无线通信网络入侵检测的模糊狮子贝叶斯系统.pdf资源-CSDN文库

需积分: 10 160 浏览量 2020-05-04 11:38:08 上传评论 1 收藏 1.87MB PDF 举报

无线通信网络入侵检测的模糊狮子贝叶斯系统在无线通信网络中，入侵检测是一个非常重要的问题，因为这些网络通常具有有限的资源，从而导致高安全威胁。为了检测和防止这些入侵，入侵检测系统（IDS）是一种非常有效的方法。本文介绍了一种基于模糊狮子贝叶斯系统（FLBS）的入侵检测机制，该机制结合了模糊聚类算法和狮子优化算法来生成概率测度，并且使用后验概率函数来识别异常节点。 FLBS系统使用模糊聚类算法将数据集分组，然后使用狮子优化算法来优化概率测度。然后，将优化后的概率测度应用于每个数据组，生成聚合数据。将聚合数据应用于狮子naive Bayes（LNB）模型中，并使用后验概率函数识别异常节点。 FLBS系统的性能使用KDD Cup 99数据进行评估，并与现有的方法进行比较分析，结果表明FLBS系统具有最高的性能，证明了该系统在入侵检测中的有效性。知识点： 1. 无线通信网络中的入侵检测问题：无线通信网络由于资源有限，容易受到攻击和入侵，入侵检测系统是防止这些攻击的有效方法。 2. 模糊狮子贝叶斯系统（FLBS）：一种结合了模糊聚类算法和狮子优化算法的入侵检测机制，可以生成概率测度并识别异常节点。 3. 模糊聚类算法：一种将数据集划分为多个簇的算法，用于FLBS系统中的数据预处理。 4. 狮子优化算法：一种优化算法，用于优化概率测度和生成狮子naive Bayes（LNB）模型。 5. 狮子naive Bayes（LNB）模型：一种基于狮子优化算法的naive Bayes模型，用于生成概率测度和识别异常节点。 6. 后验概率函数：一种用于识别异常节点的函数，基于狮子naive Bayes（LNB）模型和聚合数据。 7. 入侵检测系统（IDS）：一种防止入侵和攻击的安全系统，用于检测和防止无线通信网络中的入侵。 8. KDD Cup 99数据：一种常用的入侵检测数据集，用于评估FLBS系统的性能。 9. 性能评估：使用准确率和false acceptance rate（FAR）等指标来评估FLBS系统的性能，并与现有的方法进行比较分析。

资源推荐

资源详情

资源评论

J. Cent. South Univ. (2019) 26: 3017−3033

DOI: https://doi.org/10.1007/s11771-019-4233-1

FLBS: Fuzzy lion Bayes system for intrusion detection

in wireless communication network

NARENDRASINH B Gohil

, VDEVYAS Dwivedi

1. Shantilal Shah Engineering College, Bhavnagar, Gujarat, India;

2. C.U.Shah University, Wadhwan City, Gujarat, India

Abstract: An important problem in wireless communication networks (WCNs) is that they have a minimum number of

resources, which leads to high-security threats. An approach to find and detect the attacks is the intrusion detection

system (IDS). In this paper, the fuzzy lion Bayes system (FLBS) is proposed for intrusion detection mechanism. Initially,

the data set is grouped into a number of clusters by the fuzzy clustering algorithm. Here, the Naive Bayes classifier is

integrated with the lion optimization algorithm and the new lion naive Bayes (LNB) is created for optimally

generating the probability measures. Then, the LNB model is applied to each data group, and the aggregated data is

generated. After generating the aggregated data, the LNB model is applied to the aggregated data, and the abnormal

nodes are identified based on the posterior probability function. The performance of the proposed FLBS system is

evaluated using the KDD Cup 99 data and the comparative analysis is performed by the existing methods for the

evaluation metrics accuracy and false acceptance rate (FAR). From the experimental results, it can be shown that the

proposed system has the maximum performance, which shows the effectiveness of the proposed system in the intrusion

detection.

Key words: intrusion detection; wireless communication network; fuzzy clustering; naive Bayes classifier; lion naive

Bayes system

Cite this article as: NARENDRASINH B Gohil, VDEVYAS Dwivedi. FLBS: Fuzzy lion Bayes system for intrusion

detection in wireless communication network [J]. Journal of Central South University, 2019, 26(11): 3017−3033. DOI:

https://doi.org/10.1007/s11771-019-4233-1.

1 Introduction

Due to the technological development, a

number of sensors are developed and they are

grouped to make a WCN, and the cost of these

sensors is very low. The sensors are used to collect

the information from the various fields, and they are

in contact with the destination nodes from which

the users can access their needs [1]. Now a day,

WCNs are the popular networks, and they are easily

adaptable and reconfigurable. While designing the

sensor networks, it should be notable that the

networks consume less power for maximizing the

lifetime of the networks. Meanwhile they provide

the better sensing ability [2]. WCNs [3] are used in

several applications, such as smart city, smart

homes. The WCNs are dynamic in nature, and they

have uncertain environments, so there is a chance of

high risks [4]. The security mechanisms including

user authentication, cryptographic techniques [5],

and firewalls are inadequate to prevent the networks

from the intrusion techniques [6].

