Risk-Based Decision Support Systems for Cybersecurity in Autonomous Vehicle Operations

Abstract

The Spatiotemporal Innovation Center at George Mason University is developing an autonomous airspace management concept, creating a new level of symbiotic analysis and decision-making that fuses high-frequency answer-rich data (command and control) with real-time low-frequency query-rich data. State-of-the-art safety-of-life functions use one model to control the vehicle and a different model to monitor the vehicle, without significant sharing of information between the two models, leading to sub-optimal performance. Our approach supports risk-based decision support systems to optimize vehicle operations by constructing executable queries that guide information retrieval from a network of capabilities that maximize mission benefit within resource constraints, enabling smart sharing of information and coordination for risk-aware intelligent data.

Autonomous vehicle (AV) operation generates and uses large volumes of data in real time to ensure safe operations. Wireless data transfer rates are constrained by the airwaves' available frequencies, transmission bandwidth, and transmission power, and the amount of useful information that can be transmitted in a unit of time. This spatial confinement creates gaps in real-time data transfer capabilities and available bandwidth, requiring automated algorithms to constrain AV operation to low-latency feedback communication to carry out safety-of-life functions, preventing accidents or ensuring the AV can carry out its primary functions, avoiding potential negative impacts. Consequently, many of the required analyses and decisions must be carried out on or by the AV, remotely or both. These design choices must meet industry practice standards.