Share factors optimization in the Push-Sum algorithm for sensor networks

The convergence speed of an asynchronous point-to-point version of the Push-Sum algorithm in sensor networks is investigated both through numerical simulations and theoretical arguments. The main contribution of the paper consists in studying the application of such algorithm in realistic scenarios, represented by non fully-meshed networks. Simulations show that, in this case, convergence may be strongly dependent on the adopted share factor, whose value should be optimized as a function of the connectivity level of the network. Optimum shares are derived for some common topologies, like the ring and the random geometric graph.