Evaluation of an Automated OSPF Area Design Utilty for Wireless Battlefield Networks

Abstract

To achieve a scalable open shortest path first (OSPF) routing domain, an OSPF autonomous system is partitioned into a number of non-backbone areas interconnected by a backbone area. While the design of scalable OSPF domains represents a network engineering challenge that is important to both civilian and military environments, very little has been published about how the process of OSPF area design can be automated and optimized. An exception to this is the Dynamic Domain Optimization Agent (DDOA)-a novel network design agent that has been proposed to automate the assignment of links to OSPF areas and the selection of area border routers (ABRs). This paper describes extensions to earlier work on DDOA and reports the results of new OSPF area design experiments. Specifically, the results demonstrate the capability of DDOA to compute OSPF area designs that simultaneously satisfy multiple routing domain goals and constraints. Among the performance metrics considered include link state advertisement (LSA) control overhead, area diameters, backbone area loading, degree of ABR diversity and area node count. The OSPF area design experiments were conducted on a diverse set of network topologies consisting of node counts ranging up to more than 250 nodes. Furthermore, the experiments were performed with actual prototype code that can be integrated into network planning tools and dynamic network management agent software