One of the major limitations of the WCNs is

that they have a minimum amount of resources,

such as processor and power unit, which leads to

Received date: 2018-06-22; Accepted date: 2019-09-11

Corresponding author: NARENDRASINH B Gohil, Assistant Professor; E-mail: nbgohil1@gmail.com; ORCID: 0000-0003-4885-5844

万方数据

J. Cent. South Univ. (2019) 26: 3017−3033

3018

high-security threats. The challenges in WCNs may

have the abilities to obtain the secret information

like the secret keys by performing eavesdropping

and the denial-of-services (DoS) attacks [7]. Node

capture is also the major issue in WCNs. In contrast

to conventional networks, the WCNs have

abandoned operations and revealed nature. Hence

the attackers can easily capture the sensor nodes in

WCNs. The node capture attack interrupts most of

the security services, such as key management,

access control, secure routing. Therefore, the

actions must be taken for providing the secure

communication in the WCNs [8]. DoS attacks are

the attacks that decrease the ability of the networks

to do its normal operations, and they are very

difficult to manage. There are several reasons to

DoS attacks, namely, hardware failures, exhaustion

of resources, errors in software. The DoS attacks on

Internet are totally different from those of the DoS

attacks in WCNs. Every layer of WCNs is affected

by the various DoS attacks, and the characteristics

and nature of each attack are distinct from the other.

There is no mechanism to find and eliminate all

types of DoS attacks [9].

Therefore, the alternative approach for

detecting and preventing the attacks is the IDS [10,

11]. The IDS [12, 13] is defined as the mechanism

for securing the network from the unauthorized

access, and it plays the significant role for

providing the secure networks [6]. There are two

types of IDSs. The first one is the anomaly-based

intrusion detection. The other name of anomaly-

based detection is the outlier detection, and it

depends on the modeling of the statistical behaviors

[14]. This method detects intruders immediately

when they are entering into the boundary region of

the surveillance areas while maintaining the entire

sensing power. The limitation of this method is that

it requires a high cost. The second type of intrusion

detection mechanism allows the intruders into the

certain range of the surveillance areas and then

detects the intruders. This method requires a low

cost. If the intruder can be identified immediately,

then the detection capacity of the WCNs is high

[15]. The capacity of the IDS is determined by how

quickly it detects an intruder in the WCNs. The

IDSs are situated within the sensors, and they detect

the intruders quickly when they are entering into

the field of Interest (FoI) [16]. Optimization

algorithms [17, 18] are used in IDS.

This paper proposes the fuzzy lion Bayes

system (FLBS) for intrusion detection in WCNs.

Here, the naive Bayes classifier is integrated with

the lion optimization algorithm for generating the

optimal probabilistic measures. Initially, the data

samples are grouped into clusters by the fuzzy

clustering algorithm. Then, the model for each data

group is created by calculating the mean and

variance of the attributes of data. Then, the lion

naive Bayes (LNB) model is applied to each cluster,

and the aggregated data are generated. After

generating the aggregated data, the LNB model is

applied to the aggregated data, and the abnormal

nodes are identified based on the posterior

probability function.

The major contributions of this paper are:

FLBS: The FLBS is proposed for intrusion

detection in WCN, which uses the fuzzy clustering

and LNB classifier.

LNB classifier: The LNB classifier is proposed

by integrating the lion optimization algorithm with

the naive Bayes classifier for optimally generating

the probability measures.

Fitness function: In LNB classifier, the new

fitness function is generated by calculating the

posterior probability of the data attributes of both

positive training data and the negative training data

on the normal class and the abnormal class.

The rest of this paper is organized as follows:

Section 2 presents the motivation of the proposed

IDS; Section 3 presents the system model of the

WCN. Section 4 describes the proposed FLBS for

intrusion detection in WCNs. Results and

discussions are presented in Section 5, and Section

6 concludes the paper.

2 Motivation

A number of mechanisms have been developed

for intrusion detection in WCN over the past

decades. They detect only specific kinds of attacks,

and they have several limitations. In this section,

the various existing research works in intrusion

detection is discussed, and the advantages and

disadvantages of those methods are described. The

challenges of the IDSs are also described here.

2.1 Review of related works

Here, we present the existing research works

of IDS. MA et al [6] have proposed an IDS based

万方数据

J. Cent. South Univ. (2019) 26: 3017−3033

3019

on the spectral clustering and deep neural network

(SCDNN). This method was used to detect an

intruder in networks with large size and parallel

computing. The limitation of this method is that it

detects the parameters by using the observations

and did not use the mathematical theory. DEVI et al

[19] have presented the IDS which uses the

adaptive neuro-fuzzy inference system for detecting

the intruders in the wireless networks. The

advantage of this method is that it ensures the

device to device communication. This method was

designed for detecting only the denial-of-service

attacks. In Ref. [10], AMBUSAIDI et al have

proposed the IDS based on least square support

vector machine (LSSVM-IDS). This method was

used to remove the unrelated and redundant

characteristics of the original data, so the range of

search area was decreased. This method did not

provide the optimal results. CHEN et al [20] have

suggested the IDS called as multi-step attack

pattern recognition method (WMAPRM). This

method had the ability to find the multi-step attacks.

