Internet of Things (IoT) platforms with trigger-action capability allow event conditions to trigger actions in IoT devices autonomously by creating a chain of interactions. Adversaries exploit this chain of interactions to maliciously inject fake event conditions into IoT hubs, triggering unauthorized actions on target IoT devices to implement remote injection attacks. Existing defense mechanisms focus mainly on the verification of event transactions using physical event fingerprints to enforce security policies to block unsafe event transactions. These approaches are designed to provide offline defense against injection attacks. The state-of-the-art online defense mechanisms offer real-time defense, but extensive dependency on the inference of attack impacts on the IoT network limits the generalization capability of these approaches. In this paper, we propose a platform-independent multi-agent online defense system, namely RESTRAIN, to counter remote injection attacks at runtime. RESTRAIN allows the defense agent to profile attack actions at runtime and leverages reinforcement learning to optimize a defense policy that complies with the security requirements of the IoT network. The experimental results show that the defense agent effectively takes real-time defense actions against complex and dynamic remote injection attacks and maximizes the security gain with minimal computational overhead.

This paper proposes a novel model-and-data-driven adaptive frequency control (MDAFC) for microgrid (MG) systems. The proposed MDAFC includes two control loops, i.e., an adaptive model predictive control (AMPC) loop and a model free adaptive control (MFAC) loop. The AMPC loop not only utilizes the exact model information to improve the control performance but also employs an unscented Kalman filter (UKF) to estimate the unknown parameters of the internal prediction model to improve the robustness to a certain degree. The MFAC loop is designed to address the unmodeled dynamics, nonlinear uncertainty and disturbance of the MG system by virtue of its adaptation mechanism and the data-driven characteristics. Therefore, the MFAC loop can compensate the poor impact of the inaccuracy model information on the AMPC method. To validate the effectiveness of the proposed method, a low inertia MG system containing renewable energy sources (RESs) is considered in this research. Simulation results show that the proposed method can achieve a high control performance. It can effectively cope with the frequency fluctuation caused by RESs, large load consumption, and other unknown uncertain factors. Compared with the existing single AMPC loop and the single proportional integral control loop, the proposed dual-loop-based MDAFC performs better since the two control loops cooperate with each other to leverage their advantages and compensate for their shortcomings.

Conversational retrieval refers to an information retrieval system that operates in an iterative and interactive manner, requiring the retrieval of various external resources, such as persona, knowledge, and even response, to effectively engage with the user and successfully complete the dialogue. However, most previous work trained independent retrievers for each specific resource, resulting in sub-optimal performance and low efficiency. Thus, we propose a multi-task framework function as a universal retriever for three dominant retrieval tasks during the conversation: persona selection, knowledge selection, and response selection. To this end, we design a dual-encoder architecture consisting of a context-adaptive dialogue encoder and a candidate encoder, aiming to attention to the relevant context from the long dialogue and retrieve suitable candidates by simply a dot product. Furthermore, we introduce two loss constraints to capture the subtle relationship between dialogue context and different candidates by regarding historically selected candidates as hard negatives. Extensive experiments and analysis establish state-of-the-art retrieval quality both within and outside its training domain, revealing the promising potential and generalization capability of our model to serve as a universal retriever for different candidate selection tasks simultaneously.

Trigger-action Internet of Things (loT) platforms allow loT devices to create a chain of interactions to automate network tasks by leveraging functional dependencies between loT event conditions and actions. When network devices notify their cyber states to the loT hub by reporting event conditions, the hub utilizes this chain to invoke actions in corresponding loT devices dictated by user-defined rules. Adversaries exploit this scenario to implement remote injection attacks by maliciously reporting fake event conditions to force the hub to command target loT devices to perform invalid actions violating rule integrity. Security mechanisms in the existing literature either require complete visibility over network events to provide an effective defense against dynamic injection attacks or do not offer real-time security. In this paper, we propose Io'I'Haven, an online defense system that counters remote injection attacks at runtime. Even with partial visibility over network states, our system can discern an optimal defense policy, maximizing the overall security gain. We train an LSTM-based function approximator to determine the optimal defense action at each timestep. Experimental results show that IotHaven effectively counters attack progression at runtime with minimal computation overhead.

