AbstractDeep packet inspection has become extremely important due to network security. In deep packet inspection, the packet payload is compared against a set of patterns specified as regular expressions. Regular expressions are often implemented as deterministic finite automaton (DFA) for matching in linear time at high network link rates. We proposed a predict DFA which can accelerate the processing speed of DFA. A predict DFA uses additional information to predict several next transitions. We tested our proposal on Layer7 rule-set and validated it on real traffic traces, experiments show that our approach offers a significant performance improvement by accelerate rate factors from 1.6 to 2.8 over original DFA.