Skip to Main content Skip to Navigation
New interface

Game Theoretic Approaches to Spectrum Sharing in Decentralized Self-Configuring Networks

Abstract : This thesis deals with the analysis and design of techniques for spectrum sharing in decentralized self-configuring networks (DSCN). For such networks, spectrum sharing can be broadly divided into two consecutive phases that radio devices im- plement autonomously, and often individually. In the first phase, radio devices identify their available spectrum access opportunities (SAO). For instance, unused frequency bands, time slots, or spatial directions over which their transmissions are allowed to take place. In the second phase, radio devices determine the optimal transmit/receive configurations for exploiting the available SAOs while guarantee- ing a reliable communication. Here, such configurations are described in terms of power allocation policies, modulation-coding schemes, scheduling policies, decoding orders, etc. For the first phase, we introduce a novel notion of SAO. The main idea consist in exploiting the unused spatial directions (SD) associated with the singular values of the channel matrix of a given primary link using a water-filling power allocation (PA) policy. The method proposed in this thesis for exploiting such opportunities is called opportunistic interference alignment (OIA) and relies on the existence of multiple antennas at both transmitters and receivers. This novel spectrum sharing technique is particularly useful in highly dense networks where classical SAOs such as unused time slots and/or unused frequency bands are short-lasting rare events. For the second phase, it is well known that the main problem lies in the mutual interference arising from the simultaneous exploitation of the same set of available SAOs. Therefore, to study such a competitive interaction between the radio devices, we make use of tools from game theory. Within this framework, we adopt a particular network topology (parallel multiple access channel) to study the existence and the multiplicity of Nash equilibria (NE). The relevance of NE stems from the fact that it represents a network state where each radio device's configuration is optimal with respect to the configuration of all the other devices. In particular, we show that, paradoxically, an important gain in the global performance at the NE is observed by reducing the number of possible configurations a radio device is allowed use. Later, we introduce a novel technique that allows radio devices to achieve NE in a fully decentralized fashion based only on the periodical observation of their individual performance. This (learning) technique is independent of both the network topology and the performance metric of the radio devices. More importantly, we show that it converges to epsilon-NE in relevant types of games in wireless communications, namely potential games among others. We finally tackle the quality of service provisioning in DCSNs. We thus formalize an alternative notion of equilibrium, namely satisfaction equilibrium (SE). Contrary to the existing equilibrium concepts, we show that the SE properly models the problem of QoS in DCSNs. More importantly, we introduce learning techniques that allow achieving a SE relying only on a periodical 1-bit message from the receivers. In particular, as long as the SE exists, these techniques achieve a SE in finite time and are shown to be computationally simpler than those used to achieve NE.
Complete list of metadata
Contributor : Samir M. Perlaza Connect in order to contact the contributor
Submitted on : Tuesday, February 7, 2012 - 1:24:36 AM
Last modification on : Tuesday, October 18, 2022 - 3:33:41 AM
Long-term archiving on: : Wednesday, December 14, 2016 - 5:26:03 AM


  • HAL Id : tel-00667124, version 1


Samir Perlaza. Game Theoretic Approaches to Spectrum Sharing in Decentralized Self-Configuring Networks. Computer Science and Game Theory [cs.GT]. Ecole nationale supérieure des telecommunications - ENST, 2011. English. ⟨NNT : ⟩. ⟨tel-00667124⟩



Record views


Files downloads