Two timescale dynamic node-selfishness management in wireless networks

In this paper, a two timescale dynamic node-selfishness management mechanism is proposed with respect to the node´s available resources as well as vicinal nodes´ actions, which have different time-varying characteristics. Based on this dynamic node-selfishness mechanism, we develop a dynamical flow allocation approach. During a long-timescale interval, which consists of several short time intervals, an strategy-learning process is exploited for node to decide whether to forward the received packets in exchange for money based on the evolutionary game theory (EGT). Moreover, the evolutionary steady state is obtained by nodes with same resource level in a node-group. At two adjacent long-timescale intervals, the Markov decision process (MDP) aided optimal flow allocation is determined in harmony with the resource dynamics of intermediate nodes. Our simulation results verify the theoretical analysis.