November 14, 2018 Thesis Open Access

tewodros hailu

Citation Style Language JSON Export

{
  "DOI": "10.20372/nadre:4488", 
  "author": [
    {
      "family": "tewodros hailu"
    }
  ], 
  "issued": {
    "date-parts": [
      [
        2018, 
        11, 
        14
      ]
    ]
  }, 
  "abstract": "<p>Over-the-Top (OTT) bypass is a type of Interconnect Bypass fraud where regular<br>\nvoice calls are rerouted through OTT network and terminated as an OTT call. These<br>\ncalls are terminated using OTT applications which need user&rsquo;s Mobile Station International<br>\nSubscriber Directory Number (MSISDN) for authentication. Detecting<br>\nOTT voice call packets through different network traffic classification techniques is<br>\none subtask in the detection of this fraud.<br>\nIn this thesis, performance of three machine learning algorithms; Adaptive Booster<br>\n(AdaBoost) + J48, Repeated Incremental Pruning to Produce Error Reduction (RIPPER),<br>\nand Support Vector Machine (SVM) is evaluated in detecting MSISDN-based OTT<br>\npackets taking Viber, Tango, and Telegram as a sample. Detection of OTT traffic<br>\nand voice call packets from the OTT traffic have been treated separately as classification<br>\ntasks. Ten cross-fold and separate test data validation techniques together<br>\nwith 1.7 million labeled packets generated and captured in controlled laboratory<br>\nenvironment are used in the evaluation process.<br>\nAdaBoost + J48 achieved the best accuracy on both classification tasks compared to<br>\nthe others while using ten cross-fold validation. However, an accuracy of 48.4%<br>\nobtained in detecting voice call packets while using separate test data validation<br>\nmakes it less preferable in the classification task. Even if it takes longer time to<br>\ntrain SVM, it was the best performer (95.35% accurate) in detecting voice call packets<br>\nin separate test data validation. Considering accuracy attained by the algorithms<br>\nin separate test data validation technique together with the detection rate<br>\nof OTT voice call packets, SVM is preferable than the other two algorithms</p>", 
  "title": "Network Traffic Classification Using Machine Learning: A Step Towards Over-the-Top Bypass Fraud Detection", 
  "type": "thesis", 
  "id": "4488"
}