Wormhole Attack Behaviour in Monte-Carlo Localization for Mobile Sensor Networks

Localization is the basic requirement for network management in Wireless Sensor Networks as it helps nodes find their absolute position coordinates and in gathering information relevant to their locations. A localization algorithm has to be dynamic, scalable and should not impose high computation or communication overhead. The localization systems are also prone to attacks. We target a localization scheme for mobile sensor networks called Monte-Carlo Localization, which study its behavior under the most dangerous attack on localization called Wormhole Attack, also known as Collusion Attack and propose a modified algorithm that can help the localization system retain its accuracy level even in the presence of attacks. Our algorithm has communication cost almost equal to that of original localization algorithm (in this case MCL) in the absence of attacks.

Improving continuous-variable quantum key distribution under local oscillator intensity attack using entanglement in the middle

A modified continuous-variable quantum key distribution （CVQKD） protocol is proposed by originating the entangled source from a malicious third party Eve in the middle instead of generating it from the trustworthy Alice or Bob. This method is able to enhance the efficiency of the CVQKD scheme attacked by local oscillator （LO） intensity attack in terms of the generated secret key rate in quantum communication. The other indication of the improvement is that the maximum transmission distance and the maximum loss tolerance can be increased significantly, especially for CVQKD schemes based on homodyne detection.

基于本地差分隐私的异步横向联邦安全梯度聚合方案

联邦学习作为一种新兴的分布式机器学习框架,通过在用户私有数据不出域的情况下进行联合建模训练,有效地解决了传统机器学习中的数据孤岛和隐私泄露问题。然而,联邦学习存在着训练滞后的客户端拖累全局训练速度的问题,异步联邦学习允许用户在本地完成模型更新后立即上传到服务端并参与到聚合任务中,而无需等待其他用户训练完成。然而,异步联邦学习也存在着无法识别恶意用户上传的错误模型,以及泄露用户隐私的问题。针对这些问题,该文设计一种面向隐私保护的异步联邦的安全梯度聚合方案(SAFL)。用户采用本地差分隐私策略,对本地训练的模型添加扰动并上传到服务端,服务端通过投毒检测算法剔除恶意用户,以实现安全聚合(SA)。最后,理论分析和实验表明在异步联邦学习的场景下,提出的方案能够有效识别出恶意用户,保护用户的本地模型隐私,减少隐私泄露风险,并相对于其他方案在模型的准确率上有较大的提升。

基于文字边缘失真特征的翻拍图像篡改定位

针对翻拍文档图像的篡改定位问题,提出一种基于文字边缘失真特征的翻拍图像篡改定位方法。从文字边缘分布、边缘梯度以及待检测文本与参考文本在边缘梯度上的差异3个方面构建了文字失真特征,并训练了一个基于深度神经网络的分类器进行决策。同时,为了评估检测方法的性能,构建了一个包含120张合法图像、1 200张翻拍篡改文档图像的数据集。实验结果表明:所提出的方法在跨库实验场景下词汇级别的ROC曲线下面积(area under ROC curve,AUC)和等错误率(equal error rate,EER)分别达到了0.84和0.23;与Forensic Similarity (128×128)和DenseFCN相比,所提出的特征结合LightDenseNet的方法在翻拍篡改文档数据集的跨库协议下,词汇级别的AUC指标分别提高了0.06和0.17。

入侵意图分析下的软件定义网络DDoS攻击检测方法

为在数据样本回溯期内解决因本地信息熵值增大而造成的服务攻击问题,维护软件定义网络的运行安全性,提出入侵意图分析下的软件定义网络分布式拒绝服务(DDoS)攻击检测方法。按照软件定义网络场景重构原则,确定因果网转换标准,实现对识别参数的更新处理,完成攻击性行为的入侵意图分析,再定义DDoS数据集,根据攻击行为的时空特性,求解模型参数的取值范围,完成入侵意图分析下软件定义网络DDoS攻击检测方法的设计。实验结果表明,在该算法控制下数据样本回溯期为10 min,低于传统算法,能够较好维护软件定义网络的运行安全性。

