Συντάχθηκε 13-06-2012 10:21
από Galateia Malandraki
Email συντάκτη:
gmalandraki<στο>tuc.gr
Ενημερώθηκε:
-
Ιδιότητα: υπάλληλος ΑΡΜΗΧ.
ΠΟΛΥΤΕΧΝΕΙΟ ΚΡΗΤΗΣ
Τμήμα Ηλεκτρονικών Μηχανικών & Μηχανικών Υπολογιστών
ΠΑΡΟΥΣΙΑΣΗ ΔΙΔΑΚΤΟΡΙΚΗΣ ΔΙΑΤΡΙΒΗΣ
ΜΑΤΣΚΑΝΗ ΕΥΑΓΓΕΛΙΑΣ
με θέμα
“Από κοινού δρομολόγηση και κατανομή πόρων σε ασύρματα δίκτυα με τεχνικές κυρτής προσέγγισης”
Παρασκευή 15 Ιουνίου 2012, 10:30 π.μ.
Αμφιθέατρο, Κτίριο Επιστημών, Πολυτεχνειούπολη
Εξεταστική Επιτροπή
Καθ. Σιδηρόπουλος Νικόλαος, Πολυτεχνείο Κρήτης
Καθ. Λιάβας Αθανάσιος, Πολυτεχνείο Κρήτης
Καθ. Zhi-Quan Luo, Πανεπιστήμιο της Μινεσότα, Η.Π.Α.
Καθ. Τασσιούλας Λέανδρος, Πανεπιστήμιο Θεσσαλίας
Καθ. Πατεράκης Μιχαήλ, Πολυτεχνείο Κρήτης
Επ. Καθ. Μπλέτσας Άγγελος, Πολυτεχνείο Κρήτης
Επ. Καθ. Κουτσάκης Πολυχρόνης, Πολυτεχνείο Κρήτης
ΠΕΡΙΛΗΨΗ
The part of the thesis presented concerns the joint back-pressure routing and power control in the context of cross-layer wireless networking. Throughput-optimal multi-hop wireless network operation entails a key physical-layer optimization problem: maximizing a weighted sum of link rates, with weights given by the differential queue backlogs. This emerges in joint back-pressure routing and power control, which is central in cross-layer wireless networking. We begin by showing that the core problem is not only non-convex, but also NP-hard. This is a negative result, which however comes with a positive flip side: drawing from related developments in the digital subscriber line (DSL) literature, we propose effective ways to approximate it. Exploiting quasi-periodicity of the power allocation in stable setups due to the push-pull nature of the solution, we derive two custom algorithms that offer excellent throughput performance at reasonable, worst-case polynomial complexity. Judicious simulations illustrate the merits of the proposed algorithms.
Back-pressure power control (BPPC) is amenable to successive convex approximation strategies that deliver manifold improvements in end-to-end throughput relative to the prior art in wireless networking, as our findings suggest. A drawback is that existing implementations are centralized, whereas practical power control has to be distributed across the network. Our work fills this gap by developing distributed implementations of approximations of the BPPC problem. Two approximation approaches to the NP-hard underlying problem are utilized, while feedback requirements and consensus-on-termination issues are also addressed, in order to come up with fully decentralized protocols that allow tight approximation of the BPPC objective in all interference regimes. The first distributed algorithm proposed is based on the successive convex approximation approach, and the alternating direction method of multipliers (ADMoM) is utilized towards distributed implementation of the core step of the approach, which is the convex lower-bounding approximation of BPPC, at any operating point. Judicious simulations verify that the distributed algorithm derived matches the performance of its centralized counterpart. The iteratively weighted Minimum Mean Square Error approach to weighted sum-rate maximization for the MIMO interference channel is also exploited in our context. A second distributed implementation is then possible, providing a one-shot approximate solution to the BPPC problem. Further exploiting quasi-periodicity of the solution arising in stable setups, due to the push-pull evolution of the network in our context, and upon elaboration on distributed warm re-start, we derive custom adaptive versions of the distributed algorithms, that enjoy low average complexity with no performance loss. Extensive simulation experiments reveal the strengths of the proposed algorithms and allow for comparison of the two approximation approaches to the BPPC problem.
Συνημμένα:
-
Matskani-Anakoinwsi.doc
Μεταφορτώσεις: 215,
Μέγεθος: 45 KB application/msword
