| Contention-based access networks, such as the increasingly popular Wireless LAN, pose unique challenges in quality of service (QoS) provisioning, due to the random access nature of the underlying protocols. In this thesis we study a price-based approach towards adaptive rate control in a classic contention-based network—the Slotted ALOHA, where a common price reflects the current congestion state. The effectiveness of this approach in alleviating channel congestion and allocating bandwidth resources is investigated. Specifically, using drift analysis, we prove that the Markov chain associated with this approach is positive recurrent, which substantiates its capability of stabilizing the channel. We also show that service differentiation in terms of throughput can be realized at the stabilized configuration, which is tunable by means of a suitable pricing policy. Furthermore, we demonstrate empirically that our approach can be naturally extended to support multidimensional QoS in terms of both throughput and delay. |
