A Study on the Flow Assignment Efficiency of MENTOR Algorithm

Mesh Network Topological Optimization and Routing (MENTOR) algorithm is a low complexity and efficient partial mesh networks design algorithm. This study explores the relation between design parameters and performance of traffic assignment of MENTOR algorithm. We analyze 384 networks designed by MENTOR for 4 sets of 50 nodes each with equivalently distributed demand and randomly generated locations. For each of these networks, the performances at normal load and at congestion threshold of MENTOR flow assignment are calculated and compared with the optimum solution obtained by solving the linear programming. It is found that the performances of MENTOR strongly depend on “slack”, a design parameter represented the different between the maximum and minimum utilization. For small value of slack, s=0.1, the performances of MENTOR keep very close to that of the optimum solution, while the maximum utilization hardly impact the performances. As the slack get larger the performances of MENTOR become worse and more depend on the maximum utilization. Key-Words: MENTOR, Network Design, Mesh Networks, Flow assignment, Network optimization