Here, the threshold setting and the weight

calculations were performed through manually.

CHITRAKAR et al [21] have proposed the

IDS based on candidate support vector based

incremental SVM (CSV-ISVM) algorithm. This

method uses the learning process to find the

samples of original data and current data.

CSV-ISVM algorithm was suitable only for the

binary classification. In Ref. [22],

RAMAKRISHNAN et al have proposed the IDS

based on the layered fuzzy control language

(LFCL). This method had a number of advantages,

such as the minimum false positive rate, better

recall value, and the minimum computational time.

The disadvantage of this method is that it needs a

large amount of training. VASUDEVAN et al [23]

have suggested the local outlier factor and stronger

class classifier based hierarchical model for

detecting the intruders in the networks. This method

had the maximum rate of detection and minimum

false alarms. The limitation of this method is that it

was more complex for detecting intruders. CHENG

[14] proposed the differential game model for

detecting intruders in the WCNs. This method was

used to detect the optimal strategies and minimize

the number of intruders. This method was complex.

GUO et al [4] have proposed the multi-protocol

oriented middleware-level intrusion detection

(MP-MID) system for detecting the threats

occurring in the WCNs. This method had the ability

to detect all types of malicious attacks, and it was

adaptable. In this method, the correctness of the

detection process was decreased when the number

of attackers was increased.

2.2 Challenges

The process of detecting the intruders in the

WCNs is the challenging task because the networks

have the dynamic nature. The IDS is needed to

process a large amount of data from the networks,

and it needs high power. Therefore, it was very

arduous to attain the correct detection rate at high

speed [6].

Although, the WCNs have the advantage of

greater coverage region, security, and higher data

rate, they provide a great dare to IDSs because of

the traffic of network data. Due to the large amount

of data presented in the network, it was very

difficult to the IDSs to provide the correct detection

rate, and they have high computational complexity

[10].

The processing of large data set requires very

large memory and the time needed for training such

data set is also very long. To get the entire

information about the network, deep learning is

needed [21].

The attacks in both network data sets and

normal datasets are same, which gives confusion to

the classifiers to classify these attacks [6].

The number of features of the networks will

increase the time needed for performing

computation and decrease the performance of the

networks. Thus, there is a need for subset features

[22].

In Ref. [6], the clustering was performed by

the spectral clustering based on the affinity matrix.

When the data set was very large, the affinity

matrix computation requires more time. Here, the

data samples were categorized by the deep neural

networks which require a great knowledge for

describing the architecture.

3 System model

Figure 1 shows the system model of the WCN.

The WCN consists of a number of sensor nodes,

and each sensor node communicates with other

nodes by transmitting the data. A router connects

万方数据

剩余16页未读，继续阅读

评论收藏

内容反馈

anitachiu_2

粉丝: 31
资源: 801

无线通信网络入侵检测的模糊狮子贝叶斯系统.pdf

网络入侵检测系统.pdf

基于LightGBM的网络入侵检测系统.pdf

检测的网络入侵检测系统设计与实现.pdf

智能入侵检测系统在网络中的应用.pdf

基于深度学习的改进贝叶斯网络入侵检测算法.pdf

机器学习之贝叶斯估计.pdf

基于机器学习的KDD-CUP99网络入侵检测数据集的分析.pdf

基于贝叶斯分类算法的网络入侵检测系统设计_崔斌.caj

人工智能中的贝叶斯方法.pdf

基于网络入侵检测系统的技术改进.pdf

基于深度学习的配电网无线通信入侵检测系统.pdf

基于Linux的网络入侵检测系统.pdf

基于卷积神经网络的网络入侵检测系统.pdf

分布式网络入侵检测系统浅析.pdf

机器学习算法在网络入侵检测中的应用综述.pdf

论文研究-威布尔部件的经验贝叶斯评估.pdf

19.贝叶斯网络.pdf

回归模型中的贝叶斯分析.pdf

基于云计算和机器学习的网络入侵检测系统研究.pdf

网络监控和入侵检测系统netwatch.pdf

数据挖掘在网络入侵检测系统中的应用 (1).pdf

Linux下的网络入侵检测系统研究与实现.pdf

Linux网络入侵检测系统的设计.pdf

数据挖掘在网络入侵检测系统中的应用.pdf

面向云计算的无线通信网络异常入侵识别.pdf

机器学习算法在校园网入侵检测系统中的应用.pdf

朴素贝叶斯法.pdf

最新资源