An automated sensor nodes’ speed estimation for wireless sensor networks

Wireless sensor networks (WSNs) have become more and more notorious thanks to their numerous advantages. But, some of the WSN weaknesses, inherent to sensor nodes’ particularities (low memory, finite battery, etc.), make these networks vulnerable especially for some particular scenarios such as nodes’ mobility which alters the correct network functioning and completely compromises its normal behavior. Thus, we propose in this paper a novel mobility prediction model called the general Bayesian-based mobility prediction (G-BMP) model where sensor nodes’ speed values are derived based on a Bayesian inference paradigm and upon the occurrence of “expired links” and “non-expired links” events. Moreover, to make the implementation of G-BMP possible on sensor devices, we introduce some simplifications during the computation and the transmission of speed distributions. The evaluation of G-BMP using python illustrates the accuracy of the model in deriving the correct speed values in a timely manner. We also compare the performance of G-BMP to the native BMP model that only considers the expired link events when updating the nodes’ speed distributions. The results show that the convergence to real speed values within sensor nodes is faster with G-BMP than that with the native BMP model. In addition, all the simulations illustrate the accuracy of the simplifications used to reduce the overhead generated by the frequent exchange of speed distributions.