This final report summarizes the SmartNets research group’s work on QoS in ad-hoc networks. It describes the major lessons learned, the remaining open questions, and further research directions. Our work focuses on four major areas: Media Access Control (MAC) protocols, models for capacity estimation, routing protocols, and clustering. Our work in MAC protocols includes work on a novel “impatient” backoff algorithm that enhances fairness between contending nodes, as well as work on a multi channel MAC protocol that improves throughput. We are motivated to study capacity estimation models because many QoS routing algorithms require some methodology for determining how much capacity is available on the different paths of an adhoc network. This work has led to a clique based algorithm for residual capacity estimation that can be implemented in a distributed fashion, and that requires a manageable amount of computation. Our work in QoS routing includes routing schemes that maintain network state information, use that state as an input to our clique capacity model, and then use the capacity estimates to guide admission and routing decisions. We also look at more practical QoS schemes that make use of information from both real-time measurements, as well as from maintained state information. The scalability of QoS routing schemes can be improved by dynamically clustering the ad-hoc network into clusters, and having each node maintain detailed state information about its local cluster only, while perhaps keeping some limited global state information. Our work in clustering includes the development of a novel clustering algorithm that adapts to node mobility.