A Dual Assignment Network with Applications in Deterministic Communication Path Selection and Multi-Vehicle Target Assignment

Abstract

In this paper, a continuous-time dual neural network model for linear assignment is presented. The model is based on a dual formulation of the primal linear assignment problem. Global convergence of the dual neural network is ensured under given conditions and assumptions. The dual neural network is compact in the sense that its number of neurons is the same as the number of agents. Simulation results on selecting communication paths with deterministic delay and jitter quality of services in networks and assigning multiple vehicles to formation targets are presented to substantiate the efficacy of the dual neural network model.