Boundary cutting for packet classification

摘要: Decision-tree-based packet classification algorithms such as HiCuts, HyperCuts, and EffiCuts show excellent search performance by exploiting the geometrical representation of rules in a classifier searching for geometric subspace to which each input belongs. However, decision tree involve complicated heuristics determining field number cuts. Moreover, fixed interval-based cutting not relating actual space that rule covers is ineffective results huge storage requirement. A new efficient algorithm using boundary proposed this paper. The finds out performs according boundary. Hence, deterministic rather than involving heuristics, it more effective providing improved memory For sets with 1000-100 000 rules, simulation provides through 10-23 on-chip accesses 1-4 off-chip average.