Admission Control for Wireless Networks

With the population of wireless subscribers increasing at a rapid rate overloaded sit uations are likely to become an increasing problem Admission control can be used to balance the goals of maximizing bandwidth utilization and ensuring su cient resources for high priority events In this paper we formulate the admission control problem as a Markov decision problem While dynamic programming can be used to solve such problems the large size of the state space makes this impractical We propose an ap proximate dynamic programming technique which involves creating an approximation of the original model with a state space su ciently small so that dynamic programming can be applied Our results show that the method improves signi cantly on policies that are generally in use in particular the greedy policy and the reservation policy Much of the computation required for our method can be done o line and the real time computation required is easily distributed between the cells