A Study on Large-Scale Signal Detection Using Gaussian Belief Propagation in Overloaded Interleave Division Multiple Access

Abstract

With the progress of IoT, an explosive increase in the number of devices connected to the Internet is predicted. Among multiple access techniques for the IoT, non-orthogonal multiple access (NOMA) has been attracting attention. Interleave division multiple access (IDMA), which is one of the NOMA techniques, detects transmitted data sequences based on a property that each terminal uses its own unique interleave pattern. In this paper, we apply Gaussian belief propagation (GaBP) to IDMA signal detection of hundreds of users, and evaluate the detection performance when the repetition code rate is larger than the inverse of the number of users. In addition, the performance when terminating the iteration for the error-free users using the cyclic redundancy check (CRC) is also evaluated. The simulation results show that the GaBP with CRC achieves both good BER performance and reduction of the number of iterations in the overloaded case.