TECHNOLOGICAL BASIS OF “INDUSTRY 4.0”
Can Distributed Consensus Algorithms with Asymptotic Convergence Be Applied to Estimate Size of Multi-agent Systems?
Estimating the size of a multi-agent system (MAS) is a fundamental task with applications in routing, resource management, distributed coordination, etc. Achieving accurate network size estimation in MASs is challenging due to the absence of a central authority, lack of global identifiers, dynamic network changes, communication constraints, etc. This paper investigates the applicability of distributed consensus algorithms with asymptotic convergence for estimating MAS size. Specifically, we examine the average consensus algorithm using four different weighting schemes: Maximum Degree (MD), Metropolis-Hastings (MH), Best Constant (BC), and Convex Optimized (OW) weights. Experimental results on a random geometric graph demonstrate that all weighting schemes enable agents’ internal states to asymptotically converge to the value 1/n, where n is the total number of agents. Differences among the schemes are observed in convergence speed and early-stage oscillations, with BC providing the fastest convergence and OW exhibiting more initial fluctuations. Overall, the study confirms that distributed average consensus algorithms with the examined weights can effectively estimate network size in MASs.