With the increasing development of cloud computing and wireless technology, mobile cloud computing has been developed to alleviate the limitation of battery capacity and computing capability of the mobile device by offloading some computation-intensive tasks onto the cloud. However, the extra consumption for transmission from the mobile device to the remote cloud may lead to degradation of performance. To this end, the authors develop a Markov decision process-based computation offloading (MDPCO) algorithm to minimise the energy efficiency cost (EEC) from a global perspective by jointly optimising the resource allocation and offloading decisions. Firstly, they formulate an EEC minimisation problem for a single-chain application with M tasks. Due to the difficulty to directly solve the formulated problem, they decompose it into multiple subproblems and preferentially optimise the local computing frequency and transmission power by distributed algorithm under hard time constraints. Based on this, they proposed the Markov decision process-based offloading algorithm to preschedule the computing side for each task from a global perspective to minimise the EEC further. The simulation results show that the performance of the MDPCO algorithm is significantly superior to that of the other algorithms under different parameters.

中文翻译：

---

基于马尔可夫决策过程的计算分流算法和时间约束下的资源分配

随着云计算和无线技术的不断发展，已经开发了移动云计算以通过将一些计算密集型任务卸载到云上来减轻移动设备的电池容量和计算能力的限制。但是，从移动设备传输到远程云的额外消耗可能导致性能下降。为此，作者开发了一种基于马尔可夫决策过程的计算分流（MDPCO）算法，通过联合优化资源分配和分流决策，从全局的角度将能效成本（EEC）降至最低。首先，他们为单链应用制定了EEC最小化问题，中号任务。由于难以直接解决所提出的问题，他们将其分解为多个子问题，并在困难的时间约束下通过分布式算法优先优化本地计算频率和传输功率。基于此，他们提出了一种基于马尔可夫决策过程的卸载算法，以便从全局角度为每个任务预先安排计算端，以进一步最小化EEC。仿真结果表明，在不同参数下，MDPCO算法的性能明显优于其他算法。