Adaptive resource allocation for media services based on semi-Markov decision process

In this paper, we study adaptive resource allocation for elastic media services in service-oriented network. We form ulate the resource allocation problem as a semi-Markov decision process (SMDP) to capture the dynamics of user arrivals and departures. Based on the network resources, an optimal decision is made to maximize the overall system rewards by striking the balance between the network utilities and costs of network resources. We further analyze the network performance in terms of the service blocking probability, and the probability of different decisions or actions of resource allocation. Extensive simulations demonstrate that our proposed scheme can achieve much higher network utility and lower service blocking probability compared with a greedy resource allocation.