Towards reliable logging in the internet of things networks

The internet of things is one of the most rapidly developing technologies, and its low cost and usability make it applicable to various critical disciplines. Being a component of such critical infrastructure needs, these networks have to be dependable and offer the best outcome. Keeping track of network events is one method for enhancing network reliability, as network event logging supports essential processes such as debugging, checkpointing, auditing, root-cause analysis, and forensics. However, logging in the IoT networks is not a simple task. IoT devices are positioned in remote places with unstable connectivity and inadequate security protocols, making them vulnerable to environmental flaws and security breaches. This thesis investigates the problem of reliable logging in IoT networks. We concentrate on the problem in the presence of Byzantine behaviour and the integration of logging middleware into the network stack. To overcome these concerns, we propose a technique for distributed logging by distributing loggers around the network. We define the logger selection problem and the collection problem, and show that only the probabilistic weak variant can solve the problem. We examine the performance of the Collector algorithm in several MAC setups. We then explore the auditability notion in IoT; we show how safety specification can be enforced through the analogies of fair exchange. Next, we review our findings and their place in the existing body of knowledge. We also explore the limits we faced when investigating this problem, and we finish this thesis by providing opportunities for future work