In trigger-action IoT platforms, IoT devices report event conditions to IoT hubs notifying their cyber states and let the hubs invoke actions in other IoT devices based on functional dependencies defined as rules in a rule engine. These functional dependencies create a chain of interactions that help automate network tasks. Adversaries exploit this chain to report fake event conditions to IoT hubs and perform remote injection attacks upon a smart environment to indirectly control target IoT devices. Existing defense efforts usually depend on static analysis over IoT apps to develop rule-based anomaly detection mechanisms. We also see ML-based defense mechanisms in the literature that harness physical event fingerprints to determine anomalies in an IoT network. However, these methods often demonstrate long response time and lack of adaptability when facing complicated attacks. In this paper, we propose to build a deep reinforcement learning based real-time defense system for injection attacks. We define the reward functions for defenders and implement a deep Q-network based approach to identify the optimal defense policy. Our experiments show that the proposed mechanism can effectively and accurately identify and defend against injection attacks with reasonable computation overhead.

Conditional variational models, using either continuous or discrete latent variables, are powerful for open-domain dialogue response generation. However, previous works show that continuous latent variables tend to reduce the coherence of generated responses. In this paper, we also found that discrete latent variables have difficulty capturing more diverse expressions. To tackle these problems, we combine the merits of both continuous and discrete latent variables and propose a Hybrid Latent Variable (HLV) method. Specifically, HLV constrains the global semantics of responses through discrete latent variables and enriches responses with continuous latent variables. Thus, we diversify the generated responses while maintaining relevance and coherence. In addition, we propose Conditional Hybrid Variational Transformer (CHVT) to construct and to utilize HLV with transformers for dialogue generation. Through fine-grained symbolic-level semantic information and additive Gaussian mixing, we construct the distribution of continuous variables, prompting the generation of diverse expressions. Meanwhile, to maintain the relevance and coherence, the discrete latent variable is optimized by self-separation training. Experimental results on two dialogue generation datasets (DailyDialog and Opensubtitles) show that CHVT is superior to traditional transformer-based variational mechanism w.r.t. diversity, relevance and coherence metrics. Moreover, we also demonstrate the benefit of applying HLV to fine-tuning two pre-trained dialogue models (PLATO and BART-base).

With the emergence and fast development of cloud computing and outsourced services, more and more companies start to use managed security service providers (MSSP) as their security service team. This approach can save the budget on maintaining its own security teams and depend on professional security persons to protect the company infrastructures and intellectual property. However, this approach also gives the MSSP opportunities to honor only a part of the security service level agreement. To prevent this from happening, researchers propose to use outsourced network testing to verify the execution of the security policies. During this procedure, the end customer has to design network testing traffic and provide it to the testers. Since the testing traffic is designed based on the security rules and selectors, external testers could derive the customer network security setup, and conduct subsequent attacks based on the learned knowledge. To protect the network security configuration secrecy in outsourced testing, in this paper we propose different methods to hide the accurate information. For Regex-based security selectors, we propose to introduce fake testing traffic to confuse the testers. For exact match and range based selectors, we propose to use NAT VM to hide the accurate information. We conduct simulation to show the protection effectiveness under different scenarios. We also discuss the advantages of our approaches and the potential challenges.

Moving target defenses (MTDs) are widely used as an active defense strategy for thwarting cyberattacks on cyber-physical systems by increasing diversity of software and network paths. Recently, machine Learning (ML) and deep Learning (DL) models have been demonstrated to defeat some of the cyber defenses by learning attack detection patterns and defense strategies. It raises concerns about the susceptibility of MTD to ML and DL methods. In this article, we analyze the effectiveness of ML and DL models when it comes to deciphering MTD methods and ultimately evade MTD-based protections in real-time systems. Specifically, we consider a MTD algorithm that periodically randomizes address assignments within the MIL-STD-1553 protocol—a military standard serial data bus. Two ML and DL-based tasks are performed on MIL-STD-1553 protocol to measure the effectiveness of the learning models in deciphering the MTD algorithm: 1) determining whether there is an address assignments change i.e., whether the given system employs a MTD protocol and if it does 2) predicting the future address assignments. The supervised learning models (random forest and k-nearest neighbors) effectively detected the address assignment changes and classified whether the given system is equipped with a specified MTD protocol. On the other hand, the unsupervised learning model (K-means) was significantly less effective. The DL model (long short-term memory) was able to predict the future addresses with varied effectiveness based on MTD algorithm's settings.

