Một cách tiếp cận mới để sinh ra một kho đầy đủ các bộ dò trong hệ miễn dịch nhân tạo

Nguyễn Văn Trường và đtg Tạp chí KHOA HỌC & CÔNG NGHỆ 80(04): 121 - 125

121

A NEW APPROACH FOR GENERATING A COMPLETE DETECTOR

REPERTOIRE IN ARTIFICIAL IMMUNE SYSTEMS

Nguyen Van Truong1

, Pham Dinh Lam2

1 College of Education – TNU; 2 Board of Information Technology - TNU

ABSTRACT

Generation of the Detector set is a key problem in Artificial Immune Systems (AIS) due to cost of

time and space. A weakness in many detector generating algorithms was random generation of

detectors so that many candidate detectors may be discarded. We present a novel data structure for

extracting data from protected self set. That helps to produce all possible detectors, or a complete

detector repertoire, with lower time and space complexities compare to recent novel approaches.

Theoretical analysis and experimental results show that our approach is effective and feasible.

These new valuable characteristics can make complex AIS have the highest ability to detect data

changes and to reduce false detection.

Keywords: Artificial immune system, Intrusion detection system, self set, detector set, complete

detector repertoire, false detection

INTRODUCTION*

It is impractical for security software to detect

all intrusions in computer networks. Many

network technologies on security came into

being. The more successful one is based on

AIS, which illustrates biology immune system

on computers. It is evaluated as a new and

effective soft computing method in the field

of network security and information security.

Especially, it is suitable for building up the

Intrusion Detection Systems (IDS) that can

protect computer systems against intruders

and the destruction of computer viruses or

other malicious software system.

This paper is concerned with one aspect of

computer security: ability of detecting all

kinds of changes or attacks, both known and

unknown ones. The method explored involves

the most discussed immune models: Negative

selection algorithm. In the algorithm,

generating the detector set plays a very

important role and helps to increase the

overall performance of AIS. Our method is

to concentrate on producing complete

repertoire for negative selection in not only

security systems, but also in problem

requiring some tolerance of noise, or

involving dynamic streams of data (such as

system call sequences).

*

Tel: 0915016063; Email: [email protected]

There are many researches on intrusion

detection generating, such as Exhaustive

Detector Generating Algorithm, Linear Time

Detector generating algorithm, the Greedy

Detector Generating Algorithm [2], [7]. Our

algorithm involves a special data structure

called Location table. It plays an important

part in our algorithm and leads to reduce both

time and space complexities. Furthermore,

our approach has many advantages when

applying in dynamic environment as

mentioned in the 5th section.

The remaining of the paper is organized as

follows. In the 2nd section, we give a brief

literature review of AIS. Our technique of

generating complete repertoire, the main part of

the paper, is presented in the 3rd section. Some

analyses and experiments are given in the 4th

section. The 5th section concludes the paper.

LITERATURE REVIEW

In this section, we first described negative

selection algorithm, very commonly used

algorithm in AISs. After that, we present a bit

some recent researches in the field.

Artificial negative selection is a

computational imitation of self/nonself

discrimination, firstly designed as a change

detection method. This mechanism is first

given by Forrest et al. [1]. The outline of a

typical negative selection algorithm contains

two stages [10].