| With the explosive growth of user numbers,the generated data volume is also increasing exponentially.To alleviate the communication burden of terminal devices and provide efficient lowdelay services,computation offloading has become an effective solution that attracts attention.As an important extension of cloud computing,edge computing can further improve data transmission efficiency and service quality.Therefore,the computation offloading mechanism based on edge computing has been widely studied.However,there is still a lack of consideration for energy fairness and challenges such as high delay and energy consumption.Based on the above issues,this thesis studies an intelligent computation offloading method for edge computing.The main contributions include three aspects as follows:1)Energy fairness-based hybrid heuristic computation offloading: Regarding the issues of uneven and inefficient resource allocation in computation offloading mechanisms for edge computing scenarios,a hybrid heuristic computation offloading based on energy fairness mechanism is proposed.Specifically,based on the energy fairness of edge server,communication resource and computing resource,an optimization problem is constructed to minimize the total energy consumption of all tasks.In order to solve the mixed integer nonlinear programming problem,the optimal target server is obtained by incorporating the energy fairness into the selection criteria of the target edge server,a hybrid heuristic computation offloading decision algorithm is designed,which combines genetic algorithm and simulated annealing algorithm to improve the convergence speed and search quality and avoid the algorithm falling into local optimal solution.At the same time,the dependence of the algorithm on the initial parameters is reduced and the stability of the algorithm is improved.In the process of solution,the initial optimization problem is decomposed and the optimal allocation ratio of computing resource and communication resource in the edge server is represented by offloading decisions.Then,the hybrid heuristic computation offloading decision algorithm is used to solve the offloading decisions.Finally,the simulation results show that the mechanism has significant advantages over other benchmark methods in energy consumption and the edge server has the highest fairness in energy consumption.2)Improved particle swarm optimization-based computation offloading and caching: Regarding the problems of increased delay and energy consumption caused by the repetitive tasks offloading in real-world scenarios,as well as the degradation of system performance,an improved particle swarm optimization-based computation offloading and caching mechanism is proposed.Specially,it jointly optimizes offloading and caching decisions,local CPU computation speed and edge sever’s transmission power to minimize the total cost.To solve the problem easily,the mechanism utilizes the mathematical optization method to reduce four optimization variables to three variables and add the penalty mechanism to transform the constrained optimization problem into the unconstrained optimization problem.Then,an Improved Particle Swarm Optimization(PSO)-Based Computation Offloading and Caching Decision Algorithm(WPSO-COC)is designed.Based on traditional PSO,the algorithm adopts the adaptive inertia weight to avoid falling into local optimum and obtain the optimal policy which adapts to the dynamic network environment with the ability of autodidacticism.Finally,the simulation results demonstrate that the WPSO-COC can converge at a faster rate and reduce the total cost significantly compared with other methods.3)Deep reinforcement learning-based collaborative computation offloading and caching:Actually,the data is large-scale and high-dimensional with various types of data including both continuous and discrete variables.The network environment is dynamic and constantly changing.To further adapt to the network environment with time-varying wireless channels and dynamically changing service requests and enhance the perception and abilities of deciding,a deep reinforcement learning-based collaborative computation offloading and caching mechanism is proposed.Specifically,it formulates an optimization problem of minimizing the weighted sum of all tasks’ completion time and energy consumption under the constraints of delay,bandwidth,computing capability and energy.To solve the above mixed integer nonlinear programming problem,a Deep Reinforcement Learning(DRL)-Based Collaborative Computation Offloading and Caching Algorithm(DRL-CCOC)is developed.The algorithm jointly optimizes offloading decisions,caching decisions,the occupation ratio of the wireless channel bandwidth and the edge server’s computing capability which is allocated to the task,it can generate the optimal policy and adapt to dynamic network environment with the ability of autodidacticism.Finally,the simulation results demonstrate that the DRL-CCOC can converge at a faster rate and reduce the total cost significantly compared with other methods,they also confirm the strong dynamic adaptability of our algorithm. |
