Sensor Placement Algorithm With Range Constraints for Precision Agriculture

作     者：Visalini, K. Subathra, B. Srinivasan, Seshadhri Palmieri, Giovanni Bekiroglu, Korkut Thiyaku, S. 

作者机构：Kalasalingam Acad Res & Educ Anandnagar Ctr Res Automat Control Engn Int Res Ctr Krishnankoil 626126 Tamil Nadu India Berkeley Educ Alliance Res Singapore Singapore 138602 Singapore Univ Sannio I-82100 Benevento Italy SUNY Inst Technol Utica Rome Polytech Inst Utica NY 13502 USA Satyukt Analyt Private Ltd Bengaluru 560094 Karnataka India 

出 版 物：《IEEE AEROSPACE AND ELECTRONIC SYSTEMS MAGAZINE》 (IEEE航空航天与电子系统杂志)

年 卷 期：2019年第34卷第6期

页      面：4-15页

核心收录：

学科分类：0808[工学-电气工程] 08[工学] 0825[工学-航空宇航科学与技术] 

主　　题：Agriculture Robot sensing systems Wireless sensor networks Precision engineering Optimization Agricultural engineering 

摘      要：In precision agriculture (PA), fusing low-resolution remote sensing (RS) information with proximal sensors provided by wireless sensor networks (WSN) can help increase productivity, reduce cost, and optimize resources. However, sensor cost and their range emerge as a major bottleneck in proximal sensing. Moreover, the range is affected by weather and farming operations outages. Consequently, optimal sensor placement to reduce cost considering range constraints and having ability to fuse high-level low-resolution RS information has emerged as one of the key problems in WSNs for PA applications. Typically, optimal sensor placement is a combinatorial problem which is computationally intensive (NP-hard) and range constraints add another complex dimension to the problem. This paper proposes Sensor Placement Algorithm with range constraints (SPARC), which optimally places sensors considering their range. Furthermore, it has ability to fuse low-resolution RS information with proximal sensing of the sensors. We first show that the problem is computationally intensive and complex due to its combinatorial nature. Next, we propose a two-step approach wherein the optimal sensor placement problem is solved in the first step by using the optimization models. In the second step, the sensor locations are adjusted to address the range constraints such that the error covariance is reduced. Novelty of SPARC is the inclusion of range constraints, fusing the proximal measurements with low-resolution high-level data, and the solution technique that organizes the problems into two steps. As a result, sensor cost, physical constraints of farming practices, and maintenance cost can be reduced drastically without compromising on the monitoring capability. We also present prototype of the sensor node and illustrate the SPARC on a precision farming test-bed developed by the authors.