Nagpur: With forest fires are the main causes of concern, Indore Institute of Technology (IIT) Indore working on the pilot project to develop a wireless sensor network (WSN) with the help of artificial intelligence tools (AI) to detect forest fires in Melghat Tiger Reserve (MTR) in Amravati District.
AI is a simulation of human intelligence processes with machines, especially computer systems.
WSN Network is a group of special sensors with communication infrastructure to monitor and record conditions in various locations.
This project is the idea of ​​Abhishek Srivastava, Associate Professor with the Department of Computer Science and Engineering.
He said, “Our team seeks to compile a sophisticated supervision system utilizing machine learning to more effectively detect forest fires.” “This approach involves the spread of WSN motes (remote devices) loaded with sensors to detect fires.
Forest fires are a big cause of destruction in the wild, often causing loss of flora and fauna, and forcing the transfer of locals,” Srivastava said.
It was during a relaxed conversation with a friend who worked for conservation in Melghat that the idea was evolved and Dr.
Srivastava knew how this fire was detected.
Maharashtra is susceptible to forest fires.
As reported by TOI on June 18, 2021, from more than 3.86 lakh fires issued nationally by the Indian Forest Survey (FSI) based on satellite data, Maharashtra recorded 60,851 warnings, which reached 16% of the total incidence in this country.
In accordance with FSI data, there are records of the number of fires between January and can be compared to the past two years.
Most of these fires were reported in April and may be man-made and turned on for the collection of leaves of the Mahua and Tendu.
“This task is completely dependent on manual efforts by forest officials and field staff who have been running for a long time, and in the absence of any connectivity, running back to notify an accident, spending a long time to extinguish the fire.
In addition, the rough field Melghat makes fighting Herculean’s duties , “said Srivastava.
“We work on WSN technology but the main problem with the spread of WSN as it is that they are often susceptible to false positives.
For example, the heat sensor of a mote can be stimulated by sunlight on a very hot summer day, and this can be misinterpreted as a fire, “Srivastava said.
“The moisture sensor might fake fire on a moist day.
To overcome such a scenario, the idea is to train the machine learning algorithm appropriately to minimize false positive incidents,” he said.
The norm is to spread the algorithm on the back-end cloud with unlimited resources and send signals from individual sensors for decision making.
Given the environmental hunger of dense forest energy such as melgat, however, do this is impractical.
Team, therefore, works on the development of a very ‘lean’ machine learning algorithm that can be used on MOTH WSN which is limited by resources, and is able to detect forest fires correctly.
The slim algorithm for classification, anomalous detection, and localization has been developed and deployed on the WSN pilot network in IIT, Indore, Campus.
The distribution and testing of this algorithm in the Matghat Tiger Reserve will be carried out as soon as the limitation of the pandemic is fully lifted.
“We will talk to forest officials before taking the project,” he said.
The project was designed by Dr.
Srivastava along with his collaborator at the University of Alberta, Dr.
Osmar Zaiane, and their team of students consisting of Arun Kumar, Ankit Jain, Prarthi Jain, Uttarsh Aggarwal, and Seemandhar Jain.