Telecom Operator leverages Big Data for Real-time Fraud Detection

About Client

The client is a leader in IP networking technology with a strong track record in developing and deploying next-generation carrier-grade Session Border Controllers (SBCs). The company focuses on enabling telecom operators to transition toward modern all-IP communication infrastructures.

Its SBC platforms are widely deployed by leading telecom operators across the globe to manage and secure voice traffic. As telecom networks handle massive volumes of calls and transactions every second, the need for built-in real-time fraud detection for telcos has become increasingly critical.

Background

Telecom network operators—who are the client’s primary customers—were experiencing significant financial losses due to fraudulent activities within voice traffic and network usage.

Existing systems lacked the infrastructure necessary to run advanced analytics and clustering algorithms capable of identifying suspicious behavior in real time. As telecom networks grew in scale, detecting fraud required processing massive volumes of call detail records (CDRs) and network data streams.

The client needed a scalable analytics framework capable of supporting advanced fraud pattern recognition while operating within the resource constraints of an SBC environment. Without such capabilities, telecom operators struggled to proactively detect fraudulent calls, SIM misuse, or abnormal traffic patterns.

Objective

The objective was to design a system capable of storing and analyzing large volumes of call data directly on a single SBC box while remaining scalable enough to retain more than a year of historical data.

The system also needed to analyze this data to generate meaningful fraud patterns and identify suspicious activity. To support operational teams, the platform had to deliver real-time fraud detection for telcos through dashboards and monitoring reports. Key technical challenges included:

  • Massive Data Volume and Velocity
    Telecom networks generated more than 10,000 calls per second, requiring high-throughput data processing.
  • Complex Application Data Formats
    Call records were stored in diverse formats including flat files, Google protocol buffers, and nested data structures.
  • Limited Infrastructure Resources
    The solution needed to run within the hardware limitations of the SBC box while supporting scalability through additional nodes.

Our Solution

To meet these requirements, a scalable big data platform was designed to support real-time analytics and fraud detection directly within the SBC infrastructure.

  • A big data platform was needed to fulfill the ambitious requirements of the client
  • Apache Spark with Parquet columnar storage was selected for compression
  • Elephant Bird and Java were used for Google Protobuf Processing
  • KVM Virtualization of SBC Server to run multiple nodes with redundancy
  • Real-time scaling with addition of physical nodes
  • Unsupervised clustering & supervised classification for pattern recognition
  • Statistical parameters such as average length of call, average number of calls per month and average delays in bill payment
  • Real time fraud detection and alerting

Outcome

  • The system enabled telecom operators to store and analyze one year of call data on the same SBC box, compared to only one month previously.
  • The client enhanced its product offering by integrating advanced reporting and analytics capabilities powered by the big data platform.
  • Fraud management capabilities were embedded directly into the SBC product, allowing telecom operators to detect predefined fraud patterns in real time.
  • Operators could now proactively monitor suspicious behavior and respond quickly to fraud attempts using advanced fraud pattern recognition analytics.
BizAcuity
Privacy Overview

This website uses cookies so that we can provide you with the best user experience possible. Cookie information is stored in your browser and performs functions such as recognising you when you return to our website and helping our team to understand which sections of the website you find most interesting and useful.