Distributed Consensus Algorithms in Sensor Networks with Higher-Order Topology.

Area of Science:

• Distributed Systems
• Network Science
• Machine Learning

Background:

• Information aggregation in distributed sensor networks is crucial.
• Existing distributed consensus algorithms face communication and energy limitations.
• Non-Bayesian social learning strategies enable agents to learn from information exchange.

Purpose of the Study:

• To design a novel non-Bayesian social learning strategy using higher-order network topology.
• To theoretically analyze the convergence properties and rate of the proposed strategy.
• To demonstrate the framework's effectiveness and superior performance in sensor network applications.

Main Methods:

• Introduction of a hypergraph structure to model higher-order communication topologies.
• Theoretical analysis of convergence and convergence rate for the hypergraph social learning strategy.
• Extensive numerical simulations to validate the framework's performance.

Main Results:

• The proposed hypergraph social learning strategy demonstrates effective information aggregation.
• Theoretical analysis confirms the convergence and convergence rate of the new strategy.
• Superior performance observed in sensor network tasks like cooperative positioning.

Conclusions:

• The hypergraph social learning framework offers an efficient approach for distributed information aggregation.
• The strategy enhances communication topology design for sensor networks, improving resilience.
• The framework has broad theoretical and applied value in distributed estimation and social networks.