模糊逻辑与图理论驱动的Ad-Hoc网络数据平面安全信任系统

160 浏览量更新于2024-08-26 收藏 2.16MB PDF 举报

本文主要探讨了"用于确保Ad-Hoc网络数据平面安全的信任管理系统"的研究，发表在2016年9月的IEEE Transactions on Vehicular Technology Vol.65, No.9上。随着Ad-Hoc网络技术的快速发展，包括Vehicular Ad-Hoc Networks (VANETs)、无线传感器网络(WSNs)以及应急和军事通信等领域广泛应用，其开放性和动态拓扑结构使得数据平面面临多种安全威胁。传统的基于身份的安全机制可能不足以应对这些复杂环境中的攻击，如数据篡改、欺骗等。为了提高Ad-Hoc网络的数据平面安全性，研究人员提出了一个创新的信任管理系统。该系统利用模糊逻辑来处理不精确的先验知识，通过模糊推理来评估路径信任值。模糊逻辑的优点在于能够处理不确定性和模糊性，这对于评估网络中动态变化的信任关系非常关键。此外，结合图论，文章构建了一个新颖的信任模型，用于计算节点的信任度，这有助于识别并信任网络中的可靠节点。然而，面对信任管理系统自身可能面临的攻击，如欺骗（slandering）和庇护（harboring），研究者还提出了一种过滤算法。这个算法旨在识别并剔除试图破坏信任系统的恶意行为，从而增强系统的健壮性和抵抗能力。信任度衰减机制也是文中讨论的重要部分，它设计了一套有效的策略，根据节点的历史行为和当前行为判断其信任值的下降程度，确保系统的实时响应和适应性。这篇研究论文深入分析了Ad-Hoc网络数据平面安全的挑战，并通过融合模糊逻辑和图论技术，提出了一种全面的信任管理系统。同时，针对信任管理系统的潜在威胁，提出了针对性的防御措施，为Ad-Hoc网络的数据安全提供了理论支持和实用解决方案。这一研究成果对于保障物联网和移动通信领域中复杂网络环境的安全具有重要意义。

TAN et al.: TRUST MANAGEMENT SYSTEM FOR SECURING DATA PLANE OF AD-HOC NETWORKS 7581

its packets are forwarded correctly. Similarly, the trustwor-

thiness measurement approaches in [20], [25], and [26] also

leverage promiscuous mode to collect information. It is worth

noting that the success of such approach d epends on the ability

to access the header and content of a packet. The adoption

of trust management does not obviate the need for crypto-

graphic security mechanisms, such as packet encryption and

encapsulation. That is, the overheard packet cannot be parsed.

Consequently, trust factors can no longer be obtained. To over-

come the shortcoming, an ACK-based collection approach is

presented in this paper.

In trust management, recommendation and reputation are of-

ten considered to facilitate trust evaluation. In [27], nodes have

to provide forwarding services for others to acquire the rep-

utation value, which is used to pay for services. Consequently,

selﬁsh nodes will be isolated as they do not have enough reputa-

tion value to pay for the required forwarding services. Based on

a hidden active probe method [5], the trustworthiness of a node

is evaluated according to multiple reputation values. Similarly,

recommendation is also integrated in the trust mechanism in

[25]. Despite the advantages, the frequent use of recommen-

dation and reputation also leads to vulnerabilities, which are

exploited by two widely recognized attacks: slandering and

harboring. To defend against them, a ﬁltering algorithm b ased

on the idea of our earlier work [28] is proposed.

A trust model based on fuzzy theory is presented in [16], but

it is not based on fuzzy reasoning systems. Schmidt et al. [17]

have proposed a customizable trust evaluation model based

on fuzzy logic for multiagent systems, particularly for the

automated e-commerce markets. Another linguistic fuzzy trust

model based on fuzzy reasoning is also proposed in [29],

although these two models cannot be directly used in ad-hoc

networks because the security requirements of ad-hoc networks

and the application environments of [17] and [29] vary a lot.

Moreover, some issues such as trust decay in ad-hoc networks

should probably be addressed using different approaches.

III. T

RUST FAC TOR COLLECTION

Trust evaluation always starts from collecting trust factors.

It is crucial to decide what kind of information should be

collected. The decision should concern two aspects: 1) The col-

lected information should expose the misbehaviors and anom-

alies, and 2) the collection method should be valid and feasible.

