Excellence in Research: Actor-Based Modeling and Control of Distributed Networked Autonomous Systems with Fault-Tolerant Protocol Settings

A common vision of the future is the one where our surrounding environments are replete with smart distributed autonomous agents whose normal functioning significantly relies on tight integration of complex dynamic environments, making decisions and autonomously sensing their environments, and by doing so realizing a set of complex and often changing tasks for which these distributed autonomous agents are designed or expected to operate. The focus of this proposal is to present design and functionality analysis for autonomous agent-based systems, where agents dynamically process information received from their connected neighbors and make decisions or statistical inferences about their networked connections for reliable exchange or aggregation information across the whole distributed systems.

The findings from this project will have positive impacts on the current development and deployment of distributed autonomous agents, sensor networks and ubiquitous computing systems, and their interactions that need high confidence and fault-tolerant protocol settings. The project will also provide many opportunities and educational resources to STEM students that broaden their participation in engineering through algorithm development, testbed experimentation, performance evaluation and other project related activities.

This project aims at addressing fundamental challenges that arise from attempting to integrate a data-driven computational paradigm in settings, where distributed networked autonomous agents seek making strategic-decisions, inferences and/or establishing reliable network connections in complex uncertain environments. The basic idea is to expand efforts in modeling the structural behaviors of distributed autonomous agent-based systems, aimed at facilitating transmission of necessary information for decision-making capabilities or inferences based on fault-tolerant protocol settings.

The proposed work provides a framework for addressing the following questions: (i) how should autonomous agents dynamically revise their local connections based on the messaging patterns they receive from neighboring agents over a certain duration of time, that support global network formation with desirable structural properties? (ii) how is the resulting network formation process sensitive to noises in the communication channels or jamming activities due to adversaries? (iii) under what conditions may the network formation process never reach steady-state or how any adjustments in the messaging patterns or protocol settings can help address such issues? Overall, the proposed framework along with its theoretical algorithms and software tools that are being developed will facilitate the transition to real-world application problems.

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.