Research On Computation Offloading And Resource Allocation In Mobile Edge Computing

With the explosive growth of smart mobile devices such as smartphones,tablet computers,and virtual reality devices,the scale of the Internet of Things has expanded dramatically.Various applications and services carried by user devices have also required higher levels of security and real-time performance.However,smart mobile devices are usually limited by their battery life,computing power,and storage capacity,and it is difficult to meet the requirements for service quality when running complex applications.With respected to the above mentioned,Mobile Edge Computing(MEC)technology is considered to be a key technology to solve these problems.Compared with traditional mobile cloud computing technology,mobile edge computing usually deploys servers near users and spreads computing resources to edge servers,which means mobile edge computing uses a large number of service nodes located the edge of the network to enhance the computing and storage capabilities of mobile devices by reducing Transmission delay and energy consumption of applications running in mobile devices.Mobile edge computing has limited computing resources and storage resources in MEC server,and most research about it focuses on computing offloading and resource allocation for improving the quality of service for users and reducing the latency and low energy consumption of mobile devices in multi-user scenarios.In response to the above problems,the research work of this article includes the following two parts:(1)With respected to the problem of static computing offloading and resource allocation,a model for optimization of joint computing offloading,computing resources and cache resources under heterogeneous networks is proposed.For tasks generated by mobile devices,the total delay of mobile devices is minimized through joint offloading decisions,computing resource allocation decisions,and cache resource allocation decisions.An evolutionary algorithm based on the mayfly algorithm is proposed to solve the static computing offloading and resource allocation problems,and the coding method of this algorithm is adjusted according to this problem model.To enhance the algorithm performance of the mayfly algorithm,a mayfly algorithm based on opposition learning is proposed.Lastly,the feasibility of the proposed scheme and the superiority of the improved algorithm has been proved by simulation experiment analysis.(2)With respected to the problem of dynamic computing offloading and resource allocation,a continuous-time optimization problem model driven by task arrival events is proposed.For tasks dynamically generated by mobile devices,the user's mobility and the time-varying resource state of the network is considered,and the scheme by jointing computing offloading decisions,computing resource allocation decisions,wireless resource allocation decisions,and power allocation decision is adopted to minimal computational consumption of tasks.To solving this problem model,the reinforcement learning DDPG algorithm is proposed,and the three elements,which including state,action and reward,of reinforcement learning are designed according to the problem model.Finally,the feasibility of the proposed scheme and the superiority of the improved algorithm has been proved by simulation experiment analysis.