低轮值传感网链路自适应的缓存优化机制研究

The low-duty-cycle sensor networks face many irreliabilable problems,such as data forwarding delay and packet loss. These problems are research issues in sensor networks and IoTs. This program transforms the relationship between data forwarding and cache to a tradeoff problem between energy consumption and packet loss based on link stability. The research aim is to research the cache optimation and implement data aggregation and transmission on demand in order to optimize the above tradeoff problem in the low-duty-cycle sensor network applications. The research strategy and the creativity of this program is to define and update a cache threshhold in each sensor node based on predicting the stability of the link quality first. Then each node can compare the cache threshhold and cache capicity in order to determine the time of switching the data cache and forwarding adaptively and guranteeing the reliablity of data forwarding, which can impove the probability that more data could be received by sink node. This program mainly includes link stability measurement and prediction,cache threshhold caculating,cache threshhold dissmission based on minmal delay tree, and data cache and forwarding protocols. The main research issue includes link stability prediction, data cache threshhold caculating and high efficiency data transmitting.This program transforms the link stability problem into a time series problem, which is a fine-grained research based on the existing coarse-grained link measurement techniques. For the cache threshhold updating problem, a threhhold disseminating tree will be proposed, in order to implement adaptive cache threshhold caculating and optimization.

低轮值传感器网络面临着数据转发延迟、丢包等不可靠问题，是目前传感器网络和物联网等研究领域的热点和难点。本课题在综合考虑链路稳定性的基础上将数据缓存和转发之间的关系转化为能量开销和数据丢包之间的权衡问题。研究目的是实现一套缓存优化机制，支持低轮值传感器网络中的按需数据聚合传输,降低数据丢包率。研究思路和创新点是提出和利用缓存阈值模型动态管理数据缓存和转发,在对链路稳定性测量和预测的基础上，研究全网范围缓存阈值优化机制；在该模型的基础上，推导节点数据缓存或者转发的时机，研究数据在传输过程中的链路自适应性和可靠性理论，分析该机制条件下数据到达基站节点的机会和能量效用。本项目专门针对低轮值传感器网络的特点,将链路平稳性问题转化为时间序列，是对现有的链路质量测量技术在长时间统计过程的细粒度研究。针对缓存阈值的更新问题，主要研究如何构建和优化缓存阈值分发树，实现全网范围自适应的缓存阈值迭代计算和优化。

本项目的研究基本按照原计划执行，一些研究内容稍作调整。在项目执行期间，按计划主要研究了低轮值传感器网络中基于链路质量调整节点的缓存阈值的问题，提出了一套无线传感器网络节点数据缓存转发控制系统及方法。实现了根据节点的工作周期和链路质量的稳定性，对节点的缓存进行初步优化，自适应地调整节点转发数据的时机。基于该系统方法，针对水下低轮值传感器网络问题，重点研究了基于复合AUVs的水声传感器网拓扑优化机制，制定了基于复合自主式水下航行器的水声传感器网络拓扑智能优化策略（TO-DA机制），提高了网络连通性；进一步发现水声传感器网络拓扑结构形成的内在规律，提出了三角淡化关键节点网络拓扑优化方法，有助于网络抗毁再生。针对低轮值传感器网络节点缓存容量有限，结合物联网的背景研究并提出了一种结合可感知条码技术进行环境感知的策略。在研究的过程中，结合了部分云计算的研究思路，研究计划做了一些调整，部分经费也资助了云计算领域的研究工作。依托本项目发表论文10篇，撰写专利5篇（3篇授权），取得一项军队科技进步三等奖一项，培养研究生4名。