Data Accuracy Estimation for Cluster with Spatially Correlated Data in Wireless Sensor Networks

TL;DR

This paper develops a data accuracy model for sensor clusters sensing point events in wireless sensor networks, identifying minimal sensor sets that maintain accuracy and reducing redundancy before data aggregation.

Contribution

Introduces the first data accuracy model for sensor clusters prior to data aggregation, optimizing sensor selection and reducing communication overhead.

Findings

A minimal sensor set achieves similar data accuracy as the maximal set.

Data accuracy decreases with increasing distance from the event.

Model applies to spatially correlated data for point events.

Abstract

Objective-The main purpose of this paper is to construct a data accuracy model for the maximal set of sensor nodes that sense a point event and forms a cluster with fully connected network between them. We determine the minimal set of sensor nodes that are sufficient to give approximately the same data accuracy achieve by the maximal set of sensor nodes. Design approach/Procedure-L set of sensor nodes are randomly deployed over a region Z. Since a point event S has occurred in the region Z, M maximal set of sensor nodes wake up and start sensing the point event. The set of M sensor nodes forms a cluster with fully connected network and remaining set of sensor nodes continue to be in sleep mode. One sensor node is elected randomly as a cluster head (CH) node which can estimate the data accuracy for the cluster before data aggregation and finally send the data to the sink node. Findings - Since we simulate the data accuracy for the cluster (M set of sensor nodes) at CH node, there exist P minimal set of sensor nodes which give approximately the same data accuracy level achieve by M set of sensor nodes .Moreover we find that as the distance from the point event to the number of sensor nodes increases, the data accuracy also get decreases. Design Limitation -This model is only applicable to estimate data accuracy for the point event where the sensed data are assumed to be spatially correlated with approximately same variations. Practical implementation-Detect point event e.g. fire in forest. Inventive/Novel idea - This is the first time that a data accuracy model is performed for the cluster before data aggregation at the CH node which can reduce data redundancy and communication overhead.