Asynchronous multi-user detection using inference algorithmsAsynchronous multi-user detection using inference algorithmsΑσύγχρονη εκτίμηση πολλαπλών χρηστών με αλγόριθμους συμπερασμού Διπλωματική Εργασία Diploma Work 2017-11-132017enThis work highlights state-of-the-art probabilistic graphical models and infer- ence algorithms that exploit rather than avoid the symbol-level asynchrony in multi-user wireless communications. More specically, the recently pro- posed inference algorithm SigSag is studied, where multiple users transmit their data packets at the same time and frequency channels, with however random delays; assuming channel state information (CSI) at the receiver, linear equations are formed, which produce a probabilistic graphical model (PGM), amenable to inference algorithms. This work implements the sum- product belief propagation algorithm on the crafted PGM and a) derives the message passing equations, b) studies initialization and c) complements with CSI estimation, using linear minimum mean squared error (LMMSE) estimator. Performance was tested for 2 or 3 users. It was found that per- formance was sensitive to initialization, as expected, due to the inherently loopy nature of the crafted PGM for small packet lengths. On the contrary, bit error rate (BER) decreases with increasing packet length, at the expense of convergence time. Moreover, reduced convergence time results to higher BER.http://creativecommons.org/licenses/by/4.0/Πολυτεχνείο Κρήτης::Σχολή Ηλεκτρολόγων Μηχανικών και Μηχανικών ΥπολογιστώνTsetis_Ioannis_Dip_2017.pdfChania [Greece]Library of TUC2017-11-13application/pdf609.7 kBfree Tsetis Ioannis Τσετης Ιωαννης Bletsas Aggelos Μπλετσας Αγγελος Paterakis Michalis Πατερακης Μιχαλης Karystinos Georgios Καρυστινος Γεωργιος Πολυτεχνείο Κρήτης Technical University of Crete Networking Inference theory Probabilistic graphical models Detection & estimation theory