基于差异距离反馈的多执行体裁决算法与调度算法

Multi executor decision algorithm and scheduling algorithm based on differential distance feedback

针对现有多执行体裁决算法和调度算法主要将执行体受攻击的次数作为可信度的参考依据,难以区分每次攻击行为的恶意程度,也难以应对执行体输出为数值且允许存在误差的应用场景的问题,以电力系统状态估计异常检测作为应用场景,提出根据归一化的执行体输出差异距离调整其可信度的裁决算法,以及基于运行时长、可信度和切换开销等构造收益函数的调度算法。仿真实验结果表明,相比同等权重的裁决算法和随机切换的调度算法,所提算法在系统执行开销基本相同的情况下,可以将系统平均失效率降低43.8%,能够有效地提升工业网络防护设备的防御能力。

With the anomaly detection of power system state estimation as the application scenario,this paper proposed a decision algorithm that adjusted executor’s credibility according to the normalized output differential distance,and proposed a scheduling algorithm that constructed a revenue function based on runtime,credibility,and switching overhead,to solve the problem that the existing multi executor decision algorithms and scheduling algorithms can not distinguish the malicious degree of each attack and deal with the application scenarios where the output was numerical while allowed errors since they mainly took the attack number as the reference basis for the executor credibility.The simulation results show that compared with the decision algorithm with the same weight and the scheduling algorithm with random switching,the proposed algorithm can reduce the average failure rate of the system by 43.8%,and effectively improve the defense capability of industrial network protection equipment.