Regarding the ﬁrst aspect, objectives of the proposed trust

management should be reminisced. Since it aims at enhancing

the data-plane security of ad-hoc networks, we should concen-

trate more on data-plane information. Furthermore, the infor-

mation should reﬂect the changes of networks when attacks

occur. The data plane is responsible for successfully delivering

packets to destinations at a given timescale. Then, malicious

agents can launch three types of attacks regarding data packets:

tampering, suppressing (delaying), and dropping. To defend

against tampering, data packets should be authenticated by end

nodes to ensure integrity. This requires proper cryptographic

security primitives, which demonstrates that the a doption of

trust mechanisms does not obviate the need for traditional secu-

rity mechanisms. A tampered data packet is usually discarded

Fig. 1. Data-packet ACK.

by receivers due to authentication failure, which implies that

tampering attacks can be somehow viewed as dropping attacks

from the perspective of sender nodes. Thus, the data packet

delivery ratio (PDR) is able to indicate tampering attacks to

some degree. The suppressing attack becomes increasingly dan-

gerous for real-time services. For instance, the suppression of

alert messages may cause car accidents in VANETs. Dropping

attacks such as BlackHole is the most serious threat to the

data plane because it completely disrupts data delivery. More

seriously, the suppressing and dropping attacks cannot be effec-

tively detected and resisted by traditional security mechanisms.

According to the above analysis, we get that all three types of

attacks will affect the average delay ( AD) and delivery ratio

of transm itted data packets. Thus, the AD and data PDR are

selected as the trust factors in this paper.

The second aspect implies that collection methods should be

valid and feasible. As described in Section II, if data packets

are protected by cryptography mechanisms, such as encryption,

the monitoring node is no longer able to obtain the required

information. To avoid this incompatibility, we present a new

approach to collect the AD and PDR of the data packets

transmitted by a path. Within a path, the source node generates

and sends data packets, and is responsible for computing the

AD and PDR. Note that, in this paper, AD and PDR r efer to

the average transmission time and delivery ratio of data packets

forwarded along a path, rather than by a node.

The source node numbers and records its recently sent

packets by a record {seq, time}, where seq is a monotonically

increasing sequence number of d ata packets, and time is when

the packet is sent. After receiving N packets with a larger

sequence number than the previously acknowledged one, the

destination node will reply an acknowledgement (ACK) mes-

sage. The process is shown in Fig. 1.

In Fig. 1, seq

refers to the sequence number of the previ-

ously acknowledged packet, and seq

curr

refers to the one that is

acknowledged currently. ts

and td

denote when the success-

fully delivered ith data packet is sent and received b y the source

node and the destination node, respectively, for 1≤i≤N .

ack

is the time when the current ACK message is received.

An ACK message is comprised of {seq

curr

,N,(seq

,td

...,(seq

,td

),...,(seq

,td

)},whereseq

is the se-

quence number of the h

th received data packet after the

previous ACK, for 1 ≤ j ≤ m,1≤ m ≤ N ,and1≤ h

≤ N.

剩余13页未读，继续阅读

weixin_38696590

粉丝: 6

模糊逻辑与图理论驱动的Ad-Hoc网络数据平面安全信任系统

H3C无线局域网交换机管理系统

无人机系统安全保密体系研究.docx

无线传感器网络技术试题.docx

如何在Linux系统中通过脚本设置ad-hoc模式的Wi-Fi共享网络，并确保其他设备可以顺利连接？

如何在Android平台上实现Wi-Fi Direct的Ad-Hoc通信，并确保网络的安全性和稳定性？

如何通过编写脚本在Linux系统中创建一个ad-hoc Wi-Fi共享网络，并确保其他设备可以顺利连接？

如何在WindowsXP系统上配置锐捷RG-WG54U无线网卡以Ad-Hoc模式工作，并确保网络安全？

ad-hoc 网络 与Link16 的联系或者是区别是什么？

在Linux系统中，如何通过脚本配置ad-hoc模式Wi-Fi共享，并确保其他设备能够成功连接？

在Android平台上，如何通过编程接口实现Ad-Hoc通信，并优化其网络连接的稳定性和安全性？

最新资源

ad-hoc 网络与Link16 的联系或者是区别是什么？