A fair sampling algorithm of network flow based on dynamic count filter

The network application of flow level becomes increasingly wide, to analyze network flow completely we need fair sampling. The previous algorithm of SGS (Sketch Guided Sampling) realizes fair sampling by that sampling rate is in inverse ratio to traffic, traffic counting of which uses hash method. In this paper, for the defects of SGS flow counting and the previous counting Bloom Filters, a more efficient fair sampling algorithm called DCFS (Dynamic Count Fair Sampling) is proposed. The algorithm of DCFS applies DCF (Dynamic Count Filter) to count traffic, which is faster more space efficient than SGS, and it can be easily implemented in hardware.