Daniel McClary

Computer networks often display nonlinear behaviour when examined over a wide range of oper- ating conditions. There are few strategies available for modelling such behaviour and optimizing such systems as they run. Prole-driven regression is developed and applied to modelling and run-time optimization of throughput in a mobile ad hoc network, a self-organizing collection of mobile wireless nodes without any xed infrastructure. The intermediate models generated in prole-driven regression are used to t an overall model of throughput, and are also used to op- timize controllable factors at run-time. Unlike others, the throughput model accounts for node speed. The resulting optimization is very ective; locally optimizing the network factors at run- time results in throughput as much as six times higher than that achieved with the factors at their default levels.