Andreas L¨f o January 15, 2009
Contents
1 Introduction
1.1 Problem Description . . . . . . . . . . . . . . . . . . . . . . . . .
1.2 Document Outline . . . . . . . . . . . . . . . . . . . . . . . . . .
1
1
2
2 Concepts and Related Work
2.1 Basic Concepts . . . . . . . . . . . . . . . . . . . . . . .
2.1.1 Artificial Intelligence . . . . . . . . . . . . . . . .
2.1.2 Network Flow . . . . . . . . . . . . . . . . . . . .
2.1.3 Network Event . . . . . . . . . . . . . . . . . . .
2.1.4 Network Measurements . . . . . . . . . . . . . .
2.2 Related Work . . . . . . . . . . . . . . . . . . . . . . . .
2.2.1 Event Detection Methods . . . . . . . . . . . . .
2.2.2 Artificial Intelligence Approaches to Autonomous
2.2.3 Conclusions . . . . . . . . . . . . . . . . . . . . .
3
3
3
4
4
5
5
6
6
7
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
Networks
. . . . .
3 Research Questions
3.1 Hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2 Research Questions . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2.1 What Types of Artificial Intelligence Are Suitable? . . . .
3.2.2 Is It More Effective to Fuse Data From Several Event
Detection Methods? . . . . . . . . . . . . . . . . . . . . .
3.3 Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3.1 Creating Pre-classified Training Data . . . . . . . . . . . .
3.3.2 Choosing Event Detection Methods . . . . . . . . . . . .
3.3.3 Creating a Framework For Comparing Event Detection
Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3.4 Survey of Artificial Intelligence Techniques . . . . . . . .
3.3.5 Fusing Data From Several Event Detection Methods . . .
11
11
12
4 Thesis Outline
13
5 Timeplan
14
i
8
8
8
9
9
9
9
10
6 Other Information
16
6.1 Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
6.2 Ethics Statement . . . .