Controllability analysis of multi-agent systems with switching topology over finite fields

Abstract

In this paper, we investigate the controllability problem of multi-agent systems with switching topology over finite fields. The multi-agent system is defined over finite fields, where agents process only values from a finite alphabet. Under leader-follower structure, one agent is selected as a leader for each subsystem. First, we prove that a multi-agent system with switching topology is controllable over a finite field if the graph of the subsystem is a spanning forest, and the size of the field is sufficiently large. Second, we show that, by appropriately selecting leaders, the multi-agent system with switching topology can be controllable over a finite field even if each of its subsystems is not controllable. Specifically, we show that the number of leaders for ensuring controllability of the switched multi-agent system is less than the minimum number of leaders for ensuring the controllability of all subsystems. Finally, it is proved that the multi-agent system is controllable over a finite field if the union of the graphs is a directed path graph or a star graph.

Leader-following flocking for unmanned aerial vehicle swarm with distributed topology control

“…It is shown that γ * almost reaches 1 at β = 1.2, compared with the corresponding value γ * = 0.4 under the AAT rule (i.e., β = π/2). Furthermore, Figures 7(b) and (c) denote the standard deviation curves of relative distance and velocity between the followers and the leader during each iteration, as defined in (8) and (9), respectively. At β = 1.2, the relative distance reaches 0, less than the corresponding values at both β = 1.2 and β = π/2.…”

“…Compared with the traditional centralized control-based network architecture, in a self-organizing network (SON) architecture, because each node can rely on local interactions with its neighbors to perform its communication tasks, the network's control signaling overhead can be greatly reduced by implementing the self-organizing mechanism. The swarm can be controlled by selecting one or more members as leaders to accomplish a number of specified tasks without requiring all members in the swam to maintain connections to the control center [9]. Obviously, the self-organizing control mechanism is very suitable for the drone group.…”

Event‐triggered finite‐time reliable control for nonhomogeneous Markovian jump systems

“…The effectiveness of the proposed event-triggered finite-time controller design method has been verified by a numerical example. To further improve the proposed approach reliability and safety in practical applications, especially in networked control systems 43,44 and multiagent systems, [45][46][47][48] the problems of fading channels, [49][50][51] packet dropouts, 52,53 multiple input channels, 54 and cyber attacks [55][56][57] will be considered in our further work.…”

Multigradient recursive reinforcement learning NN control for affine nonlinear systems with unmodeled dynamics

“…According to (60) and Cauchy-Schwarz inequality, the forward different for V n1 (k) can be derived as…”