分布式卫星云雾网络及时延与能耗策略

为了解决分布式卫星的地面云计算中心架构存在的高传输时延问题，提出分布式卫星云雾网络（DSCFN）架构，由小卫星编队飞行组成卫星雾网络，根据地面站云计算得出的任务划分比例直接进行本地分布式计算，降低业务处理时延.由于卫星的计算能力较弱，时延降低将导致能耗增加，卫星工作寿命减短，为此提出均衡时延和能耗的策略，利用改进的粒子群优化（MPSO）算法，解决能耗约束下的时延优化问题，达到时延和能耗折中的目标.仿真结果表明，基于MPSO算法得出的任务比例进行分布式计算，可以在能耗约束条件下，有效地降低卫星雾网络的任务处理时延，满足时延敏感型业务的需求；由10颗小卫星组成的DSCFN处理1 Gb数据的时延相比地面云中心降低了90.7%.
Abstract: To resolve the problem of high transmission latency in the distributed satellite cloud computing center architecture, a distributed satellite cloud fog network (DSCFN) architecture was proposed. The satellite fog network consisted of small satellites, and carried out distributed computing locally according to the task partition ratio calculated from the earth station cloud. As a result, the task processing latency was reduced. The computing power of satellites was weak and the reduction of the latency would lead to incensement of power consumption, thus, the working life of satellites was shortened. A balanced strategy of latency and power consumption was proposed to achieve the tradeoff between power consumption and latency by leveraging a modified particle optimization (MPSO) algorithm. Simulation indicates that the MPSO algorithm for distributed computing reduces the task processing latency of satellite fog network efficiently and meets the demand of latency-sensitive applications under the power consumption constraints. Compared with the ground cloud computing center, the latency performance of DSCFN was increased by 90.7% when using 10 small satellites and processing 1 Gb data.