Home

Bio

Publications

Research

Courses

Media

Contact

Distributed Kalman Filtering for Range-Only Radio Networks


Sanjiban Choudhury

Overview


In this project, we design an algorithm to track an object in a radio sensor network in a robust and efficient way.

Course Details


Course Title: Math Fundamentals
Instructor: Michael Erdmann

Abstract


Advances in sensing technology and wireless communication are now headed towards designing large, interconnected, inexpensive senor grids with small sensors as its nodes. These sensors systems are aimed towards obserivng large areas with fine resolution. Specifically while addressing the problem of tracking a target there have been recent development in range-only sensing tech- nologies to utilize radio frequency signals to measure range between nodes. These range only radio nodes do away with line of sight measuring which grants a robustness of measurements against presence of obstacles such as smoke, dust or walls. Range measurements have inherent ambiguities as the state of the target (cartesian position) is not observable from one particular node and have a flip ambiguity when two nodes are there. None the less these sensors lend themselves very well with an inexpensive measurement grid, and hence it is necessary to develop estimation strategies which are specific for these naturally occurring nonlinear and multi-modal measurement distributions using the presence of many such connected nodes to advantage.
A problem with these sensor grids is collection of data from distibuted wireless sensor nodes to a single point for on-line data processing is infeasible. This is due to the inherent delay in packages (due to multi-hop transmissions) and limitation on communciation badnwidth due to energy consumption requirements. A typical example of this problem is in wireless ad-hoc sensor networks where information needs to be multi-hopped from one node to another using closer neighbors. Thus couple with the requirement of good sensors for tracking, there is a need for algorithms which do on-line estimation while reducing the communication load among all sensor nodes, being very robust to sensor node failures or replacements and packet losses, and being suitable for distributed control applications.

Details


Here is the final report.

Here is the final presentation.