We study in this paper a noncooperative game with an infinite number of players that are involved in many local interactions, each involving a randomly selected pair of players. Each player has to choose between an aggressive or a nonaggressive action. The expected lifetime of an individual as well as its expected total fitness during its lifetime (given as the total amount of packets it transmits during the lifetime) depend on the level of aggressiveness (power level) of all actions it takes during its life. The instantaneous reward of each player depends on the level of aggressiveness of his action as well as on that of his opponent. We model this as a Markov Decision Evolutionary Game which is an extension of the evolutionary game paradigm introduced in 1972 by Maynard Smith, and study the structure of equilibrium policies.