TERM3

Term Overview

For Term 3, the project progressed from baseline system validation to evaluating the resilience of the behavioural trust framework under active protocol-level attacks. The core deliverables included deploying three ESP32 sensor nodes transmitting real-time environmental data through an MQTT broker hosted on the Rock Pi gateway, alongside an independent Python-based traffic capture and monitoring system with a Dash dashboard. The Behavioural Trust Index (BTI) was extended to evaluate message frequency, inter-arrival timing, sensor data stability, and payload-size consistency. Five protocol-compliant attack scenarios—replay, timing manipulation, false data injection, traffic flooding, and behaviour mimicking—were systematically evaluated across multiple experimental iterations. The results demonstrated that false data injection produced the strongest trust degradation, with BTI reductions of up to 33.4%, while timing manipulation caused reductions of up to 14.4%. Replay attacks proved significantly more difficult to detect, with only 0.8%–4.0% of malicious messages being flagged. Flooding attacks were detected rapidly but exposed a weakness in the current weighted trust model, where increased message frequency could artificially raise the overall BTI. Overall, Term 3 established that behavioural trust scoring can effectively identify certain forms of malicious IoT behaviour, while highlighting the need for improved replay detection, frequency-metric design, adaptive thresholds, and complementary security mechanisms.