Scalable and efficient fault-tolerant protocol for mobility agents in mobile IP-based systems

Existing fault-tolerant protocols to mask failures of multiple mobility (home or foreign) agents in a network have used passive replication techniques. However, they result in high failure-free latency during registration process if the number of mobility agents in the same network increases, and force each mobility agent to manage bindings of all the mobile nodes registering with its network. In this paper, we present a new fault-tolerant protocol using checkpointing and receiver-based pessimistic message logging techniques. The protocol achieves low failure-free latency even if the number of mobility agents in a network increases and improves scalability to a large number of mobile nodes registering with each network compared with the existing protocols. Additionally, the protocol allows each failed mobility agent to recover bindings of the mobile nodes registering with the mobility agent when it is repaired even if all the other mobility agents in the same network concurrently fail. The experimental results show that the stable storage access overhead of our protocol becomes smaller than the linearization and takeover overhead of the existing protocols as the number of mobility agents in a network increases.