Institutional Repository
Technical University of Crete
Technical University of Crete
EN
|
EL
| URI | http://purl.tuc.gr/dl/dias/D7448416-5EAC-4923-9954-BCD3AF5F6F30 | - |
| Identifier | https://doi.org/10.26233/heallink.tuc.95151 | - |
| Language | en | - |
| Extent | 1.6 megabytes | en |
| Extent | 59 pages | en |
| Title | Delay-constrained distributed inference in wireless networks | en |
| Title | Κατανεμημένος συμπερασμός σε ασύρματα δίκτυα με περιορισμούς καθυστέρησης | el |
| Creator | Mitrolaris Stavros | en |
| Creator | Μητρολαρης Σταυρος | el |
| Contributor [Thesis Supervisor] | Bletsas Aggelos | en |
| Contributor [Thesis Supervisor] | Μπλετσας Αγγελος | el |
| Contributor [Committee Member] | Christopoulos Dionysios | en |
| Contributor [Committee Member] | Χριστοπουλος Διονυσιος | el |
| Contributor [Committee Member] | Deligiannakis Antonios | en |
| Contributor [Committee Member] | Δεληγιαννακης Αντωνιος | el |
| Publisher | Πολυτεχνείο Κρήτης | el |
| Publisher | Technical University of Crete | en |
| Academic Unit | Technical University of Crete::School of Electrical and Computer Engineering | en |
| Academic Unit | Πολυτεχνείο Κρήτης::Σχολή Ηλεκτρολόγων Μηχανικών και Μηχανικών Υπολογιστών | el |
| Description | Diploma Thesis | en |
| Content Summary | Belief propagation (BP)-based algorithms are powerful message-passing algorithms that exploit the conditional independences of specific variables on a carefully crafted graph and efficiently compute marginal distributions or maximum a posteriori (MAP) estimates (of the involved variables). Motivated by the convergence guarantees of Gaussian belief propagation (GBP) under high-order factorization and asynchronous scheduling, we study how this algorithm can be utilized by resource-limited wireless networks for in-network processing. We show that there are cases where a particular asynchronous variant of GBP converges in theory but diverges when transmission delays are included. A simple solution is proposed, based on a coordinator, and its performance in terms of convergence time is examined, through simulations; optimized resource allocation is performed, including the bottleneck assignment problem (BAP) formulation, taking into account transmission delays, as well as bandwidth-limited wireless channels and external interference. Next, we study the max-product BP algorithm, as a distributed solver of BAP, i.e., a matching problem between tasks and agents, with the objective of minimizing the costliest pairing, for which only a couple of distributed algorithms currently exist. We examine a line of work which addresses this problem, provided that a unique solution exists and simplify the message calculations; specifically, we propose new BP message expressions, with linear complexity (as opposed to quadratic of prior art). Furthermore, we provide an asynchronous variant and study its convergence and correctness guarantees, using the notion of generalized computation trees. Simulations results show convergence of both the synchronous and asynchronous variant. | en |
| Type of Item | Διπλωματική Εργασία | el |
| Type of Item | Diploma Work | en |
| License | http://creativecommons.org/licenses/by/4.0/ | en |
| Date of Item | 2023-03-02 | - |
| Date of Publication | 2023 | - |
| Subject | Wireless sensor networks | en |
| Subject | Belief propagation | en |
| Bibliographic Citation | Stavros Mitrolaris, "Delay-constrained distributed inference in wireless networks", Diploma Work, School of Electrical and Computer Engineering, Technical University of Crete, Chania, Greece, 2023 | en |
| Bibliographic Citation | Σταύρος Μητρολάρης, "Κατανεμημένος συμπερασμός σε ασύρματα δίκτυα με περιορισμούς καθυστέρησης", Διπλωματική Εργασία, Σχολή Ηλεκτρολόγων Μηχανικών και Μηχανικών Υπολογιστών, Πολυτεχνείο Κρήτης, Χανιά, Ελλάς, 2023 | el |
Copyright © DIAS 2013
