An Adaptive Resource Allocation Algorithm for Partitioned Services in Mobile Cloud Computing

Nowadays, much richer functionality of mobile applications encourages mobile devices to leverage the powerful cloud service for fast application execution by using the technology of Mobile Cloud Computing (MCC). To better utilize the computing resource of the cloud server, a novel resource allocation algorithm is proposed in this paper with the consideration of application partition offloading sequence while maintaining the high quality of service (QoS) of mobile users. The resource allocation problem is modeled as a semi-Markov decision process. Through maximizing the long-term discounted system reward, an optimal resource allocation policy is calculated for partitioned mobile applications using policy iteration approach, which makes a latter partition of the application more easily to obtain resource to speed up the application execution. Both theoretical and simulation results show that the system can adaptively adjust the allocation policy of whether to utilize the cloud and how much computing resource to allocate, according to the request traffic of mobile applications, the partition's position in the application, and the availability of system resources. The proposed algorithm outperforms Greedy admission control in terms of system throughput and QoS over a broad range of environments.