With the development of Security as a Service (SAAS), many companies outsource their network security functionality to security service providers. To guarantee the execution and quality of such services, a third party can help the end customer verify the enforcement of the security service level agreement (SSLA). Since individual testers often lack the capability and trustworthiness to attract many customers, a platform is needed to bridge the gap between the customers and the testers. In this paper, we investigate the incentivisation of outsourced network testing from the platform perspective. We first define the problem of cost/benefit model of the platform and identify the restriction factors. We describe multiple testing task assignment scenarios and prove that they are NP problems. Next we design heuristic algorithms for the problem. Our simulation results examine the performance of the heuristic approaches.

Physical activity (PA) plays a crucial role in leading a healthy life without chronic diseases. Among various PAs, weight lifting, one of the essential stationary exercises, is an integral part of routine workout sessions. Being aware of the intensity of the performed exercise is also an essential factor in keeping track of the workout. Inspired by this, we propose a low-cost quantitative weight lifting assessment system, PaWLA, leveraging only a single Photoplethysmography (PPG) sensor. Particularly, we design PaWLA as a mobile weight recognition system that can classify the user’s lifted weight into its corresponding label based on PPG sensor readings from the wrist region. The changes in blood volume in the radial artery due to the strain of lifting the weight are exploited via PPG sensor readings in this work. We build our custom hardware prototype using COTS components to prove the system’s feasibility. Evaluation of the system with nine volunteers shows that PaWLA can achieve an average F1 score of up to 97.4%, proving the feasibility and efficiency of the proposed method.

With the emergence and fast development of trigger-action platforms in IoT settings, security vulnerabilities caused by the interactions among IoT devices become more prevalent. The event occurrence at one device triggers an action in another device, which may eventually contribute to the creation of a chain of events in a network. Adversaries exploit the chain effect to compromise IoT devices and trigger actions of interest remotely just by injecting malicious events into the chain. To address security vulnerabilities caused by trigger-action scenarios, existing research efforts focus on validation of the security properties of devices, or verification of the occurrence of certain events based on their physical fingerprints on a device. We propose IoTMonitor, a security analysis system that discerns the underlying chain of event occurrences with the highest probability by observing a chain of physical evidence collected by sensors. We use the Baum-Welch algorithm to estimate transition and emission probabilities and the Viterbi algorithm to discern the event sequence. We can then identify the crucial nodes in the trigger-action sequence whose compromise allows attackers to reach their final goals. The experiment results of our designed system upon the PEEVES datasets show that we can rebuild the event occurrence sequence with high accuracy from the observations and identify the crucial nodes on the attack paths.

Abstract The essential of developing an advanced driving assistance system is to learn human-like decisions to enhance driving safety. When controlling a vehicle, joining roundabouts smoothly and timely is a challenging task even for human drivers. In this paper, we propose a novel imitation learning based decision making framework to provide recommendations to join roundabouts. Our proposed approach takes observations from a monocular camera mounted on vehicle as input and use deep policy networks to provide decisions when is the best timing to enter a roundabout. The domain expert guided learning framework can not only improve the decision-making but also speed up the convergence of the deep policy networks. We evaluate the proposed framework by comparing with state-of-the-art supervised learning methods, including conventional supervised learning methods, such as SVM and kNN, and deep learning based methods. The experimental results demonstrate that the imitation learning-based decision making framework, which ourperforms supervised learning methods, can be applied in driving assistance system to facilitate better decision-making when approaching roundabouts.

In open-domain dialogue systems, knowledge information such as unstructured persona profiles, text descriptions and structured knowledge graph can help incorporate abundant background facts for delivering more engaging and informative responses. Existing studies attempted to model a general posterior distribution over candidate knowledge by considering the entire response utterance as a whole at the beginning of decoding process for knowledge selection. However, a single smooth distribution could fail to model the variability of knowledge selection patterns over different decoding steps, and make the knowledge expression less consistent. To remedy this issue, we propose an adaptive posterior knowledge selection framework, which sequentially introduces a series of discriminative distributions to dynamically control when and what knowledge should be used in specific decoding steps. The adaptive distributions can also capture knowledge-relevant semantic dependencies between adjacent words to refine response generation. In particular, for knowledge graph-grounded dialogue generation, we further incorporate the adaptive distributions into generative word distributions to help express the knowledge entity words. The experimental results show that our developed methods outperform strong baseline systems by large margins.

