Location
MC 6460
Candidate
Zoya Abbasi | Applied Mathematics, University of Waterloo
Title
Consensus Investigation of Multilayer Multiagent Systems with Application
Abstract
The global spread of infectious diseases through travel and immigration significantly heightens morbidity and mortality rates, leading to major economic, social, and political upheaval. To mitigate these impacts and effectively manage disease control, it is critical to study the properties and behaviors of epidemic diseases. In this regard, researchers have extensively used network-based mathematical models and adapted Multi-Agent Systems (MASs) to the study of biological phenomena. This adaptation has given rise to a new type of epidemic modeling where cities implicated in disease outbreaks function as individual agents within a system.As the research has deepened, the complexity of actual systems has driven the adoption of more sophisticated frameworks, such as multilayer networks. These networks consist of multiple layers that each represent different types of interactions, allowing nodes with different dynamics to engage in a variety of exchanges across these layers. Nonetheless, the challenge of achieving consensus within these multilayer MASs received less attention.This ongoing research aims to establish a multi-layer framework to increase the system’s adaptability and resilience. A multi-layer multi-agent system will be introduced, characterized by different layers that each contain agents with distinct dynamics. This system will not only focus on managing the spread of diseases through agents but also consider the transmission through other means like water systems or contaminated surfaces, alongside an information layer that enhances communication between healthcare providers in different cities.This study will also tackle the consensus problem within this system. The goal will be directed at attaining a layer-based consensus, with agents having different dynamics in different layers reaching varied consensus points based on their specific dynamics and the tasks they face. It is more practical and realistic for agents across different layers to target layer-specific objectives rather than a common goal.Further developments in this research will apply a layer-based consensus protocol to a model featuring a switchable leadership strategy. This includes a dedicated layer for potential leaders who interact with other agents without direct commands or influence, and who can switch roles as necessary. This flexibility is crucial, as leaders may not always be capable of maintaining their roles due to unforeseen disruptions or damage that might compromise their ability to lead. Hence, there is an evident need for backup leaders.Ultimately, this research is expected to lead to more effective management of disease outbreaks and investigate layer-based consensus across the multiple layers of the multi-agent system, culminating in better disease control strategies and outcomes.