Dynamic Selection of Persistence and Transport Layer Protocols in Challenged Networks

This work applies a distributed algorithm utilizing Markov Random Fields (MRFs) to the problem of dynamically selecting Session and Transport layer protocols in challenged networks such as mobile ad-hoc networks. It motivates the problem by identifying the primary network properties which affect Message Delivery Ratio (MDR) in networks with varying degrees of connectivity and traffic load. Using this information, local and remote observations are used to select a set of protocols which should perform the best. Analysis shows that dynamically selecting a set of protocols can deliver up to 50% more messages in challenged environments, and never under-performs statically choosing protocols.