Enhancing Over-the-Top (OTT) Bypass Fraud Detection and Classification in Telecom Networks Using Machine Learning Algorithms

Abstract

Telecom fraud has been a significant challenge for operators and organizations worldwide, with fraudsters leveraging technological advancements to perpetrate these crimes. Interconnect bypass fraud, particularly concerning over-the-top (OTT) services, has been one of the most prevalent and damaging forms of fraud, enabling fraudsters to avoid access fees and profit from international calls. The dynamic nature of this fraud has allowed it to circumvent traditional methods like Test Call Generators (TCG) and Fraud Management Systems (FMS), necessitating more sophisticated detection approaches. Our research utilized machine learning methods to improve the detection of OTT bypass fraud. We assessed the performance of three models: Random Forest (RF), Support Vector Machine (SVM), and Logistic Regression (LR). The models were tested using two validation approaches: an 80/20 split for training and testing, as well as 10-fold cross-validation. The experiments were conducted with Call Detail Record (CDR) data from Ethio-Telecom, following feature selection and preprocessing steps like data cleaning, integration, and aggregation. The findings showed that the Random Forest model delivered the highest accuracy across both validation methods. Specifically, RF reached 99.98%, and 99.33% 99.79% accuracy with the training and testing approach, outperforming SVM and LR, which attained 99.46% and 97.85% accuracy, respectively. By incorporating new features and leveraging machine learning, our approach significantly improved fraud detection capabilities, addressing the challenges associated with large datasets and the evolving nature of fraud patterns. Detecting and classifying OTT bypass fraud through these methods offers numerous advantages for telecom companies, including mitigating financial risks, enhancing revenue assurance, improving operational efficiency, safeguarding customer satisfaction, protecting brand reputation, ensuring regulatory compliance, and establishing industry leadership.