Physarum-Inspired Enterprise Network Redesign

We have developed a Physarum-inspired redesign algorithm to redesign the existing enterprise network (EN) with an objective function to minimize the traffic through the backbone. In contrast to the classical Physarum model, which forms an initial network by exploring its surroundings, we have developed a redesign algorithm under the constraint that the topology of EN is given. The nodes and edges of EN are analogous to the food sources and the connecting tubes of Physarum network, whereas the traffic demand on the edges of EN are reversed to match the tubular length. Our redesign algorithm exploits the foraging behavior of Physarum to generate the possible maximum-flow paths by selecting the combination of each node as a source against the remaining as a destination. A factor called affinity factor is defined to select the highly associated node(s) of each source by traversing the generated paths from each source. A clustering algorithm is added to group the associated nodes together and the procedure is repeated to all nodes. The experimental results on a weighted undirected network revealed that the redesign algorithm manages to reduce EN backbone traffic by a maximum of 78% from the initial single-clustered EN by optimally distributing the nodes within the clusters.