Area‐priority‐based sensor deployment optimisation with priority estimation using K‐means. Issue 7 (12th April 2017)

• Record Type: Journal Article
• Authors: Ateş, Enes; Kalayci, Tahir Emre; Uğur, Aybars
• Is Part Of: IET communications. Volume 11:Issue 7(2017)
• Page Start: 1082
• Page End: 1090
• Publication Date: 2017-04-12
• Subjects: wireless sensor networks -- pattern clustering -- simulated annealing -- sensor placement -- estimation theory -- queueing theory -- image processing; area‐priority‐based sensor deployment optimisation -- wireless sensor network -- WSN deployment -- total cost reduction -- sensor coverage maximisation -- priority estimation method -- environment satellite image -- K‐means algorithm -- priority queue‐based technique -- simulated annealing algorithm -- total covered area priority maximisation -- sensor gap minimisation -- Google Maps; Telecommunication systems -- Periodicals; Speech processing systems -- Periodicals; 621.38205
• DOI: 10.1049/iet-com.2016.1264
• Languages: English
• ISSNs: 1751-8628

Abstract

Abstract : Deployment in wireless sensor networks (WSN) addresses maximising the coverage of sensors and reducing the total cost of deployment. The area‐priority concept for WSN deployment that the authors contributed to the literature recently allows environments with regions that have different importance or priority levels. In this study, the authors propose the first priority‐estimation method for area‐priority‐based WSN deployments. First, a satellite image of the environment that will be used in the deployment of the sensors is clustered by a K ‐means algorithm using the colour features of the regions. In the sensor deployment phase, this cluster information is used to determine the priorities of the sensor coverage areas on positions of the image. Sensors are initially deployed quickly using a priority queue‐based technique. Then, a simulated annealing algorithm is used to maximise the total covered area priority and to minimise the gaps between the sensors. Various experiments are performed for different scenarios (land, sea, and forest) on images captured from Google Maps using different parameter values. The experiments confirm that the proposed approach performs well and outperforms the random deployment of sensors.