deterministic finite automaton for scalable traffic identification: the power of compressing by...
TRANSCRIPT
![Page 1: Deterministic Finite Automaton for Scalable Traffic Identification: the Power of Compressing by Range Authors: Rafael Antonello, Stenio Fernandes, Djamel](https://reader036.vdocuments.us/reader036/viewer/2022083008/56649f325503460f94c4d79a/html5/thumbnails/1.jpg)
Deterministic Finite Automaton for Scalable Traffic Identification: the Power of
Compressing by Range
Authors: Rafael Antonello, Stenio Fernandes, Djamel Sadok, Judith Kelner, Geza Szabo
Publisher: NOMS(Network operations and management symposium), 2012
Present: Pei-Hua Huang
Date: 2014/12/24
Department of Computer Science and Information Engineering National Cheng Kung University, Taiwan R.O.C.
![Page 2: Deterministic Finite Automaton for Scalable Traffic Identification: the Power of Compressing by Range Authors: Rafael Antonello, Stenio Fernandes, Djamel](https://reader036.vdocuments.us/reader036/viewer/2022083008/56649f325503460f94c4d79a/html5/thumbnails/2.jpg)
INTRODUCTION
Decreasing the complexity of matching procedures and reducing the memory consumption of DFAs are the main goals of research studies in this field
RCDFA(Ranged Compressed DFA) is based on the following key observation: several consecutive transitions lead to the same destination state
Smart transition representations result in huge space savings over a standard DFA
Computer & Internet Architecture LabCSIE, National Cheng Kung University
2
![Page 3: Deterministic Finite Automaton for Scalable Traffic Identification: the Power of Compressing by Range Authors: Rafael Antonello, Stenio Fernandes, Djamel](https://reader036.vdocuments.us/reader036/viewer/2022083008/56649f325503460f94c4d79a/html5/thumbnails/3.jpg)
Motivational Example
Suppose the DFA created for ^\x01[\x08\x09][\x03\x04] uses the ASCII table as its input alphabet, it has 5 * 256 = 1280 transitions
Computer & Internet Architecture LabCSIE, National Cheng Kung University
3
![Page 4: Deterministic Finite Automaton for Scalable Traffic Identification: the Power of Compressing by Range Authors: Rafael Antonello, Stenio Fernandes, Djamel](https://reader036.vdocuments.us/reader036/viewer/2022083008/56649f325503460f94c4d79a/html5/thumbnails/4.jpg)
Converting DFA to RCDFA
Separating the traditional transitions from the ranged transitions decreased the number of transitions from 1280 to 2 regular transitions and 8 ranged transitions
Computer & Internet Architecture LabCSIE, National Cheng Kung University
4
![Page 5: Deterministic Finite Automaton for Scalable Traffic Identification: the Power of Compressing by Range Authors: Rafael Antonello, Stenio Fernandes, Djamel](https://reader036.vdocuments.us/reader036/viewer/2022083008/56649f325503460f94c4d79a/html5/thumbnails/5.jpg)
DFA and RCDFA Equivalence
New machine represents consecutive transitions going to the same destination state as a unique ranged transition
tm-n => ql for character ranges cm…cn , where n ≥ m and δ (qi , cj) = ql
Q is the finite set of states, Σ is a finite set of input symbols, δrcdfa(si, (cm, cn)) = δdfa(si, cj) s Q and ∀ ∈ c ∈Σ for j varying from m to n where n ≥ m
Computer & Internet Architecture LabCSIE, National Cheng Kung University
5
![Page 6: Deterministic Finite Automaton for Scalable Traffic Identification: the Power of Compressing by Range Authors: Rafael Antonello, Stenio Fernandes, Djamel](https://reader036.vdocuments.us/reader036/viewer/2022083008/56649f325503460f94c4d79a/html5/thumbnails/6.jpg)
Converting DFA to RCDFA
Computer & Internet Architecture LabCSIE, National Cheng Kung University
6
![Page 7: Deterministic Finite Automaton for Scalable Traffic Identification: the Power of Compressing by Range Authors: Rafael Antonello, Stenio Fernandes, Djamel](https://reader036.vdocuments.us/reader036/viewer/2022083008/56649f325503460f94c4d79a/html5/thumbnails/7.jpg)
Combining Models
RCDFA is orthogonal and very suitable for default transitions’ based models
RCDFA can be applied over D2FA and Fast Compression with minor adaptations
Computer & Internet Architecture LabCSIE, National Cheng Kung University
7
![Page 8: Deterministic Finite Automaton for Scalable Traffic Identification: the Power of Compressing by Range Authors: Rafael Antonello, Stenio Fernandes, Djamel](https://reader036.vdocuments.us/reader036/viewer/2022083008/56649f325503460f94c4d79a/html5/thumbnails/8.jpg)
EXPERIMENTAL RESULTS
Computer & Internet Architecture LabCSIE, National Cheng Kung University
8
![Page 9: Deterministic Finite Automaton for Scalable Traffic Identification: the Power of Compressing by Range Authors: Rafael Antonello, Stenio Fernandes, Djamel](https://reader036.vdocuments.us/reader036/viewer/2022083008/56649f325503460f94c4d79a/html5/thumbnails/9.jpg)
EXPERIMENTAL RESULTS
Computer & Internet Architecture LabCSIE, National Cheng Kung University
9
![Page 10: Deterministic Finite Automaton for Scalable Traffic Identification: the Power of Compressing by Range Authors: Rafael Antonello, Stenio Fernandes, Djamel](https://reader036.vdocuments.us/reader036/viewer/2022083008/56649f325503460f94c4d79a/html5/thumbnails/10.jpg)
EXPERIMENTAL RESULTS
Computer & Internet Architecture LabCSIE, National Cheng Kung University
10
![Page 11: Deterministic Finite Automaton for Scalable Traffic Identification: the Power of Compressing by Range Authors: Rafael Antonello, Stenio Fernandes, Djamel](https://reader036.vdocuments.us/reader036/viewer/2022083008/56649f325503460f94c4d79a/html5/thumbnails/11.jpg)
EXPERIMENTAL RESULTS
Computer & Internet Architecture LabCSIE, National Cheng Kung University
11
![Page 12: Deterministic Finite Automaton for Scalable Traffic Identification: the Power of Compressing by Range Authors: Rafael Antonello, Stenio Fernandes, Djamel](https://reader036.vdocuments.us/reader036/viewer/2022083008/56649f325503460f94c4d79a/html5/thumbnails/12.jpg)
EXPERIMENTAL RESULTS
Computer & Internet Architecture LabCSIE, National Cheng Kung University
12