Transmission and Reception Techniques for URA Based on Spreading and Compressed Sensing

Abstract

The Internet of things (IoT) is a component of the massive machine-type communication (mMTC) paradigm, which supports a vast number of autonomous communicating devices. Unsourced random access (URA) is a communication scenario where a multitude of users, potentially IoT devices, attempt to transmit short packets to a common receiver in a grant-free manner, i.e., without any pre-allocated resources such as time, bandwidth and code. At any given time, the number of active users is small relative to the potential users. It is assumed that the receiver needs to acquire the information transmitted by the devices but not their identities (IDs). In this thesis, first we study URA on a Rayleigh fading channel and URA with a multiple-antenna receiver. We employ a preamble-payload packet format for activity detection and the transmission of information for each user. The payload is spread, interleaved, and scrambled using a pattern selected randomly from a large pool of all possible patterns. These patterns are not orthogonal and are exploited to reduce cross-correlation between the users, making it possible for the receiver to detect users. This pattern is conveyed to the receiver by the preamble part. We evaluate the system’s performance analytically, taking into account channel estimation errors. Our numerical and analytic results demonstrate that the presented system outperforms related work in terms of achievable numbers of active users and signal-to-noise ratios (SNRs) supported by the system. Furthermore, a feedback mechanism has been employed in the URA system to dramatically reduce the SNR of transmitted bits. Moreover, a compressed sensing (CS)-based URA is proposed. The packet format in CS-based URA schemes contains multiple slots, and each user sends a sub-message within each slot. At the receiver, a tree decoder is employed to connect sub-messages. The proposed algorithm eliminates the need for a tree code by connecting the sub-messages in a novel method. Simulation results indicate that the proposed scheme outperforms other CS-based algorithms and superior than the other state-of-the-art URA schemes when the number of active users exceeds 200. The presented analysis framework closely predicts the obtained numerical results for the proposed scheme.