Investigation of Sensor Data Fusions using Continuous-Time Decentralized Kalman Filters (DKF)

Author Name(s): Tushar, Dr. Abhijit Nayak
Author Email: tusharvardhan@yahoo.in

Abstract

The integration of sensors deals with the accumulation of information from two or more sensors in network. However the accumulation of data is typically the subject of research in the area of image analysis. Generally the reliability of collected data might induce the problem of reconstructing to get its original form. The Various fusion technique has already been implement for text data, image data and video data. The general exploration in the sensor data fusion is to understand the behavioral pattern reformation of real time capturing with sense of accuracy. This also handle out the problems of localizing, navigation and tracing problem. This paper is to study of Kalman Filters technique with their key feature of State Vector Fusion, Measurement Fusion, Gain Fusion. The mathematical formulation for the data fusion of sensors derived through the State Vector Fusion (SVF) and covariance time propagations. The present investigation is to analysis of process of dynamically maintaining a model of the local external environment and deep exploration of fusion technique using time decentralized Kalman Filters (DKF). Fusion of perceptual information is at the heart of data fusion process.

Introduction

Fusion of sensor data is expressed as the method of accumulating the gathered data from distributed sources of nodes (DS of Sensors ) to deliver the highly accurate and complete unified data about an process, Process activities, or process terminology. SDT is a combination of many traditional disciplines sensors and sensor head (Sink Node). Figure 1 shows an overview of data integration from multiple sensors. The estimated value from a single sensor point which modeled to make original view cannot be sufficient. It need to have various data accumulation from various distributed sensors. It might possible that the deployed nodes not covers the proper area, one cover the whole while other at less area as they assigned. Fusion technique is one of the reliable, accurate and real time based technique for sensor Fusion. This technique has low error in their fusion recovery data at accumulation time. In this illustration four stages has been shown below. Sensor node connected with filter and accumulate the gathered data at one place represent as data association block and finally come into fusion block with suitable algorithms used as per data better reconstruction form. Systematic analysis of gather data are presented and model obtained are descriptively reported.

Conclusion

The sensor data fusion algorithm introduced in this paper allows to combine the collected data of different sensors over time. Kalman filtering is considered at the same time as data compilation machines in local systems. This structure provides the flexibility to reconfigure the control system. New subsystems can be easily added without having to redesign the entire system. The system does not need a central processor, so if some local subsystems (each with local processor, sensors, and triggers) fail, the public system will continue to work.

780 total views, no views today

Download PDF File

About the author: admin