On integrating Kalman filters with partial resequencing for real-time tracking systems

Abstract

This thesis deals with the communication network performance of the Automatic Vehicle Location (AVL) distributed database system. The Automatic Vehicle Location system is used to real-time track the movement of vehicles, traveling along a large geographical area. The full control of the movement of these vehicles requires the use of many communication systems such as Global Positioning system, Packet switching network and distributed database systems including vehicle position databases. The main goal of this thesis is to minimize the update response time in a real-time fully replicated distributed database system with resequencing constraints. A new method is introduced to reduce the system response time. An adaptive strategy that uses a combination of partial resequencing buffers and Kalman filtering techniques is proposed. Firstly, partial ordering techniques are used to discard any late update messages and reduce the system response time. Secondly, Kalman filters are used to predict any missing update, improve the percentage of received packets and hence improve the reliability of the system.