Abstract
This paper presents a case study that demonstrates how tools from compositional verification can be used to design and analyze complex multi-agent systems operating in dynamic and uncertain environments. The case study concerns the design of an unmanned multi-aircraft system tasked to compromise an aerial encroacher by deploying countermeasures. The constituent agents, termed defenders, are fixed-wing unmanned aircraft. To successfully compromise the encroacher, at least one defender must be within a prespecified distance from the encroacher for a certain period, and the defenders must avoid collision among themselves and with the encroacher. Verifying this global property using monolithic (system-level) verification techniques is a challenging task due to the complexity of the components (defenders) and the interactions among them. To overcome these challenges, the components are designed to have a modular architecture, thereby enabling the use of component-based reasoning to simplify the task of verifying the global system property. Results from Euclidean geometry and formal methods are used to prove most component properties. For properties where analytical tools are overly conservative, focused Monte Carlo simulations are carried out. Restricting the use of simulations (or testing) to local verification of partial component properties leads to increasing the reliability of the system.
Highlights
A fundamental challenge in deploying a multi-agent cyberphysical system (CPS), such as a network of unmanned aircraft systems (UAS), in a safety-critical application is expressed by the question: How can one assess and ensure the reliability of such a complex system? This question becomes all the more challenging when considering systems that are required to execute missions in an uncertain environment [1] while being subject to threats that could be internal as well as external
The analyst is faced with the challenge of evaluating the system for all possible inputs and operational scenarios, which can be prohibitive for a multi-agent CPS composed of highly-interacting components
The outline of the paper is as follows: Section II gives a brief overview of the compositional approach used in this work; Section III provides a short description of the temporal logic of actions (TLA) formal language and the UAS equations of motion; Section IV presents the case study and demonstrates how component-based reasoning can be used to verify the global system property; and Section V gives some concluding remarks
Summary
A fundamental challenge in deploying a multi-agent cyberphysical system (CPS), such as a network of unmanned aircraft systems (UAS), in a safety-critical application is expressed by the question: How can one assess and ensure the reliability of such a complex system? This question becomes all the more challenging when considering systems that are required to execute missions in an uncertain environment [1] while being subject to threats that could be internal (e.g., a component failure) as well as external (e.g., environmental hazards). The problem of designing an unmanned multi-aircraft system and verifying that it satisfies a global system property bears a number of challenges, and the hybrid approach adopted in this work is an attempt towards making the verification process rigorous, which would thereby lead to increased reliability of the system. The outline of the paper is as follows: Section II gives a brief overview of the compositional approach used in this work; Section III provides a short description of the TLA formal language and the UAS equations of motion; Section IV presents the case study and demonstrates how component-based reasoning can be used to verify the global system property; and Section V gives some concluding remarks
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
