Optimizing Area Coverage in Wireless Sensor Networks Ne j la Essaddi l , Mohamed Hamdi 1 , Sami Habib 2 , Noureddine Boudriga 1 Communication Networks and Security Research Lab., Tunisia Kuwait University, Kuwait Absract-Wireless Sensor Networks (WSNs) have inspired tremendous research interest in since the mid-1990s. Advance ment in wireless communication and miniature electromechanical systems (MEMSs) have enabled the development of low-cost, low power, multi-funcional, tiny sensor nodes that can sense the environment, perform data processing, and communicate with each other untethered over short distances. Most of the applications deployed over WSNs require strong coverage requirements, especially those related to the detection and tracking of distributed events. Moreover, these events are forwarded to the analysis center(s) through a set of sink nodes that locally gather data emanating from the elementary sensors. This paper proposes a coverage control scheme that adapts o the situation where multiple sink nodes are deployed within the monitored area. On the opposite to traditional coverage ap proaches that aim at guaranteeing a uniform density distribution, we place the sensor nodes in a manner that increases the coverage degree according to their proximity to a sink node. To reduce the complexiy of the optimization process, we consider a discrete search space by structuring the monitored into a uniform grid. An evolutionary algorithm is then nsed to choose whether to activate or not sensor nodes within every cell of the grid. We conducted a set of simulations in order to evaluate the performance of the proposed strategy, mainly in ensuring multi-target tracking. I. INTRODUCTION Wireless Sensor Networks (WSNs) pose a panoply of chal lenging problems related to localization, deployment, sleep scheduling, etc. The importance of these problems strongly depend on he nature of the application for which the WSN is deployed. A typical wireless sensor network consists of thousands of sensor nodes, deployed either randomly or ac cording to some predeined statistical distribution, over a geographic region of interest. A sensor node by itself has severe resource constraints, such as low battery power, limited signal processing, limited computation and communication capabilities, and a small amount of memory; hence it can sense only a limited portion of the environment. However, when a group of sensor nodes collaborate with each other, they can accomplish a much bigger task eiciently. One of the primary advantages of deploying a wireless sensor network is its low deployment cost and freedom from requiring a wired communication backbone, which is oten unfeasible or economically inconvenient. In the speciic case of event racking applications, coverage is the most prominent aspect since it has a direct incidence on the quality of surveillance provided by the network. The obvious reason behind this assertion is that we cannot detect an event if this latter occurs in a region which is not covered by the sensor network. Many approaches have been proposed in the literature to cope with coverage control. Various problems have been tackled including the following [ 1]: 1) Blanket coverage: to achieve a static rrangement of sensor nodes that maximizes the detection rate of targets appearing in the sensing ield 2) Barrier coverage: to achieve a static arrangement of sen sor nodes that minimizes the probability of undetected penetration through the barrier 3) Sweep coverage: to move a number of sensor nodes across a sensing ield, such that it addresses a speciied balance between maximizing the detection rate and minimizing the number of missed detections per unit area In this paper, we will focus mainly on the blanket coverage, where the objective is to deploy sensor nodes in strategic ways such that an optimal area coverage is achieved according to the needs of the underlying applications. Here, it is worth mentioning that the problem of rea coverage is related to the traditional art gallery problem (AGP) [2] in computational geometry. The AGP seeks to determine the minimum number of cameras that can be placed in a polygonal environment, such that every point in the environment is monitored. Similarly, the coverage problem basically deals with placing a minimum number of nodes, such that every point in the sensing ield is optimally covered under the aforementioned resource con straints, presence of obstacles, noise and varying topography. In the literature, many approaches have been devised to cope with coverage optimisation problems [?]. Our work extends the coverage control technique proposed by S. Habib in [ 1 1], which relies on dividing the region of interest into a regulr grid and model the sensor activation process as a genetic problem. Despite its eficiency, this approach does not adapt to the target tracking context and does not consider the communication overhead as an input parameter for the communication overhead. To address these limitations, we consider a uniform grid subdivision of the monitored zone and deploy a sensor activation strategy based on this grid structure. An evolutionary algorithm is considered to optimize the activation process of the sensor nodes according to their proximity to the closest sink node. The contribution of this paper is four-fold: 1) A uniform grid model-based activation strategy is pre sented in order to model the requirements for a multi event racking application