability values necessary for soft cancellation of the interfering users not modeled in each MUD-SISO. Simulation results show that, after multiple iterations, performance of the reduced-complexity algorithm converges to the single user bound at a slightly degraded SNR threshold relative to the full-complexity algorithm, and acceptable performance is achieved with significant reduction in computational complexity