Effect of probabilistic sensing models in a deterministically deployed wireless sensor network

Abstract

Wireless sensor network (WSN) is typically deployed to monitor, detect or track events either randomly or deterministically depending on the accessibility to the monitored area and the application. In either case, it is imperative to compute the optimum number of nodes to be deployed such that they collectively satisfy spatial and temporal sampling of the network area. This can be quantified in terms of a critical quality of service (QoS) parameter - Network coverage. Deterministic approach is considered to be straightforward wherein number of nodes to be deployed is estimated using simple geometrical computations assuming disk based binary sensing model. However, the effective sensing radius of a node is affected by sensing device characteristics and environmental factors which leads to non-uniform sensing. These factors are stochastic in nature and can be modeled using probabilistic sensing models. In this paper, we investigate the effect of probabilistic sensing models on the required number of nodes to be deterministically deployed using a square grid based scheme in order to attain a given percentage coverage. The analysis highlights the need for consideration of an appropriate sensing model along with the suitable parametric values for estimation of required number of nodes to be deployed for desired coverage. This study based on sensing models can be taken up as a guideline for estimation of the appropriate number of sensor nodes to be deterministically deployed to design, evaluate and implement an effective WSN.