Combining Operational and Spectrum Characteristics to Form a Risk Model for Positive Train Control Communications

Though there is a tremendous research effort towards maturing software defined radio (SDR) to cognitive radio technologies, it is narrowly focused on spectrum access or performance tuning. An alternative approach is to create a cognition cycle that facilitates risk management and improves the radio's security posture. In the instances where SDR technologies represent a critical infrastructure, this advancement would be novel and paramount. SDR technologies represent the future of radio infrastructures due to their ability to establish flexibility - the potential for users to adjust performance capability in their ever-changing operational settings. Yet with this transition to software, radios will face similar threats as current computer systems. Therefore it is necessary to begin formulating risk models that can assess this new risk environment. Though the threats are not known at this time, the risks they may present can be estimated and addressed. As mandated by the Rail Safety Improvement Act of 2008, trains are incorporating SDR technologies in their implementation of Positive Train Control (PTC) - a distributed system enhancing the train's communications with the command and control infrastructure, railway switches, and crossings to address the numerous safety considerations in rail operations. There are risk models for rail operations and PTC is designed to ensure high-risk environments are avoided or mitigated through adjustments in the operation of the train (assuming they receive the associated PTC status message). This dissertation introduces a risk engine that can be incorporated as a cognition cycle in PTC architectures currently under development. It provides awareness of heightened risk to the communication system availability based on environmental factors and, more importantly, adversarial attempts to exploit vulnerabilities in SDR technologies. To predict its performance in the intended operational environment, the model is simulated against train operations within a major urban environment and a regional expanse to demonstrate its ability to assess risk under varied conditions. To increase its relevance to the original intent of PTC - prevention of accidents - it is merged with an existing safety risk model in order to deliver risk assessments applicable to the situations where PTC is needed most. Findings are discussed to assess potential effectiveness as well as highlight challenges. A discussion of its relevance as a contribution to the research community is highlighted, specifically: risk estimates of planned rail operations, design and network planning of future PTC requirements, and initial efforts for adversary detection. The dissertation concludes with significant aspects of further research.