Random access game in ad hoc networks with cooperative and noncooperative users

Motivated by the fact that the exponential back-off algorithm is unfair towards nodes in the middle of networks, we propose two approaches to computing the optimal persistence probabilities in Medium Access Control based on game theory. The first is a cooperative game theoretic MAC with the objective of total payoff maximisation. By decomposing the overall problem into each user's subproblem, we solve the nonconvex random access problem in a distributed manner. The existence, uniqueness and stability of the Nash Equilibrium (NE) for the cooperative game are proved. The second is the noncooperative persistence scheme with an aim to minimise information exchanging among nodes. This formulation leads to a simple iterative scheme by which each user can arrive at a Pareto dominant NE. Simulation results illustrate the performance of these algorithms.