基于局部线性重叠聚类算法的网络攻击溯源分析方法

科技的进步使得不法组织可以利用各种先进的攻击手段,对特定目标进行隐匿的、长期持续性的网络攻击。当前大部分研究基于大数据、机器学习和图谱的方法进行攻击溯源检测,从而还原攻击全貌,但其存在检测识别正确率低、算力开销大等问题。为此,提出了一种基于溯源图谱的网络攻击分析方法,利用安全产品日志中的攻击特征划分攻击社团,并结合资产和攻击信息的局部线性关系进行重叠聚类,从而还原攻击路径。该算法已应用于某企业安全运行监管系统,实践证明,其能够有效地溯源系统被入侵的过程与痕迹,改善网络安全威胁感知和预警能力。

基于自编码器和Fisher矢量编码的人脸防伪检测研究

为了应对针对人脸识别系统的欺骗攻击,设计一种基于软件的方法来检测人脸欺骗攻击。首先,从多个颜色通道中随机提取局部视频块并将其向量化,随后训练并使用自编码器将提取的向量进一步编码为局部低维特征;接着,利用这些特征拟合一个高斯混合模型;然后,使用Fisher矢量编码将来自同一样本的全部局部低维特征聚合成一个全局特征向量;最后,通过支持向量机对这些全局特征进行分类。在三个标准人脸欺骗数据集上的实验结果证明了该方法的有效性。在IDIAP数据集上的半错误率为0%,在CASIA数据集和MSU数据集上的等错误率分别为3.33%和2.08%。该方法结合自编码器和Fisher矢量编码,在低维特征的情况下取得了较高的性能,可以有效地抵抗人脸识别中基于照片和视频的欺骗攻击。

Robust Image Watermarking Using Local Invariant Features and Independent Component Analysis

This paper proposes a novel robust image watermarking scheme for digital images using local invariant features and Independent Component Analysis （ICA）. Most present watermarking algorithms are unable to resist geometric distortions that desynchronize the location. The method we propose here is robust to geometric attacks. In order to resist geometric distortions, we use a local invariant feature of the image called the scale invariant feature transform, which is invariant to translation and scaling distortions. The watermark is inserted into the circular patches generated by scale-invariant key point extractor. Rotation invariance is achieved using the translation property of the polar-mapped circular patches. Our method belongs to the blind watermark category, because we use Independent Component Analysis for detection that does not need the original image during detection. Experimental results show that our method is robust against geometric distortion attacks as well as signal-processing attacks.

Cascading failures in local-world evolving networks

The local-world (LW) evolving network model shows a transition for the degree distribution between the exponential and power-law distributions, depending on the LW size. Cascading failures under intentional attacks in LW network models with different LW sizes were investigated using the cascading failures load model. We found that the LW size has a significant impact on the network's robustness against deliberate attacks. It is much easier to trigger cascading failures in LW evolving networks with a larger LW size. Therefore, to avoid cascading failures in real networks with local preferential attachment such as the Internet, the World Trade Web and the multi-agent system, the LW size should be as small as possible.

Denial of Service Due to Direct and Indirect ARP Storm Attacks in LAN Environment

ARP-based Distributed Denial of Service (DDoS) attacks due to ARP-storms can happen in local area networks where many computer systems are infected by worms such as Code Red or by DDoS agents. In ARP attack, the DDoS agents constantly send a barrage of ARP requests to the gateway, or to a victim computer within the same sub-network, and tie up the resource of attacked gateway or host. In this paper, we set to measure the impact of ARP-attack on resource exhaustion of computers in a local area network. Based on attack experiments, we measure the exhaustion of processing and memory resources of a victim computer and also other computers, which are located on the same network as the victim computer. Interestingly enough, it is observed that an ARP-attack not only exhausts resource of the victim computer but also significantly exhausts processing resource of other non-victim computers, which happen to be located on the same local area network as the victim computer.

Lightweight jammer localization algorithm in wireless sensor networks

In wireless sensor networks （WSNs）, as the shared nature of the wireless medium, jam- ming attacks can be easily launched and result in a great damage to the network. How to deal with jamming attacks has become a great concern recently. Finding the location of a jammer is important to take security actions against the jammer, and thus to restore the network communication. After a comprehensive study on the jammer localization problem, a lightweight easy-operated algorithm called triple circles localization （TCL） is proposed. The evaluation results have demonstrated that, compared with other approaches, TCL achieves the best jammer localization accuracy under variable conditions.

