Network Coding and Evolutionary Theory for Performance Enhancement in Wireless Cooperative Clusters

Cooperation over short-range wireless links among user devices, which download remote contents through cellular links, is a paradigm quickly gaining ground, given that it can answer several technological and design issues that next-generation wireless applications will raise. Among others, energy consumption, data rate and data transfer time are parameters that will likely benefit from this novel communication concept. This paper presents a WLAN-based cooperative scenario, conceived to support file-sharing services, in which a sub-set of cluster nodes access, through their cellular radio interface, portions of a file to be successively exchanged among all cluster members over wireless local area network (WLAN) links. Besides showing the beneficial effects of cooperation, this paper also focuses on the performance enhancement that can be achieved when using the network coding paradigm, whose deployment revealed to be very effective in the depicted scenario. The selection of the nodes acting as sources for file injection into the cluster is a critical aspect for the effectiveness of the whole system. Therefore, the potentialities offered by the evolutionary theory applied to the specific problem are investigated and an ad hoc conceived Genetic Algorithm (GA) designed. Either the service time (the time needed for all nodes to receive the complete file) or the energy consumption for the nodes is used as objective function, showing in both cases the fast convergence for the algorithm that makes it preferable to either exhaustive searches or random choices.