Resoure Scheduling And Energy Saving Problem Of Mobile Cloud Computing

Mobile Cloud Computing (MCC) is a novel technology which combines mobile internet and cloud computing. Users can overcome many shortages of the Mobile Terminals (MTs), which includes poor computing power, small storage capacity and short battery lifetime, by offloading the computing and storage service to the cloud. With the rapid development of MCC, users’requirement for the cloud service is growing greatly. A higher quality of service and a lower energy consumption are expected. However, mobile internet and cloud computing are two different techniques. MCC system is faced enormous challenges, because mobile internet and cloud computing usually belong to different providers, which consequently leads to a lack of unified deployment and management. In order to improve both the benefits of providers and the quality of mobile cloud service, as well as save the energy of MT, we investigate the problems of resource scheduling and energy saving in MCC, which includes scheduling of Virtual Machine (VM), collaborative task execution, access control and transmission scheduling. The main work and innovations in this paper are listed as following:(1) The computing resources scheduling methods have been studied in the MCC environment. We first consider the limited bandwidth scenario with cloud data center offering services to users in current MCC environment. Using the auction theory, a Bandwidth-constrainted VM Dynamic Scheduling (BVMDS) algorithm has been proposed, which can dynamic allocate VM to users quickly. The theory analysis and simulation results show that the proposed BVMDS algorithm can effectively improve the profit and the resource utilization of providers. Then, we focus on the future of the MCC system that can not only control bandwidth of the cell, but also analyze the trade process between provider and users. Based on the Stackelberg theory, a VM Pricing and Allocation (VMPA) algorithm has been proposed, which can control network congestion according to the VM price. The theory analysis and simulation results show that the proposed VMPA algorithm can optimize the profit of provider and users simultaneously.(2) The collaborative execution methods between mobile device and cloud side for mobile applications have been studied. The mobile application is composed of a series of tasks, which are different in computing workload, input and output data size. Moreover, offloading the task with a low computing workload and high throughput data to the cloud will cause a high energy consumption on MTs. To this end, we first study collaborative execution scheme for the mobile application with serial tasks. Using once migration characteristics, a Once Migration Genetic Algorithm (OM-GA) has been proposed, which can effectively reduce iterations and operation time. The theory analysis and simulation results show that the proposed OM-GA algorithm can effectively reduce the energy consumption while meeting the deadline. Then, we study collaborative execution scheme for the mobile application with parallel tasks. Based on the Lagrangian optimization method, a Collaborative Execution for Parallel Tasks (CEPT) algorithm has been proposed, which can rapidly obtain the optimal solution. The theory analysis and simulation results show that the proposed CEPT algorithm can make a tradeoff between energy consumption of MD and the deadline of mobile appplication, and reduce the energy consumption while meeting the deadline.(3) The access control methods have been studied in the MCC environment. According to the characteristics of service in the MCC system, we divided all services into two types:realtime and non-realtime services. First, we first discuss the access control in the user handover scenario. It’s considered that the blocking of handover service has more influence on the profit of the system than the blocking of new service does, a Dynamic Threshold Access Control (DTAC) algorithm has been proposed, which can dynamic allocate channel to handover user. The theory analysis and simulation results show that the proposed DTAC algorithm can guarantee the quality of handover service while maximizing the profit of the mobile cloud computing system. Then, we discuss the access control in multiple services arrived scenario. In order to reduce the blocking probability of realtime services, a Service Adaptive Access Control (SAAC) algorithm has been proposed. This SAAC algorithm can transmit the realtime services with a higher priority and allocate suitable channel for the non-realtime services. The theory analysis and simulation results show that the proposed SAAC algorithm not only enhances channel utilization, but also efficiently reduces the blocking probability of realtime services and the delay of non-realtime services.(4) The transmission scheduling methods have been studied in the MCC environment. We first investigate the optimal transmission energy for the multiple channels and multiple users scenario. Based on Lyapunov function, we have proposed a Two Time Scale Scheduling (T2S2) algorithm that can dynamically choose user to transmit data according to queue backlog and channel statistics. The theory analysis and simulation results show that the proposed T2S2 algorithm can make a tradeoff between queue backlog and energy consumption in MCC system, Meanwhile, the proposed algorithm significantly outperforms existing scheduling algorithms in energy consumption. Then, we concentrate on the problem of minimizing the transmission energy in the MCC networks where time-varying channel and the number of service subscription are considered. Using the Lyapunov optimization method, we propose a Combined Service Subscription and Delivery (CSSD) algorithm which not only can guide users to subscribe to services reasonably, bur also can determine whether to deliver the data as well as to whom this data is sent in the current time unit based on the queue backlog and channel state. The theory analysis and simulation results show that the proposed CSSD algorithm can effectively decrease both the transmission energy and queue backlog while decreasing the unnecessary overhead caused by unreasonable service subscribing.