Optimized Dislocation of Mobile Sensor Networks on Large Marine Environments Using Voronoi Partitions

Alessandro Puntoni,Davide Moroni,Mario D’Acunto,Ovidio Salvetti

doi:10.3390/jmse8020132

Abstract

The real-time environmental surveillance of large areas requires the ability to dislocate sensor networks. Generally, the probability of the occurrence of a pollution event depends on the burden of possible sources operating in the areas to be monitored. This implies a challenge for devising optimal real-time dislocation of wireless sensor networks. This challenge involves both hardware solutions and algorithms optimizing the displacements of mobile sensor networks in large areas with a vast number of sources of pollutant factors based mainly on diffusion mechanisms. In this paper, we present theoretical and simulated results inherent to a Voronoi partition approach for the optimized dislocation of a set of mobile wireless sensors with circular (radial) sensing power on large areas. The optimal deployment was found to be a variation of the generalized centroidal Voronoi configuration, where the Voronoi configuration is event-driven, and the centroid set of the corresponding generalized Voronoi cells changes as a function of the pollution event. The initial localization of the pollution events is simulated with a Poisson distribution. Our results could improve the possibility of reducing the costs for real-time surveillance of large areas, and other environmental monitoring when wireless sensor networks are involved.

Highlights

A growing number of applications, such as spatial distribution mapping, dynamic sensors coverage, and environmental extensive area monitoring, have motivated the development of both sensing hardware and algorithms for target-oriented mobile sensor networks [1,2]
The optimal deployment was found to be a variation of the generalized centroidal Voronoi configuration, where the Voronoi configuration is event-driven, and the centroid set of the corresponding generalized Voronoi cells changes as a function of the pollution event
Let Q ⊂ d be a convex polytope, the space in which the sensors have to be deployed, and Vi ⊂ Q be the generalized Voronoi cell corresponding to the i-th node; we introduce a continuous density distribution function φ: Q → [0,1], where the density φ(q) is the probability of an event of interest occurring in q ∈ Q, and P = {p1, p2, . . . , pN}, pi ∈ Q is the configuration of N sensors [20]

Summary

Introduction

A growing number of applications, such as spatial distribution mapping, dynamic sensors coverage, and environmental extensive area monitoring, have motivated the development of both sensing hardware and algorithms for target-oriented mobile sensor networks [1,2]. Some parts might be prioritized based on prior knowledge and due to an inherent time-constraint that prohibits an exhaustive search of the area; for instance, in the case of emergencies and search and rescue operations This situation is met during the oil spill monitoring of large marine regions, and in general, in all pollution events involving diffusion [3]. In this scenario, mobile sensors should follow a certain physical velocity field to move from their initial configurations to the optimal one.

Modeling of Pollution Events

Basic Properties for a Voronoi Partition

Optimization in the Localization of Event-Dependent Mobile Sensors Network

Conclusions