面向物联网重编程的存储优化方法研究

发布时间：2018-04-11 05:40

本文选题：物联网 + 无线重编程　；参考：《计算机学报》2017年08期

【摘要】：未来在大多数物联网应用场景中,将允许使用者根据自己的需求复用当前已部署的物联网节点设备.从而要求节点设备具备对本地软件程序重编程的能力.然而,现有重编程方法普遍存在着传输和重组开销过高的问题.在工程实践中,通过使用大容量的扩展存储单元升级代码,可以有效避免由于传输代码引起的传输开销.因此,当需要对物联网节点软件频繁升级时,重编程的能量开销将主要取决于在节点本地读写代码引起的重组开销.为此,我们提出一种基于分页机制的重编程存储优化方法——AdvCache.该方法的核心思想是将相似的代码段以函数为单位聚合为函数页,放入低功耗的缓存中保存并执行.通过提高缓存的使用率和命中率,减少对高功耗存储单元的读写次数.此外,为了应对节点软件可能发生急剧变化的情况,我们提出了一种基于函数页调用关系的缓存替换算法,通过引入程序结构因素,进一步提高缓存命中率.实验表明AdvCache能够有效地降低对高功耗存储单元的读写操作数量.与先前的工作相比,AdvCache的缓存命中率提高了22%.
[Abstract]:In the future, in most Internet of things applications, users will be allowed to reuse the currently deployed IoT node devices according to their own needs.The node equipment is required to have the ability to reprogram the local software program.However, the existing reprogramming methods generally have the problem of high transmission and reorganization overhead.In engineering practice, the transmission overhead caused by the transmission code can be effectively avoided by using a large capacity extended storage cell to upgrade the code.Therefore the energy cost of reprogramming will mainly depend on the reconfiguration overhead caused by reading and writing code locally at the node when the software of the Internet of things node is frequently updated.Therefore, we propose an optimization method of reprogramming storage based on paging mechanism-Advanced Cache.The core idea of this method is to aggregate the similar code segments into function pages and save and execute them in a low power cache.By increasing the cache usage and hit ratio, the number of high power memory cell read and write is reduced.In addition, in order to cope with the sudden change of node software, we propose a cache replacement algorithm based on function page call relationship, which can further improve the cache hit rate by introducing the program structure factor.Experiments show that AdvCache can effectively reduce the number of read and write operations for high power storage cells.Compared with previous work, the cache hit ratio of Advanced Cache has increased by 22%.
【作者单位】：浙江工业大学;
【基金】：国家自然科学基金(61502427,61379122,61502428,61379023) 浙江省自然科学基金(LY16F020034,LR16F010003)资助~~
【分类号】：TN929.5;TP391.44

【参考文献】