Distributed Access in Wireless Networks: Game Theory and Optimization
Date Issued
2012
Date
2012
Author(s)
Hsu, Fu-Te
Abstract
Medium access control (MAC) plays an important role in wireless networks. It is discovered recently that most MAC protocols for distributed access can be interpreted as a game with appropriate utility functions. In this dissertation, we study some distributed access problems using the tools of game theory and optimization. Our network models taking into account the capture effect and heterogeneity are more general than existing works in the literature, and some new interesting results and properties are revealed. We also make some performance comparisons between the game-theoretic solution and the optimal solution. These will help us understand more about the essence of distributed access in wireless networks and give us some implications on the design principles.
Same as the distributed access schemes in wireless network developed in past years, we start by analyzing ALOHA-like random access in which users independently decide when to transmit and which transmission power levels to use. The game-theoretic solution (for noncooperative users) and optimal solution (for cooperative users) are derived. We also show that under certain condition the game-theoretic solution is the conventional ALOHA (using only one power level), verifying the fact that ALOHA-like random access can be interpreted as a game.
Then, a simple CSMA-based network with the request-to-send (RTS)/clear-to-send (CTS) handshake mechanism is considered. A tight upper bound of the total power consumption in the network is provided. To be specific, the upper bound can be satisfied by one of three points in the feasible throughput region depending on the RTS fraction when all users use the same length of data transmission period.
Finally, we consider a variant of conventional ALOHA called channel-aware ALOHA, which is a joint medium access and physical layer design. In the network, users have the knowledge of their own channel state information (CSI) and utilize a CSI-dependent access scheme to meet their own throughput demands. A game-theoretic approach is employed to analyze this network model under two reception models. One is called the signal to interference plus noise ratio (SINR) capture model; and the other is the power capture model. We show that there are at most two Nash equilibrium points in the feasible region of throughput demands, and point out that under some situations Braess-like paradoxes may occur that the availability of CSI may degrade rather than improve the performance. To the best of our knowledge, our work is the first to show a Braess-like paradox analytically in a random access network. This discovery is important since it was generally believed that the additional availability of CSI should improve the network performance.
Subjects
Game theory
Nash equilibrium
performance analysis
throughput region
power consumption
medium access control (MAC)
ALOHA
CSMA
optimization
Type
thesis
File(s)![Thumbnail Image]()
Loading...
Name
ntu-101-F96942059-1.pdf
Size
23.32 KB
Format
Adobe PDF
Checksum
(MD5):e933e453aa234a1e323c5631499f37d4