中文摘要：

提出了一种对于无线传感器网络中的应用层数据进行非周期性穿级采样算法;该算法着重考虑在满足误差容限的基础上,有效降低数据的发送次数,以达到降低能耗的目的。运用Markov链权衡数据率和丢包率,以确定对信号区间进行层级划分的门限区间;以此为基础,对信号区间可分配为以均匀分布和以u-law算法实现的非均匀分布的层级。考虑丢包对于穿级采样的影响,提出了穿级采样中丢失数据的有限恢复算法。通过变电站带电场景对于3组共45个节点的物理测试,分析了穿级采样算法的可靠性和能耗,验证了该种采样算法的可用性。

英文摘要：

A novel aperiodic level crossing sampling（LCS） algorithm is proposed for the application-layer data of wireless sensor networks（WSNs）.The level crossing sampling algorithm is used for energy conservation in WSNs,and it can decrease the times of data transmission and maintain data accuracy.The Markov chain is creatively used to balance the data transmission rate and packet loss rate,and decide the threshold interval for the level partitioning in the level crossing sampling.Two versions of level quantization,which are respectively uniformly distributed quantization and non-uniformly distributed quantization with u-law,are introduced.A limited data recovery algorithm for LCS is suggested to deal with the effect of packet loss on LCS.3 groups of total 45 nodes in an experimental scenario of transformer substation were tested,and the working reliability and energy consumption were analyzed,which verifies the availability of the level crossing sampling.