Complexity reduction of the Best Multiple Spanning Tree algorithm

Abstract

Spanning Tree Protocol (STP) is an IEEE approved standard that provides path redundancy while preventing undesirable loops in the Ethernet networks. STP uses a single tree for the entire network which may result in inefficient bandwidth allocation since the non-spanning-tree links will not carry traffic at all. As a solution, the IEEE 802.1s working group has introduced a Multiple Spanning Tree Protocol (MSTP) that allows a switch to participate in multiple spanning trees, one tree per group of VLANs. In our previous work, we introduced a graph theoretic approach named Best Multiple Spanning Tree (BMST) algorithm for ranking all of the possible solutions for selection of the proper trees in a network running MSTP and for finding the best solution based on load balancing on links and switches. Although, BMST algorithm can find the best answer for small networks, but its complexity is too large in large scale networks. In this paper, we propose a simple approach named Multiple Best Spanning Tree (MBST) to reduce the computational complexity of the BMST.