Orchestrated Risk-Based Integrity Testing (ORBIT)

Abstract

Wearable healthcare devices are becoming an important part of modern IoT ecosystems, enabling continuous monitoring and personalized healthcare services. At the same time, their limited computational and energy resources make security a major challenge, especially in ensuring the integrity of embedded software. Traditional remote attestation techniques often rely on periodic full-memory verification, which can impose considerable overhead on low-power devices. To address this issue, this thesis proposes ORBIT, a policy-driven Verifier–Prover framework that dynamically adapts attestation behavior according to the runtime state of the device. The system is implemented on microcontroller-based IoT devices and supported by a server-side verifier. Runtime behavior is monitored through hardware performance counters, and lightweight machine learning models are used to detect abnormal activity and estimate workload intensity. Based on these observations, a policy engine adjusts attestation frequency and memory coverage, increasing verification rigor under suspicious conditions while reducing overhead during normal operation. In this way, the framework balances security, performance, and device availability. Evaluation under controlled attack scenarios shows that microarchitectural behavioral signals can effectively support adaptive and resource-aware integrity verification in embedded IoT systems.