Iterative Interference Cancellation, and Channel Estimation for Underwater Acoustic and Unsourced Random Access Communications
The work in this thesis is dedicated to design and analysis of iterative interference cancellation systems with applications in underwater acoustic communication and random access communications over terrestrial wireless channels. In the first part of the work a technique is proposed for signal transmission and reception over underwater acoustic communications channels targeting high spectral efficiency. The transmit data is split into multiple superimposed streams, where each stream is encoded via an error-correction code, interleaved, permuted, and modulated via Orthogonal Frequency Division Multiplexing (OFDM). Since the channel exhibits a doubly selective nature due to multi-path and fading, a significant channel variation occurs within each OFDM symbol. The task of the receiver is to iteratively refine the channel estimate, cancel the inter-carrier interference (ICI) introduced by the Doppler spread, and inter-data stream interference, and perform error-correction decoding. The receiver is shown that it can operate successfully over a range of significant Doppler spreads. Because of lack of standard underwater channel models, I propose an underwater channel model which is suited for my proposed Kalman Forward-Backward channel estimation algorithm specifically to support the iterative receiver with channel knowledge. Simulation results of the integrated iterative cancellation process with channel estimation are presented. The second part of the thesis deals with unsourced random access (URA) over Gaussian and fading channels. In a URA setting, a very large number of potential users is considered, while only a smaller subset of the users is active at any given time. The users transmit messages in a grant-free fashion, utilizing a common codebook. A new URA iterative cancellation receiver is proposed and studied, that operates in both Gaussian and fading channels. The transmitted message format is concatenation of preamble and data payload packets respectively. Payload of each user incorporates repetition and permutation, and the task of the receiver is to cancel users' interference when retrieving a user signal. The preamble of each packet carries information about the users signature and permutation sequences. The main focus is on the approximate message passing (AMP)-based preamble detection algorithms, where user’s preambles is employed for channel estimation. As in the underwater case, the URA interference cancellation multi-user detector (MUD) algorithm depends on channel estimate obtained from the preamble part. A set of new preamble detection algorithms which can perform user activity and collision detection, as well as channel estimation for large numbers of active users is presented. The results are compared with the sate-of-the art methods and demonstrate advantages over the existing algorithms for both stand-alone preamble detection and the entire URA system performance.