Abstract: Communications networks are ubiquitous. Networks now carry more complex traffic and more complex traffic mixes than ever before. Since the 1970s network architectures have been organized as a series of layers, each with well-defined responsibilities, in order to reduce implementation complexity. Significant effort has been invested to optimize individual protocols at each layer. Recent research has focused on identifying interactions among protocols. Indeed further gains in performance are being realized by cross-layer optimization. This dissertation presents a methodology to identify and quantify interactions among protocols, and to perform cross-layer optimization. Design of experiments provides statistical techniques to study interactions, identifying factors that contribute most to the performance metrics of interest. Cross-layer optimization is carried out using response surface methodology. One strength of the methodology is that it can be applied to data gathered from simulation and also from a physical system such as a test-bed. The methodology is applied to real-time service delivery in mobile ad hoc networks. The characteristics of the wireless channel and rapidly changing network topology make support of real-time traffic in decentralized mobile wireless networks an especially challenging problem. Strong interactions between the access control mechanism and the routing protocol are investigated in detail. Consequently, timer values and link failure reporting are optimized to yield dramatic improvements in performance. As the network conditions change optimal levels for factors also change. Directions for optimization in this dynamic environment are outlined. We have shown a practical methodology to identify and quantify interactions among network protocols and to optimize their performance.