Joint Optimization Of Communication And Computing Resources In Fog-Radio Access Network

For the past few years,Fog-Radio Access Network(F-RAN),as a new network architecture based on Cloud-Radio Access Network,has been proposed and studied to satisfy the demands of large scale access,low latency and high rate transmission of the fifth generation mobile communication system.F-RAN can provide nearby and distributed services,which can effectively alleviate the blockage of the fronthaul link and the heavy burden of the central processor that C-RAN will confront when there are large scale devices to access,therefore,it has important research significance.Aiming at the heterogeneous F-RAN model of mobile users,fog node,and cloud server co-existing,this thesis optimizes the wireless access control between users and fog nodes and the allocation strategy of communication and computing resources for the purpose of low-latency task off-loading.The specific research contents are as follows:Firstly,considering the F-RAN where the fog node works in hard mode(demodulation and forwarding),a communication and computing resource allocation problem is designed with the goal of minimizing the maximum delay among users.Ufortunately,the original problem is NP hard,which is difficult to solve.Therefore,an approximate algorithm based on the difference of convex(DC)programming and the weighted minimum mean square error(WMMSE)algorithm is proposed in this thesis.Simulation results demonstrate the convergence of the algorithm and the effectiveness of the resulting resource allocation strategy.Besides,compared with the traditional minimum distance access strategy,the access strategy proposed by this thesis is proved to be more effective in reducing delay.Secondly,aiming at the F-RAN where the fog node works in soft mode(quantized forwarding),the problem of communication and computing resource allocation with the goal of minimizing the maximum delay is studied.With the original problem non-convex,DC programming and WMMSE algorithm are used to transform it into a convex problem,and then it is effectively solved under the framework of block coordinate descent(BCD).The simulation results verify the convergence of the proposed algorithm,and by comparing the soft mode with the hard mode in different scenarios,it can be seen that the soft mode is more effective in shortening delay when the calculation ability of the fog node is weak.Conversely,when the calculation ability of the fog node is strong,the hard mode performs better.In addition,when the calculation rate of the fog node is much weaker than the cloud server,as the amount of data describing the task increases,the proportion of users performing calculations at the fog node in both the soft and the hard modes increases gradually.Finally,the physical layer coding is considered to ensure the communication security.Aiming at the F-RAN where an eavesdropper eavesdrops on the communication between users and fog node,the offload ratio of tasks,the transmiting precoding matrix and the computational resource allocation strategy of fog node are optimized to minimize the delay under the condition that the user task can be partially offloaded and only offloaded to the fog node.The proposed optimization problem is transformed from non-convex to convex problem by DC programming and WMMSE algorithm,and then solved by successive convex approximation(SCA).Simulation results show that the algorithm is convergent and the resulting allocation strategy is effective.