Wireless Resource Allocation among Cooperative Relays and Non-cooperative Agents

Public Deposited

This thesis focuses on resource allocation in wireless communication and networking. Resource allocation has been studied widely, for example, to maximize the system-wide throughput or to minimize the average delay per user. Moreover, the utility-based framework is becoming an important tool for addressing fairness and Quality-of-Service (QoS) for individual users. With this framework, therefore, we study wireless resource allocation for relay extensions in a cellular network and dynamic spectrum sharing among non-cooperative agents. In the first part of the thesis, we study the centralized resource allocation problems among cooperative agents. Especially, the relay extension in a cellular network is focused on, including in-band relays in <em>IEEE</em> 802.16j and out-of-band relays with Wi-Fi access points. Here, the base station optimizes the time or power allocation such that total throughput for data traffic or total number of active users for voice traffic is Maximized. For the cases considered, a significant gain for both data traffic and voice traffic is obtained. In the second part of the thesis, dynamic spectrum sharing with distributed resource allocation is considered with an emphasis on the development of mechanisms and their performance. First, we consider an ``AP deployment'' game in the commons model and show that there exists a Nash equilibrium which is efficient. In addition, we address the limitations of the commons model when interference is severe. Second, we study two auction mechanisms for resource allocation in a peer-to-peer network: The sequential second-price auction and the ``Fallback'' auction. For the sequential auction, the resource is divided into $n$ units and each unit is auctioned off sequentially according to a second-price auction. The worst-case efficiency of the sequential auction is shown to be lower bounded by $1-e^{-1}$ for a bandwidth allocation and upper bounded by $1/n$ for a power allocation. Because of the low worst-case efficiency for power allocation and the impracticality of the complete information assumption in the sequential auction, we discuss a Fallback auction, which modifies Ausubel's ascending auction. With an increasing convex utility of one agent due to interference, we show that the Fallback auction for power allocation achieves a stable outcome with minimum revenue to the seller in the core.

Last modified
  • 09/20/2018
Date created
Resource type
Rights statement