Rasel Ahmed

Seeking a PhD position

Research

Machine-learning methods for detecting intrusions in IoT networks, where devices are too small and too numerous to protect one by one.

Current project

AI-driven intrusion detection for IoT networks

Master of IT research, City University Malaysia (in progress)

Degree
Master of IT
University
City University Malaysia
Focus
IoT intrusion detection
Status
In progress

The problem

Cameras, sensors, smart plugs and industrial controllers are now on almost every network, but most of them are too small to run antivirus or a host firewall, and there are too many to patch and watch one by one. Attackers use them to get into a network or to build botnets. That makes the network itself the best place to defend them.

The question

Can a machine-learning model learn what normal traffic looks like for each kind of device, and reliably flag intrusions, including attacks it has not seen before, without burying the people who run the network in false alarms?

Approach

  1. 1

    Data

    Labelled IoT network traffic covering normal device behaviour and common attack types.

  2. 2

    Features

    Packets grouped into flows and described by timing, size and protocol features a model can learn from.

  3. 3

    Models

    Classic machine-learning baselines compared with deep-learning and anomaly-detection models.

  4. 4

    Evaluation

    Precision, recall, F1 score and false-positive rate: the numbers that matter to a network team.

Progress

Network traffic, one IoT device Learned normal range Observed traffic Flagged anomaly
Outside the learned range: flagged 00:0006:0012:0018:0024:00
Figure 1. The idea in one picture: the model learns the normal range for a device (shaded) and flags traffic that falls outside it.

The project is in progress as part of my Master of IT. Results, the model comparison and the code will be published here when the thesis is complete. If you supervise in network security or applied machine learning and would like to see the proposal, please get in touch.

Publications

  1. NowAhmed, R. AI-driven intrusion detection for IoT networks. Master of IT thesis, City University Malaysia (in progress).

Related experience

At SS Rover Sdn Bhd (2024ā€“2025) I set up system monitoring with Node-RED and Grafana connected to PLCs, and worked on network design, server administration, local database design, an LMS and email security. It gave me hands-on experience with the kind of connected industrial devices my research aims to protect.

Since 2017 I have also built and hardened WordPress and Shopify sites for clients, so I see security from the side of the people who have to run the systems.

  • Python
  • Machine learning
  • Node-RED
  • Grafana
  • PLC integration
  • Linux
  • Docker
  • Network design

Supervisors and collaborators

Iā€™m happy to share my research proposal and progress. Email is the best way to reach me.

Email me