Navigation and situation awareness are amongst the most important aspects of collaborative analysis in 3D environments. This paper investigates the latest technology of mixed reality to improve the team navigation through effective real-time communication. We develop a cross-platform collaboration system, supporting different types of devices and operating systems, including Microsoft HoloLens and iOS devices. Our system provides essential coordination and information sharing functions by leveraging on device sensors and Vuforia API of image marker to localize users inside a building. We provide a set of essential building navigation, visualization, and interaction methods to support joint tasks in the physical building environments among participants with mobile devices in real-time. We have performed a user study to evaluate different devices used in coordination tasks. Our results demonstrate the effects of immersive visualization for improving 3D navigation and coordination.

Data provenance collects comprehensive information about the events and operations in a computer system at both application and kernel levels. It provides a detailed and accurate history of transactions that help delineate the data flow scenario across the whole system. Data provenance helps achieve system resilience by uncovering several malicious attack traces after a system compromise that are leveraged by the analyzer to understand the attack behavior and discover the level of damage. Existing literature demonstrates a number of research efforts on information capture, management, and analysis of data provenance. In recent years, provenance in IoT devices attracts several research efforts because of the proliferation of commodity IoT devices. In this survey paper, we present a comparative study of the state-of-the-art approaches to provenance by classifying them based on frameworks, deployed techniques, and subjects of interest. We also discuss the emergence and scope of data provenance in IoT network. Finally, we present the urgency in several directions that data provenance needs to pursue, including data management and analysis.

Generating emotional responses is crucial for building human-like dialogue systems. However, existing studies have focused only on generating responses by controlling the agents' emotions, while the feelings of the users, which are the ultimate concern of a dialogue system, have been neglected. In this paper, we propose a novel variational model named EmoElicitor to generate appropriate responses that can elicit user's specific emotion. We incorporate the next-round utterance after the response into the posterior network to enrich the context, and we decompose single latent variable into several sequential ones to guide response generation with the help of a pre-trained language model. Extensive experiments conducted on real-world dataset show that EmoElicitor not only performs better than the baselines in term of diversity and semantic similarity, but also can elicit emotion with higher accuracy.

In open-domain dialogue systems, dialogue cues such as emotion, persona, and emoji can be incorporated into conversation models for strengthening the semantic relevance of generated responses. Existing neural response generation models either incorporate dialogue cue into decoder’s initial state or embed the cue indiscriminately into the state of every generated word, which may cause the gradients of the embedded cue to vanish or disturb the semantic relevance of generated words during back propagation. In this paper, we propose a Cue Adaptive Decoder (CueAD) that aims to dynamically determine the involvement of a cue at each generation step in the decoding. For this purpose, we extend the Gated Recurrent Unit (GRU) network with an adaptive cue representation for facilitating cue incorporation, in which an adaptive gating unit is utilized to decide when to incorporate cue information so that the cue can provide useful clues for enhancing the semantic relevance of the generated words. Experimental results show that CueAD outperforms state-of-the-art baselines with large margins.

The evolutionary multi-objective optimization algorithm was used to optimize the clustering and splitting of nodes in the construction of inverted spatial objects index tree(ITSR) in this paper. Considering the objective factors including object's coverage, overlap, group center distance, directory rectangle perimeter and the word similarity between tree nodes, etc., a multi-objective optimization model for solving the optimal ITSR construction is established. To solve the model, this paper find a novel method of constructing high efficiency inverted text-spatial R-tree index. The experimental results show that the algorithm supports the multi-objective optimal clustering of ITSR with high efficiency and accuracy .

Weichao Wang, Shi Feng, Daling Wang, Yifei Zhang. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.

Cloud manufacturing (CM) is an open and service-oriented platform that virtualizes distributed design, machining, and assembly resources together in order to provide a seamless, adaptive, and high quality transaction of manufacturing procedures. Despite the results in resource discovery and planning, the research in robustness and security of the systems falls behind in many aspects. The lack of such knowledge puts a serious challenge for the wide deployment and adoption of cloud manufacturing which is naturally connected to the Internet and exposed to cyber attacks. To bridge this gap, we study a specific type of equipment contention attack in cloud manufacturing. Through requesting extra shares of scarce resources, an attacker can gain advantage in competition with other users and reduce the efficiency of the overall system. We design a mechanism to mitigate such attacks through measuring the remaining machining capabilities in the cloud. The cloud administrator can control cost difference between the attacker and subsequent service requesters. Simulations of a mid-size city manufacturing cloud are conducted and the results show that our approach will incur low increases in cost for benign users while discouraging the equipment contention attacks.

With the rapid increasing of smart phones and the advances of embedded sensing technologies, mobile crowd sensing (MCS) becomes an emerging sensing paradigm for large-scale sensing applications. One of the key challenges of large-scale mobile crowd sensing is how to effectively select the minimum set of participants from the huge user pool to perform the tasks and achieve a certain level of coverage while satisfying some constraints. This becomes more complex when the sensing tasks are dynamic (coming in real time) and heterogeneous (with different temporal and spacial coverage requirements). In this paper, we consider such a dynamic participant selection problem with heterogeneous sensing tasks which aims to minimize the sensing cost while maintaining certain level of probabilistic coverage. Both offline and online algorithms are proposed to solve the challenging problem. Extensive simulations over a real-life mobile dataset confirm the efficiency of the proposed algorithms.

Sentiment expression in microblog posts can be affected by user's personal character, opinion bias, political stance and so on. Most of existing personalized microblog sentiment classification methods suffer from the insufficiency of discriminative tweets for personalization learning. We observed that microblog users have consistent individuality and opinion bias in different languages. Based on this observation, in this paper we propose a novel user-attention-based Convolutional Neural Network (CNN) model with adversarial cross-lingual learning framework. The user attention mechanism is leveraged in CNN model to capture user's language-specific individuality from the posts. Then the attention-based CNN model is incorporated into a novel adversarial cross-lingual learning framework, in which with the help of user properties as bridge between languages, we can extract the language-specific features and language-independent features to enrich the user post representation so as to alleviate the data insufficiency problem. Results on English and Chinese microblog datasets confirm that our method outperforms state-of-the-art baseline algorithms with large margins.

Being able to join roundabouts correctly is crucial for an autonomous vehicle to maintain not only its own safety but also a normal traffic order for others. In order to know the right time and speed for entering roundabouts, the location, speed and direction of the approaching vehicles need to be taken into consideration. This study investigated the feasibility of leveraging computer vision and machine learning to help autonomous vehicles decide to wait or to enter when reaching roundabouts. A grid-based image processing approach with a single camera at normal roundabouts (GBIPA-SC-NR) is proposed in this paper to characterize traffic situations that can be used for machine learning algorithms to learn the roundabout joining criteria. Video road clips recorded when human drivers reach and then join various roundabouts at different locations were utilised for this learning process, with a selection of four supervised classification algorithms (i.e. the Support Vector Machines, Random Forests, K-Nearest Neighbours, and Decision Tree). The trained classifiers using the proposed approach were evaluated on 507 test videos captured at roundabouts, where the SVM showed the best performance with a 90.28% classification accuracy. This result suggests that the proposed grid-based image processing method can be applied to effectively help autonomous vehicles made the right decision when reaching a roundabout.

Medical sensors are usually attached to or implanted inside patient body. Since sensing results of Body Area Networks (BAN) can directly impact the control of medical equipment, the authenticity and integrity of sensing data is essential for safety of patients. Restricted by the limited resources available to BAN sensors, researchers have referred to Physical Unclonable Function of the nodes to achieve authentication. Existing approaches focus on the authentication between control unit and sensors. Mutual authentication among body sensors has not been carefully studied. In this paper, we propose to design a lightweight mutual authentication mechanism for BAN sensors with physical unclonable functions (PUF). Using control unit as a middle point, a pair of body sensors can establish shared secrets so that authenticity of exchanged data can be protected. The proposed approach does not require sensors to conduct any encryption operations, which suits the restricted resources available to BAN nodes. The analysis shows that the proposed approach has very low overhead and does not introduce new vulnerabilities into the system.

Auction based participant selection has been widely used for mobile crowd sensing (MCS) to achieve user incentive and assignment optimization. However, mobile crowd sensing problems solved with auction-based approaches usually involve participants' privacy concerns because a participant's bids may contain her private information (such as location visiting patterns), and disclosure participants' bids may disclose their private information as well. In this paper, we study how to protect such bid privacy in a temporally and spatially dynamic MCS system. We assume that both sensing tasks and mobile participants have dynamic characteristics over spatial and temporal domains. Following the classical VCG auction, we carefully design a scalable grouping based privacy-preserving participant selection scheme, which leverages Lagrange polynomial interpolation to perturb participants' bids within groups. The proposed solution does not affect the operation of current MCS platform. Both theoretical analysis and real-life tracing data simulations verify the efficiency and security of the proposed solution.