面向ASPE的抗合谋攻击图像检索隐私保护方案

现有的基于非对称内积标量保留加密算法(ASPE)实现了云计算下图像检索中的隐私保护,但由于检索时云服务提供商和检索用户不可信以及外部敌手的存在,无法抵抗恶意用户与云服务器的合谋攻击,可能导致包含敏感信息的图像数据泄露。针对多用户场景,提出了一种面向ASPE的抗合谋攻击图像检索隐私保护方案。首先,该方案通过代理重加密解决因向不可信用户传输私钥导致的图像密钥泄露问题;其次,在客户端添加对角矩阵加密解决云服务提供商和检索用户之间因合谋攻击导致的特征密钥泄露问题;最后,通过线性判别分析来解决局部敏感哈希构建索引时因降维导致的检索准确率下降问题。通过安全性分析证明,该方案是安全且有效的,不仅能够抵抗来自云服务提供商和不可信用户的合谋攻击、唯密文攻击、已知背景攻击和已知明文攻击,同时能够实现对检索过程中图像与私钥的保护。实验结果表明,所提方案在保护图像隐私及保证检索效率的前提下,密文域和明文域的检索精度仅相差约2%。

An Algorithm for Generation of Attack Signatures Based on Sequences Alignment

This paper presents a new algorithm for generation of attack signatures based on sequence alignment. The algorithm is composed of two parts: a local alignment algorithm-GASBSLA (Generation of Attack Signatures Based on Sequence Local Alignment) and a multi-sequence alignment algorithm-TGMSA (Tri-stage Gradual Multi-Sequence Alignment). With the inspiration of sequence alignment used in Bioinformatics, GASBSLA replaces global alignment and constant weight penalty model by local alignment and affine penalty model to improve the generality of attack signatures. TGMSA presents a new pruning policy to make the algorithm more insensitive to noises in the generation of attack signatures. In this paper, GASBSLA and TGMSA are described in detail and validated by experiments.

Practical Security of the Continuous-Variable Quantum Key Distribution with Locally-Generated Local Oscillators

Continuous-variable quantum key distribution (CVQKD) with the local local oscillator (LLO) is confronted with new security problems due to the reference pulses transmitted together with quantum signals over the insecure quantum channel. In this paper, we propose a method of phase attack on reference pulses of the LLO-CVQKD with time-multiplexing. Under this phase attack, the phase drifts of reference pulses are manipulated by eavesdroppers, and then the phase compensation error is increased. Consequently, the secret key rate is reduced due to the imperfect phase compensation for quantum signals. Based on the noise model of imperfect phase compensation, the practical security of LLO-CVQKD under phase attack is analyzed. The simulation results show that the practical security is reduced due to the phase attack, yet it is still tight when system parameters are estimated by training signals.

基于显著区域优化的对抗样本攻击方法

在计算机视觉任务中,以卷积神经网络为基础的图像分类模型得到广泛应用,但因其自身的脆弱性容易受到对抗样本的攻击。目前的攻击方法大多会对整张图像进行攻击,产生的全局扰动影响了对抗样本的视觉质量。针对这一问题,提出一种基于显著区域优化的对抗样本攻击方法,利用显著目标检测技术为每张原始图像生成显著图,并将其二值化为显著掩模,将该掩模与对抗扰动相结合,使显著区域内的对抗扰动保留下来,实现对抗扰动的局部添加。通过引入Nadam优化算法,稳定损失函数更新方向并动态调整学习率,提高损失函数收敛速度,从而在保持较高黑盒攻击成功率的同时,有效降低对抗扰动的可察觉性。在ImageNet数据集上分别进行单模型和集成模型环境下的对抗攻击实验,并对各方法生成的对抗样本图像质量进行对比分析,结果表明,与基准方法相比,该方法在集成模型攻击中的隐蔽性指标实现了27.2%的性能提升,黑盒攻击成功率最高达到了92.7%的水平。

基于改进联合概率数据关联的AUV定位算法

基于水下无线传感器网络的位置估计是解决水下自主机器人定位问题的有效方法,然而由于水下无线传感器网络中的传感器节点一般部署在开放水域中,外来的恶意节点在基于到达时间的网络信道中注入恶意噪声延迟会对定位精度产生不利影响。针对这个问题,本文提出了一种改进联合概率数据关联算法。首先,我们考虑了一种异构的网络架构,包括水面浮标节点、水下信标节点和待定位的AUV目标节点,并获取网络架构中信标节点和AUV之间的距离测量;然后,我们设计了一种基于矩阵相似度的恶意噪声识别机制,识别并丢弃受恶意噪声污染的距离测量,并且对检测为未受到恶意噪声污染的距离测量进行分组,利用最小二乘法完成对AUV的初步定位;接着,为了丰富所得到的样本测量点以提高AUV的定位精度,我们在预测点以及检测出未受到恶意噪声污染的测量点的附近生成一系列服从高斯分布的虚拟测量;最后,对这些测量点进行数据关联处理与更新,完成对AUV目标节点的精确定位。仿真结果表明,所提出的定位方法与其他工作相比,在有恶意噪声攻击的情况下仍能保持较好的定位精度。

基于局部邻域滤波的对抗攻击检测方法

目前,卷积神经网络在语音识别、图像分类、自然语言处理、语义分割等方面都取得了良好的应用成果,是计算机应用研究最广泛的技术之一。但研究人员发现当向输入中加入特定的微小扰动时,卷积神经网络(CNN)模型容易产生错误的预测结果,这类含有微小扰动的图像被称为对抗样本,CNN模型易受对抗样本的攻击。对抗样本的出现可能会对安全敏感的领域带来潜在的应用威胁。已有较多的防御方法被提出,其中许多方法对特定攻击方法具有较好的防御效果,但由于实际应用中无法知晓攻击者采用的攻击方式,因此提出不依赖攻击方法的通用防御策略是一个值得研究的问题。为有效地防御各类对抗攻击,本文提出了基于局部邻域滤波的对抗攻击检测方法。首先,通过像素间的相关性对图像进行RGB空间切割。其次将相似的图像块组成立方体。然后,基于立方体中邻域的局部滤波进行去噪,即:通过邻域立方体的3个块得到邻域数据的3维标准差,用于Wiener滤波。再将滤波后的块组映射回RGB彩色空间。最后,将未知样本和它的滤波样本分别作为输入,对模型的分类进行一致性检验,如果模型对他们的分类不相同,则该未知样本为对抗样本,否则为良性样本。实验表明本文检测方法在不同模型中对多种攻击具备防御效果,识别了对抗样本的输入,且在mini-ImageNet数据集上针对C&W、DFool、PGD、TPGD、FGSM、BIM、RFGSM、MI-FGSM以及FFGSM攻击的最优检测结果分别达到0.938、0.893、0.928、0.922、0.866、0.840、0.879、0.889以及0.871,结果表明本文方法在对抗攻击上具有鲁棒性和有效性。

面向人体姿态估计的不可察局部对抗攻击

尽管深度神经网络在很多任务上取得了良好的结果,但是它们对于微小的对抗扰动却很容易出现预测错误.然而在人体姿态估计的对抗攻击任务中,通常需要添加较大的扰动噪声才能攻击成功,这使得其不可察性变差;减少扰动噪声又会削弱攻击效果.为了克服该矛盾,提出一种面向人体姿态估计的两阶段局部对抗攻击方法.所提方法首先通过预攻击估计出扰动关键区域,然后利用不可察性约束在关键区域内生成扰动.方法不仅可以对人体姿态进行有效攻击,而且还能确保最终扰动区域具有低可察性.采用COCO2017作为对抗扰动实验数据集并使用PCK(percentage of correct keypoints)作为评价指标,比较在人体姿态估计模型中IGSM和C&W方法的攻击效果,其PCK降低值分别提高了15.4%与2.8%.实验结果表明所提方法在保证攻击的低可察的同时,能够取得较好的攻击效果.

局部针刺法为主联合患侧面肌功能性电刺激治疗风寒袭络型周围性面瘫疗效观察

目的观察应用局部针刺法为主联合患侧面肌功能性电刺激对风寒袭络型周围性面瘫病的临床康复疗效。方法将符合纳入标准的65例周围性面瘫患者随机分为对照组33例、观察组32例。对照组予常规治疗,观察组行局部针刺法为主联合患侧面肌功能性电刺激治疗,采用面神经功能分级(H-B)、面部残疾指数量表(FDI)评分、中医症状积分评分及临床疗效作为观察指标。结果2组患者在疗程结束后比较,2组H-B评分分级均显著改善,且观察组明显优于对照组(P<0.05);2组的FDIP评分均优治疗前(P<0.05),FDIS评分低于治疗前(P<0.05);中医症状积分均低于治疗前(P<0.05),且观察组优于对照组(P<0.05);观察组治疗总有效率显著优于对照组(P<0.05)。结论运用熊芳丽教授局部针刺为主联合患侧面肌功能性电刺激治疗,可明显加速康复周围性面瘫患者症状、面部表情肌功能恢复,较常规针刺治疗疗效优越。