融合缓存管理与资源分配的能耗均衡内容服务研究

Cache Management and Resource Allocation Based Energy-Balancing Content Service Mechanism

为了构建高效与长生命周期的内容服务物联网,基于雾节点决策、云服务决策及雾节点与云服务器带宽占比的联合优化考量,本文规划了一个最小化内容服务总能耗的优化问题。同时,提出了融合LRU-2和牛顿动量的公平能耗最小化算法(LRU-2 and Nesterov Momentum based Fair Energy Minimization Algorithm,LNM-FEM)用于解决该优化问题。具体地,基于雾节点的历史平均能耗、剩余能量及距离设计了公平性度量指标,根据该度量与雾层缓存情况获得能耗公平性最优的雾节点决策及云服务决策;在雾层没有命中的情况下,当云层将内容反馈给相应雾节点后,雾节点根据本文提出的缓存策略,将内容缓存至雾节点;基于所获得的最优雾节点决策与云服务决策,通过牛顿动量法,联合优化雾节点和云服务器的带宽占比,达到最小化内容服务总能耗的目的。最后,仿真结果表明所提算法具有收敛速度快、命中率高等特点,且与其他三种基准方案相比,总能耗最低、雾节点能耗均衡性最高、网络寿命分别平均提升了23%、28.7%和34%。

In order to construct an efficient and long lifetime Internet of Things(IoT)for content service,based on the joint optimization consideration of fog node decision,cloud service decision,and the bandwidth occupation ratios of fog node and cloud server,an optimization problem is formulated to minimize the total energy consumption in the content service.Meanwhile,a LRU-2 and Nesterov momentum based fair energy minimization algorithm is developed to solve such optimization problem.Specifically,a fairness metric is designed based on the historical average energy consumption,the residual energy and the distance between fog node and the IoT device.According to this metric and fog layer’s caching state,the fog node decision with the optimal fairness of energy consumption and cloud service decision can be obtained.In the case of failed hit in the fog layer,the fog node will cache the feedback content from cloud according to the developed caching strategy in this paper.Based on the optimal fog node decision and cloud service decision,Newton momentum is employed to jointly optimize the bandwidth ratios of fog node and cloud server and achieve the minimization of the total energy consumption in the content service.Finally,the simulation results show that the proposed algorithm possesses the fast convergence speed and high hit rate,and compared with three other benchmark solutions,it has the lowest total energy consumption,the highest energy consumption balance of fog node,and the average network lifetime is improved by 23%,28.7%and 34%,respectively.