基于资源预测的智能终端资源缓存算法

发布时间：2019-01-16 03:49

【摘要】：针对智能电视终端应用间资源竞争导致的系统性能下降问题,基于资源消耗预测,提出一种智能终端资源缓存算法。根据系统记录的各应用程序的资源消耗统计数据,应用Markov模型预测下一时间段可能出现的资源瓶颈和应用的资源状态,利用应用的资源状态动态调整应用权重,并以最小化应用切换时间为目标,将资源缓存问题转化为多维多选择背包问题,采用轻量级的启发式算法求解资源缓存问题。仿真实验结果表明,在智能终端中该算法对于资源消耗的预测精确度比其他算法提高5.4%,而应用响应时间缩短约45%。
[Abstract]:Aiming at the degradation of system performance caused by resource competition among intelligent TV terminal applications, an intelligent terminal resource cache algorithm is proposed based on resource consumption prediction. According to the resource consumption statistics of each application program recorded by the system, the Markov model is used to predict the resource bottleneck and the application resource state in the next time period, and the application weight is dynamically adjusted by the application resource state. Aiming at minimizing the application switching time, the resource cache problem is transformed into a multi-dimensional multi-selection knapsack problem, and a lightweight heuristic algorithm is used to solve the resource cache problem. Simulation results show that the prediction accuracy of the algorithm for resource consumption in intelligent terminals is 5.4 higher than that of other algorithms, while the application response time is shortened by about 45%.
【作者单位】：中国科学院声学研究所国家网络新媒体工程技术研究中心;中国科学院大学;
【基金】：国家科技支撑计划基金资助项目“电视商务综合体新业态运营支撑系统开发”(2012BAH73F01) 中国科学院先导专项课题基金资助项目“智能电视平台与服务支撑环境研制”(XDA06040501)
【分类号】：TP333

【共引文献】