Abstract

Ad hoc wireless networks are increasingly being used to provide connectivity where a wired networking infrastructure is either unavailable or inaccessible. Many deployments utilize group communication, where several senders communicate with several receivers; multicasting has long been seen as an efficient way to provide this service. While there has been a great deal of research on multicast routing in ad hoc networks, relatively little attention has been paid to the design of multicast transport protocols, which provide reliability and congestion control. In this dissertation we design and implement a complete multicast transport architecture that includes both routing and transport protocols. Our multicast transport architecture has three modules: (a) a multicast routing and state setup protocol, (b) a mobility detection algorithm, and (c) a hop-by-hop transport protocol. The multicast routing and state setup protocol, called ASSM, is lightweight and receiver-oriented, making it both efficient and scalable. A key part of ASSM is its use of Source Specific Multicast semantics to avoid broadcasting when searching for sources. ASSM also uses routes provided by the unicast protocol to greatly reduce routing overhead. The second module, MDA, solves the problem of determining the cause of frame loss and reacting properly. Frame loss can occur due to contention, a collision, or mobility. Many routing protocols make the mistake of interpreting all loss as due to mobility, resulting in significant overhead when they initiate a repair that is not required. MDA enables routing protocols to react to frame loss only when necessary. The third module is a hop-by-hop multicast transport protocol, HCP. A hop-by-hop algorithm has a faster response time than that of an end-to-end algorithm, because it invokes congestion control at each hop instead of waiting for an end-to-end response. An important feature of HCP is that it can send data at different rates to receivers with different available bandwidth. We evaluate all three components of this architecture using simulations, demonstrating the improved performance, efficiency and scalability of our architecture as compared to other solutions.

Degree

PhD

College and Department

Physical and Mathematical Sciences; Computer Science

Rights

http://lib.byu.edu/about/copyright/