Information-Centric Networking (ICN) is proposed to address the inefficiency of content delivery of IP networks from the perspective of architecture. In contrast, Content Delivery Network (CDN) is an overlay solution in current IP networks. We believe that even though ICN is fully deployed, there is still a role for CDNs to play in ICN networks. Since ISPs in ICN will replicate and forward contents according to their policies and interests, it may not align with the objectives of Content Providers (CPs). Therefore, CPs are willing pay a third party (i.e., CDN providers) a certain fee to meet their own requirements. In this paper, we propose to use the inventory model of Supply Chain Management (SCM) in logistics to formulate the content delivery process of ICN networks. The product-centric model of SCM is well-suited for the content-centric content delivery process of ICN networks. Also, we propose the system framework of inventory Centric Delivery Network (iCDN). Simulation results show that the average cost and link usage of the SCM-based algorithm can be reduced by 52% and 15% respectively compared to the baseline approach.

With the rapid increasing of smart phones and their embedded sensing technologies, mobile crowd sensing (MCS) becomes an emerging sensing paradigm for performing large-scale sensing tasks. One of the key challenges of large-scale mobile crowd sensing systems is how to effectively select the minimum set of participants from the huge user pool to perform the tasks and achieve certain level of coverage. In this paper, we introduce a new MCS architecture which leverages the cached sensing data to fulfill partial sensing tasks in order to reduce the size of selected participant set. We present a newly designed participant selection algorithm with caching and evaluate it via extensive simulations with a real-world mobile dataset.

In cloud computing, quality of services is often enforced through Service Level Agreement (SLA) between end users and cloud providers. While SLAs on hardware resources such as CPU cycles or bandwidth can be monitored by low layer sensors, the enforcement of security SLAs stays a very challenging problem. Several high level architectures for security SLAs have been proposed. However, details still need to be filled before they can be deployed. In this paper, we propose to design mechanisms to detect violations of security SLAs. Specifically, we focus on unauthorized accesses to memory pages of a virtual machine and violation of the memory deduplication policies. Through measuring the accumulated memory access latency, we try to derive out whether or not the memory pages have been swapped out and the order of accesses to them. These events will then be compared to access commands issued by the local VM. In this way, unauthorized memory accesses or violation of deduplication policies can be detected. Compared to existing approaches, our mechanisms do not need explicit help from the hypervisor or third parties. Therefore, it can detect SLA violations even when they are initiated by the hypervisor. We implement our approaches under VMWare with Windows virtual machines. Our experiment results show that the VM can effectively detect the violations with small increases in overhead.

Mobile mashups promise great data aggregation and processing capabilities for all end users. During the data collection procedures, some data providers fail to protect confidentiality and privacy of user queries and transmit information in plain text. This enables attackers to eavesdrop on networks and compromise user information. Since mobile mashups can adopt server-side, client-side, or hybrid architectures, no one-size-fits-all solutions can be designed to solve this problem. In this paper, we propose to design two mechanisms using mobile clouds to preserve data query privacy in mobile mashups. For server-side mashups, we propose to use dynamically created virtual machines as proxies to process data collection and aggregation in order to prevent information leakage through eavesdropping. For client-side mashups, we propose to use live migration of the application level virtual machines into mobile cloud to hide the data collection and aggregation procedures from attackers. We will evaluate the proposed approaches through both analysis and experiments on real platforms.

As protocol narrations are widely used to describe security protocols, efforts have been made to formalize or devise semantics for them. An important, but largely neglected, question is whether or not the formalism faithfully accounts for the attacker's view. Several attempts have been made in the literature to recover the attacker's view. They, however, are rather restricted in scope and quite complex. This greatly impedes the ability of protocol verification tools to detect intricate attacks.

Although various past efforts have been made to characterize and detect guessing attacks, there is no consensus on the definition of guessing attacks. Such a lack of generic definition makes it extremely difficult to evaluate the resilience of security protocols to guessing attacks.

The accurate estimation of component loads in a helicopter is an important goal for life cycle management and life extension efforts. This paper explores the use of evolutionary computational methods to help estimate some of these helicopter dynamic loads. Thirty standard time-dependent flight state and control system parameters were used to construct a set of 180 input variables to estimate the main rotor blade normal bending during forward level flight at full speed. Evolutionary computation methods (single and multi-objective genetic algorithms) optimizing residual variance, gradient, and number of predictor variables were employed to find subsets of the input variables with modeling potential. Clustering was used for composing a statistically representative training set. Machine learning techniques were applied for prediction of the main rotor blade normal bending involving neural networks, model trees (black and white box techniques) and their ensemble models. The results from this work demonstrate that reasonably accurate models for predicting component loads can be obtained using smaller subsets of predictor variables found by evolutionary-computation based approaches.

