PMSW: a passive monitoring system in wireless sensor networks

Network passive monitoring is an important method in debugging and analyzing wireless sensor networks. The passive monitoring method for wireless sensor networks (WSN) has the following problems: monitoring traces are incomplete and contain only limited information because some packets may be misheard; how to infer implicit information about network protocol behavior and how to detect network events from incomplete traces are challenging issues in passive monitoring system. To resolve the above issues, we propose the use of time tuning in trace merging and a finite state machine for inferring missing packets, implicit information and network event detection. (1) In order to enhance the integrity of traces, we propose clock-adjusting strategies including clock fine-tuning and clock self-adjustment to solve clock drifting in trace merging. (2) For inference of missing packets and implicit information, we devise and implement a finite state machine to infer missing packets and packet receiving state for multi-hop network protocols. (3) For network event detection, we propose an XML-based description of event rule and network event visualization methods. These methods are implemented in PMSW (passive monitoring system for WSN). We verify the accuracy of merging and inference modules with monitoring traces in the Collect Tree Protocol sensor network. Experiments show that the finite state machine can infer 40-60% of missing packets and more than 90% of the receiving state of data packets, when the trace has more than 50% of the data packets captured in the network. Copyright (C) 2011 John Wiley & Sons, Ltd.