Scholarlink

<p>Voice over Internet Protocol (VoIP) security has advanced in detecting attack patterns and signaling behaviors, but most efforts focus on identifying malicious traffic rather than avoiding misclassification of legitimate calls. Traditional voice activity detection (VAD) systems rely on fixed thresholds to distinguish speech from silence, often misinterpreting natural pauses as threats, leading to high false positives and disruptions in essential services such as business, finance, and healthcare. VAD-based methods require constant threshold adjustments, which makes them unreliable in real-world scenarios. This study proposes an adaptive VoIP call classification model using a hidden Markov model (HMM) to address these limitations. The HMM-based model improves accuracy, reduces false positives, and adapts dynamically to changing call conditions without the need for manual tuning. Unlike VAD, it effectively handles call anomalies and adjusts to silence sensitivity in real time. By improving the accuracy and adaptability of threat detection, this approach improves VoIP security and ensures uninterrupted communication.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Misclassification in voice over internet protocol honeypots: distinguishing genuine calls from malicious traffic

  • 期刊论文

  • 发表日期：

    2025年9月25日

查看全文
  • Muhammad Muneeb,
  • Ayesha Altaf,
  • Faiza Iqbal,
  • Ayesha Afzaal,
  • Nagwan Abdel Samee,
  • Imran Ashraf

摘要

Voice over Internet Protocol (VoIP) security has advanced in detecting attack patterns and signaling behaviors, but most efforts focus on identifying malicious traffic rather than avoiding misclassification of legitimate calls. Traditional voice activity detection (VAD) systems rely on fixed thresholds to distinguish speech from silence, often misinterpreting natural pauses as threats, leading to high false positives and disruptions in essential services such as business, finance, and healthcare. VAD-based methods require constant threshold adjustments, which makes them unreliable in real-world scenarios. This study proposes an adaptive VoIP call classification model using a hidden Markov model (HMM) to address these limitations. The HMM-based model improves accuracy, reduces false positives, and adapts dynamically to changing call conditions without the need for manual tuning. Unlike VAD, it effectively handles call anomalies and adjusts to silence sensitivity in real time. By improving the accuracy and adaptability of threat detection, this approach improves VoIP security and ensures uninterrupted communication.