Artificial Intelligence

and its
Use-cases  in Telecom sector

By:- Naveen Jakhar, ITS

ADG (AI & DIU)
DoT HQ
Topics of Discussion:

Basics of AI - Present status of AI in Telecom Sector

Network Vs Subscriber View- QoE

AI + Drone Cell Tower Monitoring and EMF Audits

Customer Care Applications and Portals - Call Drop, QoS analysis

ASTR (अस्त्र) and its impact

 Traditional Programming vs ML, DL &AI

Traditional ML, DL and

Programming AI

Data Data
Computer Output
Computer Program
Program Output
 Why Artificial Intelligence?

 Present day and Future Considerations –Data Generation and its 5 V’s -
Velocity, Volume, Value, Variety and Veracity – beyond human brain’s
processing powers.
 Technological advances – more data processing powers and storage
 Will provide capabilities beyond problem solving
 AI in telecom sector-Present situation

•Summary Report -76 TSPs, operating in 51 countries.

•18% have adopted AI for their products and services.
 AI in telecom sector – Driving factors

Major Driving Force for AI in your organization?

For selling new

products to existing Reducing OpEx by
Customers using automation

For delivering betters

Customer Experience
 Network Versus vs Customer View: QoE

• How does TSP do Network Management

How TSPs • KPIs Statistics
• Dashboard

Manage ? • Alarm identification – equipment's

monitoring The big question
is: how to make
customer QoS
experience an
• It takes me around 10s to open a website integral part of
page.
NMS?
Customers’ •
•
I cannot receive my email
It takes a long time to connect to customer

View
care executive.
• My phone is showing the network but I
cannot access internet.
 AI+Drone enabled Cell Tower Monitoring

• Verizon and AT&T - for monitoring their cell sites

• Tower inspection algorithms for checking assets - 6,000 videos, > 1 million cell
tower photos as datasets – 200 elements identified
• Telecom tower sites located in remote areas are usually left unattended.
• Fuel and expensive telecom equipment
on these sites - susceptible to theft
and fiddling.
• Drones provide a bird’s view eye
• Can be very effective during
disaster management.
• Helpful for inspecting surroundings
AI for EMF emissions audits and public awareness
of EMF emission

Lat/long of mobile towers- available on Tarang Sanchar portal.

AI + Drone based monitoring of mobile towers - lat/long, radius details to

drone

EMF emission data of telecom towers may be collected with these AI

enabled drones

Like Air Quality Index - Safe, Critical, Bad, hazardous to eradicate fear of
radiations levels among Public - Real time data may be fed to LED displays

Data pushed to a centralized server for mapping emission levels to different

categories for EMF Compliance purposes
 Present setup for grievances redressal

Access •Contact to any TSP’s Customer Care is – through Call or email or SMS

Time Delay
•Time taken by any subscriber to reach to customer care executive after wading through IVRS is more than 60s
•Unnecessary wastage of resources, public telecom network
•Poor QoS experienced by subscriber even during lodging complaints/grievances

Chatbot • Some TSPs websites have implemented interactive chatbots

• TSP downloads its APP on user mobile device
 AI enabled Chatbots & self-assessment APPs
TSP App may have an inbuilt
button of Health Check wherein
following parameters will be Datasets of all subscribers:
automatically checked for •To be used for finding subscriber
identifying issue: receiving worst service;
i. Type of network and technology • A benchmark may be set as a
ii. Signal strength threshold value.
iii. Data speed – uplink and downlink • If that criteria is not fulfilled, the App
iv. Location
will make an automatic call to call
centre where the grievance may be
directly routed to the customer care
executive.

The datasets may be used for machine learning for taking better decisions.
 AI for Disaster Management

IMD issues warnings and alerts for natural hazards & disasters

Common Alert Protocol (CAP), an indigenous application designed

by C-DOT, is being used for sending alert SMSs to the subscribers
present in affected areas of expected disaster

Number of people dialing the helpline codes increases exponentially – results in network
congestion

CAP and EWS + Location Based System (CQAS) - extract information about approx. count of
people in area - evacuation
ASTR (अस्त्र ) : AI and Facial Recognition powered Solution
for Telecom SIM Subscriber VeRification
NextGen Digital Intelligence Tool
for
Detecting and
weeding out Prevention Catalyzing
fake/forged SIMs of cyber the
before used for crimes and investigation
cyber crime and frauds of cyber
frauds crimes

By:- Naveen Jakhar, ITS

ADG (AI & DIU)
DoT HQ
The Menace of Cyber Frauds and Crimes
 Fake SIMs and Problem of Cyber Frauds, Crimes

50000 Cyber-crimes India lost Rs 1.25 Lakh

cases in 2020, 30142 cases of
financial frauds ~ 60% crores due to cyber crimes
52974 cases in 2021 in last 3 years.

