Research On Resource Allocation Algorithm For Edge Computing System Based On Hybrid Energy Harvesting

Adopting radio frequency(RF)energy harvesting technology to harvest energy from electromagnetic wave to provide power for mobile users is a new green and sustainable power supply mode.Moreover,mobile user can offload computing tasks to edge server for computation by introducing edge computing technology.In this thesis,for the energy harvesting communication system combined with edge computing,the hybrid energy harvesting method is considered to provide energy for users,and the resource allocation strategy is investigated to improve the energy consumption,energy efficiency and computing delay performance of the system.The main contributions are summarized as follows:(1)The resource allocation strategy in edge computing system is investigated to minimize the total energy consumption of mobile users.By deploying multiple magnetic induction energy quick charging stations within the coverage area of the base station,mobile user will supplement energy at nearby magnetic induction energy quick charging stations when the user is about to run out of the energy harvested from ambient RF sources.Mobile user can offload computing tasks to edge server for computation.The objective of resource allocation strategy is to minimize the total energy consumption of users under the constraints of maximal computing capability and battery energy of user,maximal computing resources of edge server,and computing delay of tasks.The suboptimal solution is obtained by using the quantum-behaved particle swarm optimization algorithm.Simulations indicate that the suggested method has less energy consumption compared with the standard particle swarm optimization algorithm and the equal allocation method of computing resources of edge server.(2)The resource allocation strategy in relay-user-assisted edge computing system is investigated to maximize the energy efficiency.By employing magnetic induction-based wireless reverse charging technology,mobile user can supplement extra energy from nearby users when the energy harvested from ambient RF sources is insufficient.When the computing resources of edge server under one base station are saturated,mobile users who cannot continue to offload tasks can employ a neighbor user under another base station as a relay node,and transfer their tasks to another server with abundant computing resources under the base station for edge computation.The objective of the resource allocation strategy is to maximize the energy efficiency under the constraints of computing delay and energy harvesting.A suboptimal solution is obtained by adopting the quantum-behaved particle swarm optimization algorithm.Simulations reveal that the suggested approach can effectively increase the energy efficiency compared with no relay node method.(3)The resource allocation strategy in multi-technologies-assisted edge computing system is investigated to minimize the computing delay.Mobile user can connect with nearby users for auxiliary computation by introducing device-to-device communication technology.When the computing resources of edge server under one base station are saturated,an unmanned aerial vehicle(UAV)with a server is introduced.If fewer users are unable to continue offloading tasks,these tasks will be offloaded to the UAV server for edge computation.If there are too many such users,the UAV will also act as a relay node,and transfer too many tasks to another server with abundant computing resources under the base station for edge computation.By introducing caching technology,the computing results of popular tasks can be acquired directly from caching server.The objective of the resource allocation strategy is to minimize computing delay by jointly optimizing the flying trajectory of UAV,the distance decision and transmitting power of users.Combining quantum-behaved particle swarm optimization algorithm with binary quantum-behaved particle swarm optimization algorithm is employed to acquire the suboptimal solution.Simulations show that the proposed method has lower computing delay than the no caching and no transfer tasks methods.