CAJ | 학술논문

研究有QoS偏好要求的云存储任务调度。首先，分析云存储与云计算的差异，用存在矩阵避免无效解的产生。其次，归纳云存储的QoS需求为时间、代价与质量3大类，并据此修改PSO算法的适应度函数用权重因子调节QoS偏好。实验发现，在不同优先级任务分布不均的情况下，分布广的任务的偏好会掩盖其他任务的偏好，因而不适宜用PSO进行整体性调度，而必须进行分级调度。实验证明，改进后的分级PSO算法对QoS偏好具有较好的感知能力。
연구유QoS편호요구적운존저임무조도。수선，분석운존저여운계산적차이，용존재구진피면무효해적산생。기차，귀납운존저적QoS수구위시간、대개여질량3대류，병거차수개PSO산법적괄응도함수용권중인자조절QoS편호。실험발현，재불동우선급임무분포불균적정황하，분포엄적임무적편호회엄개기타임무적편호，인이불괄의용PSO진행정체성조도，이필수진행분급조도。실험증명，개진후적분급PSO산법대QoS편호구유교호적감지능력。
The task scheduling algorithm in cloud storage area with QoS preference awareness was studied. Firstly, the differences between cloud storage and cloud computing were pointed, and exist matrix was used to avoid the creation of useless solutions. Secondly, QoS factors were summarized into three classes:time, cost and quality classes. Based on the three classes, the fitness function of the PSO was improved by using weighted factors to adjust the QoS preference. The simulation results release that PSO is not fit for the overall scheduling when the task distributions of different degrees are unevenness because the QoS demand of wide distribution tasks may cover the QoS demand of other tasks. So a hierar-chical scheduling based on degree is necessary. The simulation results show that the improved hierarchical PSO schedul-ing algorithm can aware the QoS preference relevantly.