Research Article
Optimal Relay Node Placement and Trajectory Computation in Sensor Networks with Mobile Data Collector
@INPROCEEDINGS{10.1007/978-3-642-17994-5_27, author={Ataul Bari and Fangyun Luo and Will Froese and Arunita Jaekel}, title={Optimal Relay Node Placement and Trajectory Computation in Sensor Networks with Mobile Data Collector}, proceedings={Ad Hoc Networks. Second International Conference, ADHOCNETS 2010, Victoria, BC, Canada, August 18-20, 2010, Revised Selected Papers}, proceedings_a={ADHOCNETS}, year={2012}, month={5}, keywords={}, doi={10.1007/978-3-642-17994-5_27} }- Ataul Bari
Fangyun Luo
Will Froese
Arunita Jaekel
Year: 2012
Optimal Relay Node Placement and Trajectory Computation in Sensor Networks with Mobile Data Collector
ADHOCNETS
Springer
DOI: 10.1007/978-3-642-17994-5_27
Abstract
Most sensor network architectures typically assume that nodes are stationary after deployment. However, a number of recent papers have shown that the use of mobile nodes or mobile data collectors (MDC) can significantly improve the performance of a network. In this model, the network can be viewed as a three-tier architecture, where the lowest-tier consists of a set of sensor nodes. The middle-tier contains a number of higher powered relay nodes, each acting as a cluster head for a number of sensor nodes in the tier below, and one or more mobile data collector(s), constitute the upper-tier. For such hierarchical architectures, there are a number of important design problems such as determining the number of relay nodes that are needed and their locations, determining the appropriate buffer capacities in the relay nodes to ensure there is no data loss due to buffer overflow and calculating a suitable trajectory for each MDC. In this paper, we first propose an integrated integer linear program (ILP) formulation that calculates the optimal number and positions of the relay nodes in the middle-tier, along with the requisite buffer sizes. We then present an algorithm for calculating the trajectory of the MDC, based on the relay node locations and the load on each individual relay node, in a way that minimizes the maximum energy dissipation of the relay nodes. Experimental results demonstrate that our approach is feasible for networks with hundreds of sensor nodes.