Accurately determining component loads on a helicopter is an important goal in the helicopter structural integrity field. While measuring dynamic component loads directly is possible, these measurement methods are not reliable and are difficult to maintain. This paper explores the potential of using computational intelligence methods to estimate some of these helicopter dynamic loads. Thirty standard time-dependent flight state and control system parameters were used to construct a set of 180 input variables to estimate the main rotor blade normal bending during forward level flight at full speed. Unsupervised nonlinear mapping was used to study the structure of the multidimensional time series from the predictor and target variables. Based on these criteria, black and white box modeling techniques (including ensemble models) for main rotor blade normal bending prediction were applied. They include neural networks, local linear regression and model trees, in combination with genetic algorithms based on residual variance (gamma test) for predictor variables selection. The results from this initial work demonstrate that accurate models for predicting component loads can be obtained using the entire set of predictor variables, as well as with smaller subsets found by computational intelligence based approaches.

OS fingerprinting tries to identify the type and version of a system based on gathered information of a target host. It is an essential step for many subsequent penetration attempts and attacks. Traditional OS fingerprinting depends on banner grabbing schemes or network traffic analysis results to identify the system. These interactive procedures can be detected by intrusion detection systems (IDS) or fooled by fake network packets. In this paper, we propose a new OS fingerprinting mechanism in virtual machine hypervisors that adopt the memory de-duplication technique. Specifically, when multiple memory pages with the same contents occupy only one physical page, their reading and writing access delay will demonstrate some special properties. We use the accumulated access delay to the memory pages that are unique to some specific OS images to derive out whether or not our VM instance and the target VM are using the same OS. The experiment results on VMware ESXi hypervisor with both Windows and Ubuntu Linux OS images show the practicability of the attack. We also discuss the mechanisms to defend against such attacks by the hypervisors and VMs.

Previous research on the security of network coding focuses on the detection of pollution attacks. The capabilities of network coding to detect malicious attacks have not been fully explored. We propose a new mechanism based on physical layer network coding to detect the Sybil nodes. When two signal sequences collide at the receiver, the starting point of the collision is determined by the distances between the receiver and the senders. When the distance between two receivers is large enough, they can combine their interference sequences to recover the original data packets. On the contrary, the Sybil nodes attached to the same physical device cannot accomplish the data recovery procedure. We have proposed several schemes at both physical and network layers to transform the idea into a practical approach. The investigation shows that the wireless nodes can effectively detect Sybil nodes without the adoption of special hardware or time synchronization.

Previous research on physical layer network coding (PNC) focuses on the improvements in bandwidth usage efficiency. In this paper, we propose a PNC-based node localization mechanism. When two signal sequences collide at the receiver, the starting point of collision is determined by the distances between the receiver and senders. When the signal interference results from two receivers are combined together, we can determine a hyperbola with two senders as the respective focal points. In this way, by using multiple pairs of anchor nodes as senders, we can determine multiple hyperbolas and the node position will be at the intersection point of these hyperbolas. The proposed approach does not require the wireless nodes to be equipped with any special hardware such as synchronized clocks. We propose several schemes at the physical and network layers to transform the idea into a practical approach. We also investigate the overhead, localization accuracy, and safety of the approach.

Many security protocols are vulnerable to type flaw attacks, in which a protocol message may be forged from another message. The previous approaches focus on heuristic schemes to protect specific protocols but fail to expose the enabling factors of such attacks. In this paper, we investigate the relationship between the type flaw attacks on the security protocols and the knowledge of the principals. We formalize the notion of recognizability that characterizes the fact that a message could not be type-flawed. The approach helps us better understand security protocols and gives insights into the detection and prevention of type-flaw attacks.

This paper presents a new approach to intrusion detection that supports the identification and analysis of network anomalies using an interactive coordinated multiple views (CMV) mechanism. A CMV visualization consisting of a node-link diagram, scatterplot, and time histogram is described that allows interactive analysis from different perspectives, as some network anomalies can only be identified through joint features in the provided spaces. Spectral analysis methods are integrated to provide visual cues that allow identification of malicious nodes. An adjacency-based method is developed to generate the time histogram, which allows users to select time ranges in which suspicious activity occurs. Data from Sybil attacks in simulated wireless networks is used as the test bed for the system. The results and discussions demonstrate that intrusion detection can be achieved with a few iterations of CMV exploration. Quantitative results are collected on the accuracy of our approach and comparisons are made to single domain exploration and other high-dimensional projection methods. We believe that this approach can be extended to anomaly detection in general networks, particularly to Internet networks and social networks.

