Multi-User Detection of Overloaded Systems with Low-Density Spreading

Abstract

Future wireless networks will have applications that require many devices to be connected to the network. Non-orthogonal multiple access (NOMA) is a promising multiple access scheme that allows more users to simultaneously transmit in a common channel than orthogonal signaling techniques. This overloading allows for high spectral efficiencies which can support the high demand for wireless access. One notable NOMA scheme is low-density spreading (LDS), which is a code domain multiple access scheme. Low density spreading operates like code division multiple access (CDMA) in the sense that users use a spreading sequence to spread their data, but the spreading sequences have a low number of nonzero chips, hence the term low-density. The message passing algorithm (MPA) is typically used for multi-user detection (MUD) of LDS systems. The MPA detector has complexity that is exponential to the number of users contributing to each chip. LDS systems suffer from two inherent problems: high computational complexity, and vulnerability to multipath channels. In this thesis, these two problems are addressed. A lower complexity MUD technique is presented, which offers complexity that is proportional to the number of users squared. The proposed detector is based on minimum mean square error (MMSE) and parallel interference cancellation (PIC) detectors. Simulation results show the proposed MUD technique achieves reductions in multiplications and additions by 81.84% and 67.87% with a loss of about 0.25 dB with overloading at 150%. In addition, a precoding scheme designed to mitigate the effects of the multipath channel is also presented. This precoding scheme applies an inverse channel response to the input signal before transmission. This allows for the received signal to eliminate the multipath effects that destroy the low-density structure.