Opportunistic detection of relative mobility in wireless sensor networks

In highly mobile ad hoc networks the neighbourhood status evolves rapidly. Routing tables which are populated with an “out of date” view of the neighbourhood produce an overhead in the routing process which can affect the QoS of the communication and the stability of the network interconnectivity. To maintain correct network state information, periodic “Hello” messages are often employed. In many scenarios these may be expensive in terms of communication and energy. Providing a node with an ability to estimate mobility has been proposed as a mechanism for regulating the transmission of discovery “Hello” messages and for prompt updates of its neighbourhood view. In this paper, a mechanism for gauging the relative mobility of a node with respect to its neighbourhood is presented for mobile Wireless Sensor Networks. The algorithm employs a set of Bloom filters to encapsulate opportunistically eavesdropped data from neighbours. A set of mobility states is defined using a model based on the number of neighbours recently cached, a membership comparison of Bloom filters, and a probabilistic expression. The mechanism detects changes in the neighbourhood using only the available information. The evaluation of the system has been performed (using a gradient-based routing protocol) in the TinyOS simulator, Tossim.