Non-cooperative behavior in wireless networks

Existing cellular networks are centrally managed and require a tremendous initial investment. With the advancement of new wireless technologies, the operators of traditional networks have to face new competition. New technologies make it possible to provide wireless services with substantially less investment. There are two interdependent trends that enable this progress: the evolution of technology and the evolution of spectrum policy. These two trends point towards a more colorful landscape of wireless communication technologies. In particular, new wireless technologies enable users and small operators to deploy their own networks and to compete with the large network operators that run traditional wireless networks. Because the participants have an increased control over their devices, they might be tempted to adjust their devices in order to benefit more from the network. This selfish (i.e., non-cooperative) behavior can dramatically decrease the efficiency of the operation of the network or even paralyze it completely. In this thesis, we consider various aspects of non-cooperative resource management in wireless networks using the methods provided by game theory. In the first part of the thesis, we present a comprehensive tutorial on game theory to facilitate the understanding of this theory as a tool. To emphasize the appropriateness of game theory in wireless networking, we present a set of selected examples along with their game-theoretic formalization. In the second part of the thesis, we are concerned with the non-cooperative behavior of users. More precisely, we focus on ad hoc wireless networks and assume that users can alter the default programming of their communication devices to improve performance. First, we consider the problem of multi-radio channel allocation and show that a Nash equilibrium driven by the selfish behavior of users achieves load-balancing. We propose two algorithms, each based on a different set of available information, to achieve the characterized Nash equilibria. Furthermore, we discuss other properties such as efficiency, fairness and coalition-proofness. In the remainder of this second part, we focus on the fundamental problem of packet forwarding in ad hoc networks. In static networks, we prove that cooperative Nash equilibria exist, but the set of conditions that enable selfish participants to mutually forward each others' packets are very restrictive. We also show that in dynamic networks, mobility promotes cooperation. In the third part of the thesis, we model the non-cooperative behavior of wireless network operators. We assume that they make strategic decisions to maximize the performance of their network. We also assume that the networks of different operators mutually affect each other, hence we model the decisions as network operator games. First, we study the scenario of co-located sensor networks and show that the network operators can mutually increase the lifetime of their network by cooperating. We also show that cooperation in terms of sharing sinks is more beneficial than providing packet forwarding of their sensors only. Second, we focus on the pilot power control of cellular networks assumed to operate in a shared spectrum. We identify Nash equilibria for different parameter values in a single-stage game and show that the cooperative solution can be enforced in a repeated game. Third, we consider the coexistence of cellular networks along national borders. We show that the operators have an incentive to be strategic at their borders, or, in other words, to adjust the transmit power of their pilot signals. We show the efficiency of the Nash equilibria for different user densities. Finally, we extend the payoff function to include the cost of using high pilot powers and relate this extended game to the well-known Prisoner's Dilemma.

Hubaux, Jean-Pierre
Lausanne, EPFL
Other identifiers:
urn: urn:nbn:ch:bel-epfl-thesis3791-3

 Record created 2007-04-25, last modified 2018-03-17

Texte intégral / Full text:
Download fulltext

Rate this document:

Rate this document:
(Not yet reviewed)