基于网络断层扫描的传感器网络测量技术

提出一种基于断层扫描技术的传感器网络测量方法，通过在网络边界节点获取网络自身传输的数据报文，分析推测网络的性能。针对传感器网络的特点，提出在被动测量条件下的报文丢失率模型，给出MLE-EM算法的实现。仿真试验证明，算法在不增加节点负担的前提下，能够有效推测链路的报文丢失率，随着网络规模的扩大，仍能准确地反映网络内部的链路性能。

Sensor Network Measurement Technology Based on Network Tomography

This paper proposes a performance measurement method based on network tomography techniques in sensor network. It analyses the data which is transfered in sensor network received by the sink node and infers the sensor network performance characteristics. In response to the nature of sensor network, it introduces the loss model based on passive measurements and solves the loss inference problem by using MLE-EM algorithm. Simulation results demonstrate that the methodology can infer the link loss rate accurately without any additional overhead for sensor network, and scale to large sensor network.