Abstract
Since effective Denial of Service (DoS) solutions are based on the quite expensive commercial devices that perform stateful filtering, in general, they are not always available for stateless traffic, and are not suitable for all organizations. The random port hopping (RPH) by Badishi {\it{et al.}} (2005, 2007) provides a robust communication protocol to decentralize influences by malicious DoS attacks, and is regarded as a low-cost and dependable packet filtering, where the port number used for communication is changed randomly. However, the RPH has not been used yet as a standard communication protocol in the real world, because the utility and limitation of RPH against general DoS attack patterns are still unclear. In this paper, we develop quantitative dependability models of RPH by means of the discrete-time Markov chain (DTMC) and refine the existing RPH protocol in terms of the communication success rate.