Hyderabad: Thanks to an algorithm created by the International Institute for Information Technology, Hyderabad for Bharat Electronics Ltd., the Indian Air Force's target scopes and ground-based radars are set to get sharper and more precisely detect enemy and friendly aircraft.

When combined with the existing Air Force tracking system during research, the algorithm and software, which has been transitioned to BEL and is currently being tested out in their simulation environment, were found to improve object identification accuracy by radars by up to 96 per cent, from the existing 91 per cent.

Ground-based radars built by BEL detect and track flying objects in Indian airspace. The Indian Air Force's current tracking system employs a Multi-Sensor Tracking mechanism that utilises radars stationed around the country. The Air Situation Picture (ASP) is a thorough inventory of all aircraft in the airspace with their corresponding flight numbers and flight plans, compiled using information such as location and velocity coordinates acquired by each radar.

Also Read: Telangana's Mana Ooru Mana Badi Scheme Introduces BaLA Learning Concept

Due to overlapping radars sensing the same aircraft at times and delays in sensor communication, two common errors emerge the "merging" error, in which multiple aircraft in close proximity are incorrectly identified as one, and the "splitting" error, in which a single aircraft is sensed as multiple and erroneously flagged as a threat. As a result, the ASP generated is not always correct.

A team from BEL Ghaziabad approached IIITH to address these challenges and has been involved in many discussions about developing an automated solution.

With 11 days of anonymized and tagged data collected from 17 million data points captured by multiple radars, the IIITH research team led by Prof. Praveen Paruchuri of the Machine Learning Lab, IIITH, and a Masters student Anoop Dasika built a machine learning model.

While the original tracking system was 91% accurate, the IIITH team developed an AI-assisted tracker that improved the accuracy by 5%. According to the team, one of the benefits is that it aids in thorough radar analysis.

The research findings will be presented at the 34th Annual Conference on Innovative Applications of Artificial Intelligence later this month in a paper titled "CB NN Ensemble to Improve Tracking Accuracy in Air Surveillance."