Research On Computation Offloading And Resources Allocation Of Mobile Edge Computing

With the popularity of smart mobile equipment and the rapid evolution of new computationally intensive and delay-sensitive applications,the delay and energy consumption of cellular heterogeneous networks are becoming more prominent,as well as,put forward higher requirements to data rate demands for cellular network and the implementation of green communications.To cope with these requirements,a emerging concept known as mobile edge computing(Mobile Edge Computing,MEC)has been prompted.Deploying computational capabilities on base stations at the edge of mobile network,combined with the two-layer structure of the cellular heterogeneous network,so that the mobile device users can offload their computational tasks to the MEC servers for processing,thereby efficiently reduce user-aware delay and energy consumption.The core of this thesis is focused on the three challenges of user-oriented use cases in MEC system,which are to determine the users' computation offloading,spectrum resource allocation and MEC computation resource allocation.In view of the shortcomings of existing strategies,this paper investigate three different system models under cellular heterogeneous networks.With the goal of minimizing delay and energy consumption,two different computation offloading strategies and one joint computation offloading,spectrum allocation and computation resource allocation strategy are proposed.Firstly,an offloading strategy based on users' selection is proposed.The MEC system has a macro cell base station(Macro cell base station,MBS)deploying a MEC server and a small cell base station(Small cell base station,SBS)that can serve as a relay.An optimization problem is formulated to minimize delay and energy consumption of the MEC system.Limited by the binary and inequality constraints,the problem is NP-hard.Therefore,this strategy adopts an optimized greedy algorithm.Simulation results validate the utility and effectiveness of this strategy.Secondly,a computation offloading strategy based on alternating direction method of multipliers(Alternating Direction Method of Multipliers,ADMM)algorithm is proposed.In this section's MEC system,the MEC servers are deployed on both the MBS and the SBS in the system.The optimization problem minimizing user-aware delay and energy consumption is formulated and transformed into convex.Then,the ADMM is used to achieve efficient solution.The simulation results show that this strategy can save more delay and energy consumption than the first strategy,and the speed of solving decision results is significantly improved.Finally,a strategy jointly consider computation offloading,spectrum resource allocation and computation resource allocation,which is based on ADMM and Global Consensus is proposed.The MEC servers are deployed on MBS and multiple SBS in hierarchical and distributing way in this system.The optimization problem is formulated to minimize the user-aware delay and energy consumption,and it is transformed into convex.The Global Consensus concept is adopted to solve the optimization problem in distributed manner.The simulation results show that this strategy performs better than the former two strategies in terms of saving delay and energy consumption,and the distributed solution is more efficient.With regard to the latter two strategies,this thesis also discusses the influence of different values of the factors affecting the algorithm convergence on the speed of solving the decision results through MATLAB experiments.