Machine learning and Kalman Filter based Target Tracking in Wireless Sensor Network

Abstract

Target tracking is one of the nontrivial application of wireless sensor network(WSN) which can be deploy corresponds to applications like monitoring, surveillance, indoor buildings and its application can be extended to health traffic and many other consumers and industrial areas. We summarize a framework for collaborative signal processing (CSP) in distributed sensor network. The thoughts are displayed with regards to tracking a different moving object in a sensor field. The key steps included in the following technique incorporate event detection, target classification, estimation and prediction of the target area.
Here we portray a unique way of following an object in WSNs. The projected strategy consolidates machine learning (ML) algorithm with a Kalman filter to determine the instantaneous location of an object in motion. The object’s increasing velocities, alongside data from the network, are utilized to get an exact approximation of its position. To execute this, radio fingerprints of received signal strength indicator (RSSIs) are first gathered over our region of interest.
The acquired information is then utilized with machine learning algorithm to process a function that gauges the location of the object using just RSSI data. The kernel based ridge regression (RR) and the vector output regularized least squares (vo-RLS) are used as a part of the learning algorithm. The Kalman filter is then utilized to combine first approximation of the target positions along with acceleration information for obtaining better accuracy. The outcome of this method is considered for various scenario and alikeness with a well-known techniques also provided.