Strategieën voor het aanpakken van kwetsbaarheden in embedded systemen

Internet of Things (IoT) devices are ubiquitous in consumer environments, yet their security is inadequate, exposing users and infrastructures to significant risk. The consumer IoT ecosystem is highly fragmented, with devices that frequently depend on diverse external libraries, communication protocols, and cloud services, which are often integrated without systematic security analysis. For example, real-time communication (RTC) technologies provide low-latency media transport in many IoT applications, but their implementations are frequently under-tested and complex. Threat modeling offers a principled way to reason about such complex systems and to identify security and privacy risks early in the design process. However, existing threat-modeling approaches are difficult to apply across the fragmented IoT architectures with diverse threats, and they are rarely validated against deployed systems. At the same time, economic pressures, limited awareness, and a lack of security expertise disincentivize vendors, particularly SMEs, from investing in IoT security. This dissertation investigates how resource-constrained, heterogeneous consumer IoT systems can be designed, analyzed, and operated to achieve and sustain a robust security posture. To tackle these challenges, this dissertation pursues four objectives: an in-depth empirical analysis of the security of consumer IoT devices; a focused investigation of RTC technologies, in particular WebRTC, to support systematic security testing of IoT devices; the development and validation of threat modelstailored to distinct consumer IoT architectures; and the design of an educational session to raise IoT developers' security awareness and skills. To address the first objective, we perform reverse engineering, network traffic analysis, and vulnerability assessments of real-world consumer IoT devices. These empirical analyses reveal that many devices exhibit severe security issues, including vulnerabilities that can be chained together, enabling lateral movements within an ecosystem and fully compromising the devices and ecosystem. For the second objective, this dissertation introduces RTCInspect, a novel automated tool for analyzing the security of RTC deployments. This tool addresses the lack of automated security solutions for RTC technologies, exposing insecure proprietary and flawed WebRTC implementations, while also supporting developers in testing and hardening their RTC communications. Building on these insights and tackling the third objective, this dissertation proposes three architecture-specific threat models derived from the analysis ofreal-world devices. We validate these models by mapping the modeled threats to vulnerabilities uncovered in practice, demonstrating their ability to predict which threats are likely to manifest as exploitable weaknesses. Finally, to address the fourth objective, this dissertation designs and delivers an educational session that has been delivered to 300 students and industry practitioners. This session provides hands-on training to integrate security into the IoT development lifecycle, thereby raising developers' awareness and capabilities. Together, these contributions advance the state of the art in IoT reverse engineering, security assessment, threat modeling, and developer education.