Tree-based link-state routing in the presence of routing information corruption Yih Huang a, * , Philip K. McKinley b a Department of Computer Science, George Mason University, Fairfax, VA 22030-4444, USA b Department of Computer Science and Engineering, Michigan State University, East Lansing, MI 48824, USA Received 7 August 2002; accepted 7 August 2002 Abstract Traditionally, link-state routing (LSR) uses two costly techniques to achieve its robustness and responsiveness: message forwarding on every communication link in the broadcast of network status updates, and the periodic broadcast of local status by every router. In this paper, we present a novel LSR protocol, called Tree-based LSR (T-LSR), which reduces the operational overhead of LSR as follows. A leader router is elected to periodically broadcast network status on behalf of all the other routers in the network, and a spanning tree is constructed to support these broadcasts. The spanning tree is used for most flooding operations, although the protocol reverts to conventional flooding during leader election and spanning tree construction. The T-LSR protocol distinguishes itself from previous tree-based, lightweight LSR methods by its fault-tolerance features: in addition to surviving network partitioning, the T-LSR protocol is shown to maintain consistent routing information and leader preferences throughout the network in the presence of undetected transmission/information corruption problems. The results of a simulation study demonstrate that the T-LSR protocol imposes a small fraction of the overhead of conventional LSR. q 2002 Elsevier Science B.V. All rights reserved. Keywords: Link-state routing; Protocol; Tree-based LSR 1. Introduction One of the most important aspects of a computer network is its underlying routing protocol, the set of rules that routers use to exchange network status information in order to compute paths for relaying communication traffic. One widely used routing method is link-state routing (LSR) [1], which makes complete knowledge of the network available to all routers in the network. To do so, each router periodically broadcasts, or floods, its local status to the rest of the network; such broadcasts are often termed link- state advertisements (LSAs) [2,3]. Based on received LSAs, each router locally maintains a complete image of the network, which it uses to make routing decisions. To accommodate dynamics in network status, every router also broadcasts changes in local status (for example, the failure of an incident link) in an event-driven manner. The Open Shortest Path First (OSPF) protocol [2] of the Internet is one of the most well-known LSR protocols. LSR has also been adopted as the routing method for Asynchronous Transfer Mode (ATM) networks [3]. The method used to flood an LSA must be highly robust so as to guarantee delivery to all the nodes reachable from the source of the LSA. The ‘conventional’ flooding protocol [1] works as follows. In order to flood an LSA, the source router sends the LSA to all its neighboring routers. When an LSA is received by another router for the first time, it is forwarded on all incident links, except the one on which it arrived. Copies of LSAs that have already been seen by a router will not be forwarded. Using this procedure, a given LSA traverses every communication link at least once, and each router x has to processd x copies of the LSA, where d x is the degree (the number of incident links) of router x. An example of this flooding protocol is depicted in Fig. 1. A major design challenge of LSR protocols lies in fault tolerance: LSR must be able to survive rare but potentially catastrophic situations, including partitioning of the net- work and the corruption of routing information. The latter problem refers to the corruption of an LSA or other control 0140-3664/03/$ - see front matter q 2002 Elsevier Science B.V. All rights reserved. PII: S0140-3664(02)00202-5 Computer Communications 26 (2003) 691–699 www.elsevier.com/locate/comcom * Corresponding author. E-mail addresses: huangyih@cs.gmu.edu (Y. Huang), mckinley@cse. msu.edu (P.K. McKinley).