ERI: Enabling Faulted Agent Detection through Biologically Inspired Design

Embry-Riddle Aeronautical University, Daytona Beach FL

Investigators

Abstract

Modern systems often approach problems by connecting many smaller agents, rather than using a single, more expensive platform. For example, it is often advantageous to have a fleet of lower-cost unmanned aerial vehicles (UAVs) searching an area than a single, highly capable platform (airship). The nodes in these sophisticated networks, however, are tempting targets for bad actors. Although recognized as a vulnerability for multi-agent systems, current fault-detection methods have significant limitations. This Engineering Research Initiation (ERI) award supports research that examines the issue of faulted-agent identification through the lens of Biologically Inspired Design. The objective of this proposal is to design and evaluate a new biologically inspired approach based on ant colonies to identify faulted agents within a swarm, thus increasing resilience. Despite a lack of central control and low individual processing power, ant colonies have sophisticated approaches to identify intruders. The application of this biologically-inspired response to colony intruders will enhance the ability of swarms to identify faulted agents over current approaches. This approach is decentralized, scalable, only requires each agent to have local knowledge, and can be implemented without onboard machine learning. The research serves the interests of the United States in its potential to establish and demonstrate a framework to identify faulted agents without requiring centralized control; it is also scalable and does not require sophisticated agents such as on-board machine learning. These developments have the potential to provide a solution-neutral framework applicable to a variety of vulnerable systems. Increased resilience protects citizens, improves services, and reduces damage after faults; the approach investigated in this work may also help to protect combat drone swarms from infiltration, thereby enhancing our national defense. Finally, outreach activities will involve the local student veteran population, exposing this group to Systems Engineering research and enhancing the future STEM workforce. The goal of this research is to design and validate a faulted agent detection algorithm inspired by the use of Cuticular Hydrocarbon Profiles to identify intruders in ant colonies. The first of three initiatives to achieve this goal is to create a mathematical model of the Cuticular Hydrocarbon Faulted Agent Detection Algorithm (CHC FADA). A solution-neutral faulted agent identification approach will be explored through analysis and modeling of Cuticular Hydrocarbon (CHC) spread in ant colonies. Secondly, the effectiveness of the CHC FADA will be analyzed through a series of simulation-based tests that will examine both equipment failures and Byzantine faults. Third, CHC FAFA will be validated by performing real-world tests with a robotic swarm. Validation includes a student-focused cybersecurity challenge that targets high performing cybersecurity students, as well as veteran students. The outcome of this work is threefold. First, it will document a solution-neutral formulation of a eusocial insect-inspired fault detection approach. The significance of this outcome is a new approach that is decentralized, scalable, only requires each agent to have local knowledge, and can be implemented without onboard machine learning. Secondly, the work will provide measurement of the effectiveness of the new approach. A series of validation tests will be performed with both simulations and our physical test platform. Finally, this work seeks to accomplish initial integration of the local veteran population into Systems Engineering research through the real-world validation testing. This testing will leverage veterans’ unique understanding of multi-agent systems, resilience, and system thinking due to their operational experience. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.