Interference and Outage in Clustered Wireless Sensor Networks with Cluster-Centric Data Collectors

In many wireless sensor networks, the distribution of sensor nodes involved in data transmission may be clustered as induced by the underlying geographical factor or protocol design. Instead of using the homogeneous Poisson Point Process (PPP), related work has investigated the Poisson Cluster Process (PCP) for modeling the location distribution of sensor nodes and obtaining analytical results such as the aggregate interference and outage probability for such networks. Many research endeavors, however, assume that data collectors are randomly deployed independently of the sensor nodes. While such an assumption lends itself for mathematical tractability, it is not typically how data collectors are deployed to relay data from sensor nodes to the backbone network. To address this pitfall, in this paper we consider the scenario where data collectors are deployed at the centers, or parent points, of the clusters in PCP. Since the locations of data collectors and sensor nodes are correlated, the independence assumption adopted in most related work cannot be applied. We first derive the analytical expression of the Laplace transform of the aggregate interference at each data collector and then obtain the closed-form lower bound of the transmission success probability for each sensor node to transmit data to the nearby data collector. Numerical evaluation shows that the derived lower bound matches the simulation results very well. In addition, we have also shown that placing data collectors at cluster centers, while mathematically involved for analysis, can achieve significant performance gain compared to conventional scenarios where data collectors and sensor nodes are distributed independently without any coordination.