2025 Theses Doctoral

Sparsity-exploiting Transmitter and Receiver Algorithms for Uplink Massive Machine-type Communications

Liu, Jiaai

This paper introduces novel transmission schemes and receiver algorithms aimed at improving the current multiple access protocols in MIMO and massive MIMO channels, with applications to emerging wireless systems such as massive machine-type communications (mMTC) and IRS-assisted networks. For UMA schemes, we first propose hybrid modulation schemes for both MIMO and massive MIMO to reduce receiver complexity, where sub-blocks of coded bits are modulated using a mix of nonlinear and linear approaches.

We also develop sparsity-exploiting blind receiver algorithms that leverage codeword and channel sparsity for efficient channel estimation and signal decoding. In addition, we introduce a sparse-graph-based transmission scheme for UMA, where transmitters use a Tanner graph to select sub-slots for repeated transmission. Iterative decoders are employed to decode one or more codewords per iteration, with novel blind channel estimation techniques used for decoding. Density evolution analysis is performed to assess the asymptotic performance of these schemes, showing their advantage over traditional compressed-sensing (CS)-based UMA methods.

On the other hand, we fix the Tanner graph for transmission and propose novel grant-based schemes based on known Tanner graphs combining non-orthogonal multiple access (NOMA) and random access. These schemes employ message-passing decoders to fully exploit the diversity across multiple resource blocks, with a neural network-based decoder further improving performance by learning from training epochs. In mMTC, we address the challenge of low-latency and high-reliability communication by designing a physical-layer transceiver that utilizes sparse Tanner graphs and hybrid modulation, ensuring efficient blind channel estimation and robust decoding.

Moreover, we consider an IRS-assisted system where each IRS element is equipped with two 1-bit analog-to-digital converters (ADCs) and propose a novel CS-based channel estimator with the 1-bit received signals at the IRS. This method can also be extended to the task of joint activity detection and channel estimation (JADCE), reducing pilot overhead and allowing for scalable system operation. Simulation results demonstrate the superior performance of our proposed solutions in various network scenarios.