Analysis and classification of traffic in wireless sensor networks

Abstract

Wireless sensor networks have been widely researched for use in both military and commercial applications. They are especially of interest to the military planners as they can be deployed in hostile environments to collect vital information safely and cheaply. In view of this interest, there is a need to capture and categorize the data effectively under different operational conditions. This thesis captured traffic and data from sensor motes to analyze and present characteristics of the traffic in a meaningful manner. Specifically, this thesis studied the traffic generated by wireless sensor networks by setting up two different commonly used network topologies, namely a direct connection to the base and a daisy-chain connection to base. A total of six experiments were conducted, three for each topology. The data traffic between the nodes was captured over an extended duration of time. Using the captured information, analysis was performed to categorize and identify the information through anomalies and variations of traffic patterns. Data were also analyzed to study self-similarity and statistical distribution. The experimental results have shown that it is possible to differentiate the two different topologies by numbering the traffic distribution or by analyzing the types of messages sent. The status of the nodes can also be determined with the traffic collected. Examples include new nodes joining the network and operational status of the nodes. Statistical analysis has also been done and found that wireless sensor network traffic is not self-similar except for the interarrival time of the direct connection mode.