面向电力物联网异构信息安全的多任务调度算法

Multi-task Scheduling Algorithm for Heterogeneous Information Security of Power Internet of Things

新型电力系统的建设促使电力信息安全的服务呈现海量多元化,单一物理机或云计算虚拟机均无法满足各类服务任务执行的安全性和时效性需求。因此,面向电力物联网信息安全服务体系中特有的物理密码机与虚拟密码机共存的新型混合底层硬件模式,通过对异构计算资源的协同调控,实现在确保任务安全性的约束条件下的系统执行效率优化。首先,设计了一种三级双映射的电力物联网信息安全服务多任务调度架构,描述多轮交互保密任务和异构计算资源的映射规则;然后,基于加权深度嵌入聚类算法确定安全服务任务的执行标签;最后,采用朴素贝叶斯和量子粒子群优化算法完成运算节点的创建与任务映射。仿真结果表明,所提聚类映射调度算法能利用任务映射规则匹配新型异构安全服务计算资源,与双阶段调度算法和非支配排序遗传算法相比,其任务安全保障系数最大可提升58.4%,总完成时间最高可缩短22.24%。

The construction of a novel power system has promoted massive diversification of the power information security service.Neither the single physical machine nor the cloud computing virtual machine can meet the security and timeliness requirements of the execution of each type of service tasks.Therefore,for the new hybrid underlying hardware pattern with the coexistence of the physical cryptographic machine and the virtual cryptographic machine in the information security service system of power Internet of Things,the system execution efficiency can be optimized under the constraint of ensuring the task security by coordinating the heterogeneous computing resources.First,a three-level double-mapping multi-task scheduling architecture for the information security service of power Internet of Things is designed,which can describe the mapping rules between multi-round interactive secure tasks and heterogeneous computing resources.Then,the execution label of the security service task is determined based on the weighted deep embedding clustering algorithm.Finally,the naive Bayes and quantum particle swarm optimization algorithms are used to complete the creation of computing nodes and the task mapping.The simulation results show that the proposed clustering mapping scheduling algorithm can use the task mapping rules to match the new-type heterogeneous security service computing resources.Compared with the two-stage scheduling algorithm and the non-dominated sorting genetic algorithm,the task safety factor of the proposed algorithm can be increased by up to 58.4%,and the total makespan can be shortened by up to 22.24%.