Deterministic Finite Automaton for scalable traffic identification: The power of compressing by range

Practical regular expression matching free of scalability and performance barriers

“…Generally, the related solutions can be categorized into four directions: regex grouping [7][8][9], alternative representation [10][11][12][13], hybrid construction [14][15][16][17], and transition compression [18][19][20][21][22][23][24][25][26][27][28][29].…”

“…Transition compression algorithms [18][19][20][21][22][23][24][25][26][27][28][29] focus on eliminating redundant transitions, which are identical inside or among the states, to reduce the memory consumption of DFA. D 2 FA proposed by Kumar et al [19] and A-DFA proposed by Becchi et al [21], which can achieve over 90% compression ratio, are the typical algorithms to remove the duplicated transitions among states by introducing the default transition.…”

“…D 2 FA proposed by Kumar et al [19] and A-DFA proposed by Becchi et al [21], which can achieve over 90% compression ratio, are the typical algorithms to remove the duplicated transitions among states by introducing the default transition. RCDFA proposed by Antonello et al [29] is the representative algorithm to compress the consecutively identical transitions inside states into the single ranged transition, with over 97% compression ratio and no additional memory lookups. The transition compression solutions only work on the premise that DFA has no state explosion.…”

“…Furthermore, the algorithms used for the comparison with our tailor-made transition compression method is D 2 FA [45], A-DFA [21], and RCDFA [29]. They are the most representative software algorithms to reduce the duplicated inter-and intra-state transitions of DFA.…”

“…For RCDFA, it can compress the DFA without introducing additional memory lookups. Besides, D 2 FA combined with RCDFA [29], A-DFA combined with RCDFA [29], and our method combined with RCDFA are compared as well. Table 4 displays OF statistics.…”