A Framework for Near Real-Time SIMbox Fraud Detection: the Case of Ethio Telecom
No Thumbnail Available
Date
2023-12
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
Telecom fraud is a major concern for telecom operators as well as for governments especially in Africa and Asia. Bypass fraud is one of the most fertile and costly frauds in today’s mobile industry making mobile operators and telecom regulators face a staggering annual revenue losses due to these fixed/VoIP to GSM/CDMA/Fixed line gateway equipment’s, which are used to terminate international inbound calls to local calls to local Subscribers by deviating traffic away from the legal interconnect gateways.
Bypass fraud is more rampant in the countries where the cost of terminating international call exceeds the cost of a national call by a considerable margin or the countries where government carriers monopolize international gateways. Fraudsters through the use of different bypass mechanisms, sell capacity to terminate calls cheaply in these countries, on the open market or through direct connections with interconnect operators. Operators sending outbound international traffic are then attracted by these interconnect operators with lower interconnect rates. This leads to lost revenue for terminating network operators.
While several attempts have been made to fight against Bypass Frauds the common approaches have been the use of monitoring calling patterns and profiles through Fraud Management Systems and the use of Test Call Generators. Both approaches have their own set of limitations coping up on frequently changing fraudster’s techniques and have short shelf life. In addition, those approaches took couple of months to detect a single fraudulent service number.
The general approach used to perform this research is a design science research methodology.
The proposed system works on processing the near real-time data using Spark Streaming. The objective is to build features to process the near real time data with Spark Streaming to reduce the workload on the node(s), achieve low latency to provide a better execution plan for a scalable and fault-tolerant processing of data. The Proposed framework targeted to improve response time and to give real-time solution to real time problem.
Domain experts made evaluation In order to assure the proposed system has met the requirement needed. In this work, we intended to create a fraud detection framework, which detects frauds based on big data technology, precisely Apache spark and using its Machine learning libraries in order to minimize the latency and to process transactions in a real-time.
Description
Keywords
Simbox Fraud, Big Data, Stream Processing, Real-Time Data Processing, Spark