In this paper, we investigate secure intra and inter group information sharing in a network consisting of multiple node groups. We develop a mechanism for the establishment and maintenance of multicast structures, which enables flexible topology changes and efficient information distribution. We develop a key distribution and update method for secure information sharing in the same group and among different groups. It adopts polynomials to support the distribution of personal key shares and employs LKH (logical key hierarchy) to achieve efficient key refreshment. We also investigate the overhead and safety of the proposed mechanism and demonstrate its advantages over previous approaches.

In wireless networks,the authenticity and uniqueness of node identities are essential to the fundamental operations such as routing, resource allocation, and intrusion detection. In this paper, we investigate Sybil attack, an attack in which a malicious node illegitimately acquires multiple identities and performs as these nodes simultaneously. We propose an effective approach to monitoring and detecting such attacks by integrating network security and visualization methods. The security component explores the time-varying network topology and its statistical and geometry information to detect the existence of Sybil attacks. The visualization component incorporates the detection results and provides an effective mechanism to illustrate abnormal topology patterns and locate fake identities. These two components are integrated into a practical system that takes advantage of both interactive visualization and intelligent security methods. Experimental studies are conducted to investigate the impacts of the network parameters such as node connectivity on the detection capability of the proposed mechanism.

Privacy preservation in a peer-to-peer (P2P) system tries to hide the association between the identity of a participant and the data that it is interested in. This paper proposes a trust-based privacy-preservation method for P2P data sharing. It adopts the trust relation between a peer and its collaborators (buddies). The buddy works as a proxy to send the request and acquire the data. This provides a shield under which the identity of the requester and the accessed data cannot be linked. A privacy measuring method is presented to evaluate the proposed mechanism. Dynamic trust assessment and the enhancement to supplier's privacy are discussed

Abstract In ad hoc networks, malicious nodes can deploy wormhole attacks to fabricate a false scenario on the proximity relationship among mobile nodes. A classification of the attacks according to the format of the wormholes is proposed. This forms a basis to identify the detection capability of various approaches. An analysis shows that earlier approaches focus on the prevention of wormholes among neighbors that trust each other. As a more generic approach, we present an end‐to‐end scheme that can detect wormholes on a multi‐hop route. Only the trust between the source and the destination is assumed. The mechanism uses geographic information to detect anomalies in neighbor relations and node movements. To reduce the computation and storage overhead, we present a scheme called cell‐based open tunnel avoidance (COTA) to manage the information. COTA requires a constant space for every node on the path and the computation overhead increases linearly to the number of detection packets. We prove that the savings do not deteriorate the detection capability. Various schemes to control communication overhead are studied. The simulation and experiments on real devices show that the proposed mechanism can be combined with existent routing protocols to defend against wormhole attacks. Copyright © 2006 John Wiley & Sons, Ltd.

We investigate the performance issues of destination-sequenced distance vector (DSDV) and ad-hoc on-demand distance vector (AODV) routing protocols for mobile ad hoc networks. Four performance metrics are measured by varying the maximum speed of mobile hosts, the number of connections, and the network size. The correlation between network topology change and mobility is investigated by using linear regression analysis. The simulation results indicate that AODV outperforms DSDV in less stressful situations, while DSDV is more scalable with respect to the network size. It is observed that network congestion is the dominant reason for packet drop for both protocols. We propose a new routing protocol, congestion-aware distance vector (CADV), to address the congestion issues. CADV outperforms AODV in delivery ratio by about 5%, while introduces less protocol load. The result demonstrates that integrating congestion avoidance mechanisms with proactive routing protocols is a promising way to improve performance.

This paper compares the security properties of ad hoc on-demand distance vector (AODV) and destination sequence distance vector (DSDV) protocols, especially the difference caused by on-demand and proactive route queries. The on-demand route query enables the malicious host to conduct real time attacks on AODV. The communication overhead of attacks on DSDV is independent of the attack methods and the width of attack targets. A single false route propagates slower in AODV than in DSDV. The detection of false destination sequence in AODV heavily depends on the mobility of hosts. False distance vector and false destination sequence attacks are studied by simulation. The delivery ratio, communication overhead, and the propagation of false routes are measured by varying the traffic load and the maximum speed of host movement. The anomalous patterns of sequence numbers detected by destination hosts can be applied to detect the false destination sequence attacks.