Stochastic node placement improving fault tolerance in wireless sensor networks

One of the most important issues in wireless sensor networks is to develop the technology for improved fault tolerance, since sensor nodes are prone to failure (for mechanical reasons, for example) and have limited energy storage. The strategy for determining the positions in which sensor nodes are placed is very important, because it affects the, likelihood that a target will be within the sensing range of a node, and that there will be a communication path from that node back to the base station. We have considered that when many sensor nodes are to be placed in a wide area, an effective approach is stochastic node placement, whereby sensors may be scattered in a controlled manner, such that their approximate positions are characterized by a probability density function. Then, as the first step of our research, we have proposed stochastic node placements in which the degrees of the nodes follow a power-law distribution ("power-law placement"). Through simulation studies, we have shown that power-law placement, with well-tuned parameters, shows a higher tolerance against random failures and failure through battery exhaustion than general stochastic node placement. However, this can be difficult to implement. As the second step of our research, this paper proposes an alternative method of stochastic node placement that has as high a fault tolerance as power-law placement and can be achieved with a reasonable degree of complexity. To this end, we first investigate the requirements necessary for other types of stochastic node placement to exhibit as high a fault tolerance as power-law placement. Next, we propose ways of achieving stochastic node placement that meets these requirements. Our proposal has theoretical basis that any stochastic node placement can be achieved by scattering sensor nodes many times from the air in a number of appropriately controlled operations. (C) 2006 Wiley Periodicals, Inc.