The Graph Coloring Problem-Review of Algorithms & Neural Networks and a New Proposal

By virtue of its large range of applications, the graph coloring problem has received considerable research interest from mathematicians and engineers alike. Both algorithmic & hardware methods to color a graph subject to adjacency constraints have been explored resulting in a wide variety of options. The actual selection of any one of the software/hardware methods would depend on the specific requirements of a particular application. This paper reviews the major developments that have occurred both in the algorithmic and the hardware domains pertaining to the solution of the gaph coloring problem. Further, a new neural circuit employing non-linear feedback in the form of unipolar comparators is presented which is able to color a graph more effectively than other existing neural networks for the same task. PSPICE simulations confirm the validity of the approach.