A traffic sampling model for measurement using packet identification

A new sampling model for measurement using packet identification on IP network is provided under a principle of the PSAMP, a working group of IETF (to be set up), that a good sampling model should work for all purposes of measurement applications at the same time with a simple way. After researching and analyzing huge amounts of packet headers captured randomly on the CERNET backbone, the result shows that 16 bits of identification field in the IP packet header is enough for matching bits of the sampling mask. Randomization and statistical attributes of the sampling are analyzed, and the experiment also reveals that this sampling method can be used not only in traffic measurement but also for network behavior analysis.