Dynamic resource allocation for service in mobile cloud computing with Markov modulated arrivals

Abstract

Mobile Cloud Computing (MCC) is a modern architecture that brings together cloud computing, mobile computing and wireless networks to assist mobile devices in storage, computing and communication. A cloud environment is developed to support a wide range of users that request the cloud resources in a dynamic environment with possible constraints. Burstiness in users service requests causes drastic and unpredictable increases in the resource requests that have a crucial impact on policies of resource allocation. How can the cloud system efficiently handle such burstiness when the cloud resources are limited? This problem becomes a hot issue in the MCC research area. In this paper, we develop a system model for the resource allocation based on the Semi-Markovian Decision Process (SMDP), able of dynamically assigning the mobile service requests to a set of cloud resources, to optimize the usage of cloud resources and maximize the total long-term expected system reward when the arrival process is a finite-state Markov-Modulated Poisson Process (MMPP). Numerical results show that our proposed model performs much better than the Greedy algorithm in terms of achieving higher system rewards and lower service requests blocking probabilities, especially when the burstiness degree is high, and the cloud resources are limited.