Collaboration Between Computation And Communication In Fog Radio Access Network

Fog radio access network(F-RAN)is a new type of RAN with centralized management and distributed computing capability.F-RAN has the characteristics of cooperative radio and computational resource virtualization,which contributes to large capacity,low latency,and high energy efficiency.Naturally,F-RAN plays the key role in the fifth generation mobile network(5G)and beyond 5G.F-RAN is comprised of three key components: radio cloud centre(RCC),fog node(FN),and transport network.Therein,the RCC,constructed by general purpose processors(GPPs),is responsible for the collection and management of global network information,which enables coordinated multi-point(Co MP)at FNs and coallocation for entire network resources(e.g.,computational and communication resources).The FN,formed by the remote radio head and GPPs,is capable of edge computing and radio frequency for fast and efficient computing and communication response.Moreover,the connection between the RCC and the FN is built by the transport network.In such a novel RAN,reasonable resource virtualization and allocation can improve communication resource(including radio resource and transport network resource)and computational resource utilization,reduce system energy consumption,and meanwhile provide guaranteed service for a variety of emerging mobile services.In F-RAN,CoMP at FNs can be leveraged to improve link reliability and provide largecapacity data transmission service for data-intensive mobile services.On the other hand,distributed computing between the RCC and FNs can be exploited to migrate services to network edge and provide low-latency application-level data processing service for computation-hungry mobile services.Therefore,it is challenging for F-RAN to reduce system energy consumption while guaranteeing the quality of service(Qo S)for various emerging mobile services.In detail,1)relationships between processing and communication processes are different for mobile services.For instance,data packet transmission is the core of data-intensive mobile services,during which the data processing is before the data transmission.Whereas,the key point for computation-hungry mobile services is application-level data processing,before which data transmission happens.Then,the first challenge is how to configure computing functions in the RCC and FNs through virtualization technology to support various mobile services.2)Qo S requirements are different for mobile services.For example,average packet delay is guaranteed for delay-tolerant and data-intensive mobile devices,transmission rate is guaranteed for delay-sensitive and data-intensive mobile devices,and application execution delay is guaranteed for delay-sensitive and computation-hungry mobile services.Then,the second challenge is how to build the Qo S-based computation-communication resource transformation model in F-RAN to lay a foundation of flexible network resource provisioning for significant system energy consumption reduction,with guaranteed Qo S requirements for mobile services.In this regard,this paper studies the collaboration between computation and communication in F-RAN,considering various emerging mobile services.The details are given as follows:1.For delay-tolerant and data-intensive mobile services,a green computation and communication collaboration method is proposed based on minimum-weight virtual machine(VM)matching and hybrid FN clustering,by which system power consumption is minimized subject to users' average packet delay.Firstly,through virtualization technology,user-level and cell-level processing functions used for radio frequency are deployed in the RCC and FN,respectively,that is,VMs are formed in F-RAN to support the required processing functions for data packet transmission,to mitigate the burden on transport network while enabling Co MP at FNs.Then,the tandem queueing model is adopted to depict the user-level processing in the RCC,as well as,the cell-level processing and wireless transmission at FNs.Based on the deduced average packet delay,a computation-communication resource transformation model is built in F-RAN.Finally,by jointly optimizing VM allocation and FN clustering,computational and communication resources are flexibly provisioned to minimize system power consumption under the constraint of users' average packet delay.Simulation results show that the proposed method can strike a balance between computation and communication power consumption,resulting in significant system power savings.2.For delay-sensitive and data-intensive mobile services,a green computation and communication collaboration method is proposed based on the iterative water-filling-like computation effort(CE)distribution algorithm and FN beamforming,by which system power consumption is minimized subject to users' signal-to-interference-plus-noiseratio(SINR).Firstly,through virtualization technology,VMs are respectively formed in the RCC and FNs to complete the user-level and cell-level processing functions for data packet transmission,to mitigate the burden on transport network while enabling Co MP at FNs.Then,through Intel's fitting function,which depicts the relationship among CE,number of transmit antennas,and bit efficiency,the computationcommunication resource transformation model in F-RAN is deduced on the basis of users' SINR.Finally,by jointly optimizing VM placement and FN beamforming,com-putational and communication resources are flexibly provisioned to minimize system power consumption under the constraint of users' SINR.Simulation results show that the proposed method can significantly reduce the number of active processors and system power consumption.3.For delay-sensitive and computation-hungry mobile services,a green computation and communication collaboration method is proposed based on the iterative adjustment of VM frequency and radio transmit power,by which system energy consumption is minimized subject to users' application execution delay.Firstly,through virtualization technology,VMs are formed in both the RCC and FNs to complete the processing functions for data communication and application-level data processing,to enable parallel computing between the RCC and FNs while mitigating the burden on transport network.Then,due to the known data transfer amount and CE,computational and communication resources can be transformed into the unified time dimension to build the computation-communication resource transformation model in F-RAN on the basis of users' application execution delay.Finally,by jointly optimizing VM frequency and radio transmit power,computational and communication resources can be flexibly provisioned to minimize system energy consumption under the constraint of users' application execution delay.Simulation results show that the proposed method can not only reduce the minimum affordable application execution delay,but also reduce the system energy consumption while guaranteeing the application execution delay.