This project develops AI agents to autonomously generate actionable insights for campus network operation by analysing real-time network telemetry measured from the university campus, such as the usage of signalling packets like DNS, volumetric telemetry of subnets and hosts, and performance telemetry of video, live stream, gaming, conferencing, GenAI, and other services. For example, the AI agents will be capable to make reasoning and decision on actionable insights for network capacity planning by measuring the trends in popular video streaming services (Netflix, Disney+, Stan, Prime, etc.), Live Streaming (Twitch, sporting games), Online Gaming (CS:GO, CoD, Fortnite, etc.), conferencing (Zoom, Teams, etc.), and GenAI (ChatGPT, Github Copilot, etc.), in terms of viewing patterns, and how they change by time-of-day, day-of-week, week-of-term. etc, or the AI agents can make real-time detection on cyberattacks and suggest remedying actions from volumetric and signalling packet statistics.
School
Electrical Engineering and Telecommunications
Research Area
Telecommunications networks | Data analytics | AI
- Research environment
- Expected outcomes
- Supervisory team
- Reference material/links
This project will be carried out in a vibrant group that includes not just PhD and honours-thesis students at UNSW, but also commercial personnel from UNSW spin-out Canopus Networks that is building truly disruptive network traffic analytics platforms. You will get to play with live network traffic and develop AI agents for network operators, and your solutions will be tested and deployed in real operational networks.
Expected outcomes include: (a) analysis of network traffic characteristics in the UNSW campus for the use case you work on; (b) development of AI agents to extract actionable insights from network telemetry for network capacity planning, trouble shooting, or cybersecurity.
- The UNSW research team has written many research articles which can be found at Dr. Minzhao Lyu’s website