> 11 lakh complaints on > 2.5 lakh SIMs Pendency of cases > 70%
NCRP in last 3 years involved in cyber crimes by end of 2020

Huge Vulnerable Attack Major cyber crime Direct threat to national

Surface of 1.16 billion syndicates - from security, public law & order
telecom subscribers Jamtara and Mewat. and safety

Fake SIMs - Direct Threat to Digital Economy and Digital Inclusion

 Nexus b/w Cyber Crimes and Fake SIMs
Fake SIMs in financial frauds,
cyber crimes, extortion, dacoity,
kidnapping, hoax, threat calls,
advanced crimes like sextortion

Why cyber crime

investigation is
complex ?
• Anonymity, Untraceability,
Trans border syndicates
•70% cases pending
investigation
• <5% conviction rate
Cyber crime hotspots in India

Mewat - Haryana
Jamtara -Bihar Jharkhand
Jammu Kashmir
North Eastern States
Rajasthan
West Bengal & Kolkata
Odisha
Assam
Punjab
Delhi NCR
Gujarat
Uttar Pradesh
 SIM Subscription Process in India
Paper Digital -KYC E-KYC using
Aadhaar
Live Image

Live Image
Live Photo,
KYC – Name,
DOB, PoI/PoA,

Photo,
KYC – Name,
DOB, PoI/PoA,
Tele-verification

SIMs – 18 crore, SIMs – 63 crore, SIMs – 53 crore,

13.5% of total 47% of total 39.5% of total

Total SIMs in India as on 31.12.2022 – 134 crores

 Existing SIM Subscriber Verification - Limitations

No centralized SIM Database, TSPs - individual database

> 30 PoI/PoA documents – for connection

No online mechanism for PoI/PoA verification for SIM activation

Text Based Analysis by LEAs and DoT LSA field units

Creation of fake/forged PoI and PoA by fraudsters

0.1% and 0.2% sample CAFs audit by DoT LSA units

 Existing SIM Subscriber Verification - Limitations

LEA reports to DoT carries out

DoT CAF compliance Mobile
Mob Num for this number. Number
involved in Details on CAF Found
cyber crime - match with Compliant
xxxxxx9380 PoI/PoA used

Expertise level of the fraudsters in creation of fake Proof of Identity/Address documents

Can never be detected by human beings by analyzing single case in isolation

Consequence: Mostly SIMs found compliant in conventional audit/investigation for individual

case –
so even after being reported by citizens / LEAs, cyber crime gangs and their SIMs remain
functional
 PoI - VALID AADHAAR DOCUMENT
KYC DETAILS: NAME, DOB, ADDRESS

PHOTOGRAPH AADHAAR NO.

 ASTR: NextGen Digital Intelligence Tool

Live Image
Live Image

68 facial landmarks 68 facial landmarks

detected Cosine product
detected
comparison, face
matches at
99.25%

Name - Atul Name Similarity Check, matches Name - Raju

38% - different Names
Fake PoI/PoA- Fake PoI/PoA-
Aadhaar Aadhaar
Fraudulent subscribers –
SAME Face but different
names other details
 ASTR: NextGen Digital Intelligence Tool

Naveen
Navien
Face comparison with very high grade Norman
accuracy Nakeem

Quantum jump from 0.1% to 100% SIM

verification – 1000 times Fraudulent subscribers – Individual
SAME Face but different Subscribers >
ASTR – agnostic to PoI/PoA names other details prescribed count
Presenting the Power of ASTR: 7 SIMs, 3 Names
Male and Female – One Person in Mewat, Haryana

Sample Case -1
 ASTR: Impact in Cybercrime hotspots
Sample Case -2

213 SIMs across

different TSPs
100 Names
100 Forged PoI/PoA

ASTR detects Fraud

ONE PERSON
 ASTR: Impact in
Cybercrime hotspots
Sample Case -3

52 SIMs, 29 Names

Male and Female

29 Forged PoI/PoA

ASTR detects Fraud

- One Person
 FACE RECOGNITION BASED SIM
ANALYSIS RESULTS
Total Number of suspected fake/forged SIMs detected
Total Number of more than prescribed SIMs disconnected
10000000

2655735

NUMBER OF CONNECTIONS
1000000

100000

15748

10000
1
ALL TSPs
 ASTR – for Breaking the Chain
SIMs on
fake/forged
of Cyber Crimes and Frauds Docs or in
3rd parties
Name
Un
Bank
Traceabili accounts,
A tool to curb cyber ty and social media
Anonymit handles-
crimes/frauds y in SIMs
Whatsapp,
Facebook,
Breaking the
during Telegram
investigati Chain Payment
Pro active Analysis and APPs
on Cyber
Communication Crimes/Frauds
Intelligence using ASTR Cycle
-
Vishing,
Detecting and weeding Instant Phishing,
out the fraudulent SIMs Money SMiShing
Withdrawal Attack for
even before they are by Fraudsters Financial
used to carry out any Fraud
cyber crime/fraud
 AI for Subscriber On-boarding

For checking whether the Point of Sale are working as per instructions

Sudden rise in SIM sale in an area, Ghost PoS

Inputs to the Activation Officer before SIM Activation

In Subscriber on-boarding APP – live human face detection, de-

duplication of subscriber using KYC details and image
“I believe this artificial intelligence is going to be our partner. If we
misuse it, it will be a risk. If we use it right, it can be our partner.”
--Mashayoshi Son

Thank You !!
Your queries please.