Paralleled packet classification based on network processor

Abstract

There are a large number of rules in large networks.This feature leads to overlong and sparse bit vectors of BV algorithm,which requires a large amount of storage resources and too many read cycles in network processor.An improved method,Tuple-BV algorithm,was put forward,which combined tuple space partition with the BV algorithm to reduce the length of bit vector and take advantage of parallel processing mechanism and hardware acceleration units in network processor.The improved algorithm shortens the length of bit vector and reduces storage space.The experiments prove that Tuple-BV algorithm is better than BV algorithm in terms of maximum delay and average delay when the number of rules is over one thousand.