Research On Computation Offload Method In Maritime Mobile Edge Network

With the development of blue economy,the number of marine intelligent mobile terminals(such as unmanned patrol ships,buoy sensors,intelligent mobile devices,etc.)is increasing exponentially,and the demand for computing resources of marine mobile terminals is increasing.The rapid development of land communication technology makes people adapt to the life of having the Internet.The high dependence on the Internet in the land environment makes people involved in the sea highly dependent on the marine Internet of things.Using Internet technology,the marine Internet of things connects intelligent devices with each other,constructs an Internet of things covering the marine environment,obtains a large number of marine data and carries out real-time analysis and processing,so as to realize the systematic management of marine equipment.A large amount of data generated by the marine Internet of things needs to be transmitted and processed.How to use limited communication and computing resources to deal with a large number of marine mobile terminal tasks is a hot issue.By providing computing,communication resources and cloud services,mobile edge computing has become an effective way to support computing intensive tasks in the Internet of things.This paper studies the computing offload of mobile edge computing in the network in order to reduce the computing offload cost of offshore mobile terminals.The main contents include:1)Aiming at the problem of high computational offload cost and difficult trade-off between various performances in the offshore network structure with multi terminal and single edge nodes,a traffic tariff aware computational offload cost function model is proposed.The model introduces traffic tariff into computational offload cost in order to trade off delay,energy consumption and traffic tariff.A computing offload algorithm based on deep Q network is proposed.The algorithm realizes the joint computing offload and computing resource allocation algorithm under the constraint of task delay.Simulation results show the effectiveness of the algorithm,which can balance the delay,energy consumption and traffic charges,and reduce the cost of task offloading.2)Aiming at the high cost of computing offload and the high dynamics of network nodes in the new maritime mobile edge network,a computing offload algorithm based on reinforcement learning is proposed to realize the decision-making of computing offload,the allocation of computing resources and communication resources.An agent is trained by reinforcement learning actor critical algorithm.The agent can weigh three performance indexes: delay,energy consumption and traffic charge,and control task offloading decision and resource allocation according to system parameters.Finally,simulation results show that the proposed algorithm can effectively reduce delay,energy consumption and traffic charges.3)In order to further improve the experimental design of this paper and more intuitively show the effectiveness of this algorithm,based on the above original experimental scheme,a simulation platform is designed to show the calculation offloading decision and calculation offloading performance.Then,this paper studies embedding the algorithm into the simulation platform.Among them,the mobile terminal and the edge node are distinguished by different graphics,and the calculation and offloading decision of the terminal task is represented by color.Finally,the simulation platform is implemented on the experimental prototype.By testing the simulation platform,it is proved that the simulation platform can intuitively display the calculation offloading decision,so as to verify the effectiveness of the algorithm in this paper.