Performance Improvement on Probabilistic Packet Marking by using History Caching

Probabilistic packet marking (PPM) is one of the methods on IP traceback, which embeds (i.e., marks) the address of an intermediate router into the header of the sampled packet. However, because the PPM does not have a mechanism to avoid an override of marking, there is a problem that it requires a lot of packets to get the packet mark the router far from the victim node. In this paper we propose a new packet marking scheme which uses the history of previously marked packets to decide for marking the current sampled packet. Through simulation results, we show that our proposed method can reduce the required number of packets significantly to reconstruct the paths for attacking nodes compared with the original PPM