Print Friendly, PDF & Email



Source: IE


Context: The Department of Telecommunications (DoT) has launched the Artificial Intelligence and Facial Recognition powered Solution for Telecom SIM Subscriber Verification (ASTR) tool to detect and block fraudulent mobile connections.


About ASTR:

PurposeTo detect and block fraudulent mobile connections
OriginIn 2012, the DoT ordered telecom operators to share their subscriber database, including user pictures. These images form the basis for the facial recognition algorithm used in ASTR. The ASTR project was developed by the DoT’s unit in Haryana between April 2021 and July 2021.
Technology UsedArtificial Intelligence (AI) and Facial Recognition
FunctionalityAnalyzing subscriber databases to identify multiple connections associated with the same person
Example of its workingSuppose there is a suspected face associated with multiple mobile connections. ASTR will compare this face against a database of images to identify other connections linked to the same face. If there are more than nine connections against a single individual’s photograph or if the same person has taken SIMs under different names, ASTR will flag it as potentially fraudulent.
Connection LimitCurrently, a maximum of nine mobile connections can be taken using a single identity proof
Action TakenThe list of fraudulent connections will be shared with telecom operators, banks, payment wallets, and social media platforms
Description of technology used:
Image ProcessingFaces in subscriber images are encoded using convolutional neural network (CNN) models
Face ComparisonA face comparison is carried out for each face against all faces in the database to group similar faces under one directory
Matching AccuracyFaces are considered identical if they match to the extent of at least 97.5%
Name Matching“Fuzzy logic” is used to find similarity or approximate matches for subscriber names, accounting for typographical errors
What is fuzzy logic?Fuzzy logic is a mathematical approach that deals with uncertainty and imprecise information. It allows for the representation of vagueness and partial truth, unlike traditional binary logic.