View Proposal

---

Proposer

Inkar Zhumay

Title

On-Gateway Botnet Detection for Consumer IoT Networks

Goal

To determine whether IoT botnet activity can be detected in real time on a home gateway with constrained resources, by deploying anomaly detection at the network edge and measuring detection quality against CPU, memory and latency cost.

Description

For Aktobe Students Only. Build a detection service that runs on a gateway-class device, observes traffic from connected IoT devices, and flags compromise indicators — scanning behaviour, command-and-control beaconing, and outbound flooding. Train per-device behavioural models on benign traffic only, since each device type has a narrow and highly predictable traffic profile and deviations are informative. Evaluate against recorded botnet traffic covering distinct infection stages, and report detection latency separately for each stage: detecting a device that is already flooding is far less valuable than detecting one that is scanning. Measure the resource footprint on the gateway — CPU load, memory, and any added forwarding latency — and identify the throughput at which detection can no longer keep pace with traffic. Provide an interface showing which device was flagged, on what evidence, and with what confidence.

Resources

Background

Url

Difficulty Level

Moderate

Ethical Approval

None

Number Of Students

1

Supervisor

Inkar Zhumay

Keywords

botnet detection, internet of things (iot), network security, gateway-based detection, machine learning, cybersecurity, network traffic analysis, intrusion detection system (ids)

Degrees

Bachelor of Science in Computational Sciences and Software Engineering