Mobility Assisted Positioning in Wireless Sensor Networks

Finding the physical location of sensor nodes has already been considered in many directions. Typically, physical characteristics of radio and/or ultrasonic wave propagation are exploited to measure absolute or relative distances or orientation of nodes in Euclidean space. With that information at hand plain trigonometric computations, geometric approximations, multilateration techniques, or multidimensional scaling are applied in order to estimate the sensor nodes’ positions. Refer to [1] for an overview of sensor node positioning techniques. We consider one new such positioning technique, first described in [2], which exploits the mobility in robot supported wireless sensor networks. In the original publication of that technique a quite astounding precision has been reported. The studies of that work were based on simulation. Thus, it remains open what happens if the system is actually applied in a real world setting. In this work we give some answers on that question. Extending the simplified method described in the original publication, we propose different techniques to determine the point when a mobile robot is closest to a sensor; which is a necessary requirement for that positioning scheme. We show that opposed to the simulation results reported, less precision is to be expected in a real world setting.