Skip to content

Master’s Thesis: Intelligent Failure Analysis of “Silent Calls” in Streaming Real-Time Systems

Teach a Distributed System to Recognise When a Data Stream Has Failed—and Restore It Safely

In real-time systems, it is not enough for a session to appear active—the underlying stream must actually work. Imagine that the control-plane signalling indicates that everything is normal while the audio, video, or sensor data has disappeared. How can the system determine whether this is a genuine outage, temporary network variation, or legitimate silence? In this master’s thesis project, you will develop intelligent failure analysis and self-healing capabilities for streaming data, using voice calls as the practical application domain.

The Technical Challenge

  • Detect anomalies in continuous media streams with extremely low latency
  • Distinguish genuine failures from normal variation, silence, and codec-specific behaviour
  • Model relationships between multiple parallel streams and interconnected sessions
  • Make safe recovery decisions in an asynchronous, event-driven system

What You Can Explore and Build

  • A real-time engine that analyses RTP statistics, traffic patterns, and events from multiple nodes
  • Rule-based, statistical, and AI-based methods for anomaly detection
  • A model for complex communications flows involving multiple call legs and dependencies
  • A coordinated recovery mechanism that avoids race conditions and incorrect remediation actions
  • A reproducible test environment for measuring accuracy, false-positive rates, detection latency, and resource utilisation

Why This Is an Exciting Master’s Thesis Project

The project brings together signal analysis, observability, algorithm design, and distributed coordination. You will work on a real problem in a production-grade platform, where milliseconds, concurrency, and incorrect assumptions have a major impact on quality and user experience.

Who Are We Looking For?

  • A master’s-level engineering student in computer engineering, electrical engineering, engineering physics, information technology, or a related field
  • A curiosity about real-time systems, distributed systems, streaming, networking, or intelligent failure analysis
  • An interest in Java or another modern systems development language
  • Strong analytical skills and an interest in experimenting with real-world data and hard-to-reproduce failures
  • You do not need to be a VoIP expert—SIP and RTP will be tools you learn to use during the project

Your assignment: Design, implement, and evaluate a prototype that detects when a real-time stream has failed even though the session still appears active—and that can initiate a safe recovery. The work will be carried out against a real communications platform and will give you the opportunity to advance the state of the art in self-healing real-time systems.

 

Apply here: