移动边缘计算中的分层资源部署与共享策略

Hierarchical resource deployment and sharing strategy in mobile edge computing

随着5G时代的到来,万物互联逐渐成为现实,然而,物联网中的很多终端因计算资源不足导致功能受限。移动边缘计算是一种新型网络计算模式,其终端设备可以将资源敏感型应用卸载到附近计算资源充足的边缘节点,从而降低运行成本/时延,提高服务质量(QoS,quality of service)。针对移动边缘计算资源部署问题进行研究,考虑网络服务提供商(ISP,Internet service provider)支付部署资源的费用,同时向终端设备出租资源以获得收入的场景。由于物联网终端设备的资源需求是快速随机变化的,而边缘计算资源的部署是一个长期缓慢变化的过程,因此,合理的资源部署策略至关重要。提出了一个基于5G网络模型的分层网络结构,其中,计算资源可以部署在基站端、汇聚端以及云端等不同的网络层级。基于该分层网络架构,最优资源部署问题被构建为一个以ISP的利润最大化为优化目标的混合整数规划问题,并基于MATLAB中的CVX工具箱对该问题进行求解。仿真结果表明,所提出的移动边缘计算资源分层部署方案在ISP收入和部署成本方面的性能均优于平坦型资源部署方案。

With the coming of 5G,Internet of things(IoT)has gradually become a reality.However,the function of many terminals is limited to insufficient computing resources in IoT.Mobile edge computing(MEC)is a new network paradigm,where terminals can offload the resource sensitive applications to nearby edge services with rich resources,and thus reduced the operational cost/delay and increased the quality of service(QoS).The resource deployment of MEC was studied,a scenario was considered that the Internet service provider(ISP)paid for deploying the MEC resources,and meanwhile gained revenue from leasing resources to terminals.Note that the deployment of MEC resource is a long-term strategy,while the demands from terminals are time varying.Therefore,it is critical to deploy the MEC resource properly.A hierarchical architecture built upon the 5G radio access network for MEC resource deployment and sharing was proposed,where the MEC resource could be deployed at different network levels.Based on the hierarchical network architecture,the optimal MEC resource deployment problem was formulated and solved as a mixed integer programming problem,aiming at maximizing the ISP’s revenue.And the CVX toolbox in MATLAB was used to solve the problem.Simulation results demonstrate that the proposed solution outperforms the flat resource deployment solution in terms of both the ISP’s revenue and the deployment cost.