On Stability and Sampling Schemes for Wireless Sensor Networks

We consider a wireless sensor network in which the sensor nodes are sources of delay insensitive traffic that needs to be transferred in a multi-hop fashion to a common processing center. We consider two data sampling schemes: a) the sensor nodes have a sampling process independent of the transmission scheme, and b) the sensor nodes sample new data only when it has a opportunity of transmitting the data. The first system is like the packet radio network for which exact analysis is not available. We also show that the stability condition proposed in the PRN literature is not accurate. A correct stability condition for such a system is provided. It is also observed that the second scheme gives a better performance in terms of delays and moreover is amenable to analysis. We then address the problem of optimal routing that aims at minimizing the end-to-end delays. Since we allow for traffic splitting at source nodes, we propose an algorithm that seeks the Wardrop equilibrium instead of a single least delay path. The algorithm is implemented in TinyOS and numerical results from the implementation are provided.