Performance and fairness in wireless and wireline backhaul networks

This thesis studies end-to-end performance and fairness in two classes of backhaul networks, namely multihop wireless backhaul networks and metropolitan backbone networks. Current media access and transport protocols in both wired and wireless backhaul networks can result in severe inefficiency and poor performance. In particular, flows that are an increasing number of hops away from a wired entry point or Internet backbone experience extreme unfairness and in some cases even starvation. The contributions of this thesis are as follows. First, we develop a formal reference model that characterizes performance objectives of multihop wireless backhaul networks and perform an extensive set of simulation experiments to quantify the impact of the key performance factors towards achieving these goals. For example, we study the roles of the MAC protocol; end-to-end congestion control, antenna technology, and traffic types. Next, we develop and study a distributed layer 2 fairness algorithm which targets to achieve the fairness of the reference model without modification to TCP. We also study the critical relationship between fairness and aggregate throughput. Next, we study the performance of utility maximization congestion control algorithm over multihop CSMA-based networks. We develop a framework to study key issues in such networks that are not incorporated by prior models, yet are critical to the performance of congestion control algorithms. Our study provides a deeper understanding of the impact of CSMA-based MAC on the performance of utility maximization congestion control algorithm. Finally, we introduce a new dynamic bandwidth allocation algorithm for metropolitan backbone networks called Distributed Virtual-time Scheduling in Rings (DVSR). To evaluate DVSR, we develop an idealized fairness reference model. With simulations, we find that compared to current techniques, DVSR's convergence times are an order of magnitude faster (e.g., 2 vs. 50 msec), oscillations are mitigated (e.g., ranges of 0.1% vs. up to 100%), and nearly complete spatial reuse is achieved (e.g., 0.1% throughput loss vs. 33%)