Biblio

SQL Injection continues to be one of the most damaging security exploits in terms of personal information exposure as well as monetary loss. Injection attacks are the number one vulnerability in the most recent OWASP Top 10 report, and the number of these attacks continues to increase. Traditional defense strategies often involve static, signature-based IDS (Intrusion Detection System) rules which are mostly effective only against previously observed attacks but not unknown, or zero-day, attacks. Much current research involves the use of machine learning techniques, which are able to detect unknown attacks, but depending on the algorithm can be costly in terms of performance. In addition, most current intrusion detection strategies involve collection of traffic coming into the web application either from a network device or from the web application host, while other strategies collect data from the database server logs. In this project, we are collecting traffic from two points: at the web application host, and at a Datiphy appliance node located between the webapp host and the associated MySQL database server. In our analysis of these two datasets, and another dataset that is correlated between the two, we have been able to demonstrate that accuracy obtained with the correlated dataset using algorithms such as rule-based and decision tree are nearly the same as those with a neural network algorithm, but with greatly improved performance.

Security issues are crucial in a number of machine learning applications, especially in scenarios dealing with human activity rather than natural phenomena (e.g., information ranking, spam detection, malware detection, etc.). In such cases, learning algorithms may have to cope with manipulated data aimed at hampering decision making. Although some previous work addressed the issue of handling malicious data in the context of supervised learning, very little is known about the behavior of anomaly detection methods in such scenarios. In this contribution, we analyze the performance of a particular method–online centroid anomaly detection–in the presence of adversarial noise. Our analysis addresses the following security-related issues: formalization of learning and attack processes, derivation of an optimal attack, and analysis of attack efficiency and limitations. We derive bounds on the effectiveness of a poisoning attack against centroid anomaly detection under different conditions: attacker's full or limited control over the traffic and bounded false positive rate. Our bounds show that whereas a poisoning attack can be effectively staged in the unconstrained case, it can be made arbitrarily difficult (a strict upper bound on the attacker's gain) if external constraints are properly used. Our experimental evaluation, carried out on real traces of HTTP and exploit traffic, confirms the tightness of our theoretical bounds and the practicality of our protection mechanisms.

In Industrial Control Systems (ICS/SCADA), machine to machine data traffic is highly periodic. Past work showed that in many cases, it is possible to model the traffic between each individual Programmable Logic Controller (PLC) and the SCADA server by a cyclic Deterministic Finite Automaton (DFA), and to use the model to detect anomalies in the traffic. However, a recent analysis of network traffic in a water facility in the U.S, showed that cyclic-DFA models have limitations. In our research, we examine the same data corpus; our study shows that the communication on all of the channels in the network is done in bursts of packets, and that the bursts have semantic meaning---the order within a burst depends on the messages. Using these observations, we suggest a new burst-DFA model that fits the data much better than previous work. Our model treats the traffic on each channel as a series of bursts, and matches each burst to the DFA, taking the burst's beginning and end into account. Our burst-DFA model successfully explains between 95% and 99% of the packets in the data-corpus, and goes a long way toward the construction of a practical anomaly detection system.

The continuous advance in recent cloud-based computer networks has generated a number of security challenges associated with intrusions in network systems. With the exponential increase in the volume of network traffic data, involvement of humans in such detection systems is time consuming and a non-trivial problem. Secondly, network traffic data tends to be highly dimensional, comprising of numerous features and attributes, making classification challenging and thus susceptible to the curse of dimensionality problem. Given such scenarios, the need arises for dimensional reduction, feature selection, combined with machine-learning techniques in the classification of such data. Therefore, as a contribution, this paper seeks to employ data mining techniques in a cloud-based environment, by selecting appropriate attributes and features with the least importance in terms of weight for the classification. Often the standard is to select features with better weights while ignoring those with least weights. In this study, we seek to find out if we can make prediction using those features with least weights. The motivation is that adversaries use stealth to hide their activities from the obvious. The question then is, can we predict any stealth activity of an adversary using the least observed attributes? In this particular study, we employ information gain to select attributes with the lowest weights and then apply machine learning to classify if a combination, in this case, of both source and destination ports are attacked or not. The motivation of this investigation is if attributes that are of least importance can be used to predict if an attack could occur. Our preliminary results show that even when the source and destination port attributes are used in combination with features with the least weights, it is possible to classify such network traffic data and predict if an attack will occur or not.

In the paper, we demonstrate a neuromorphic cognitive computing approach for Network Intrusion Detection System (IDS) for cyber security using Deep Learning (DL). The algorithmic power of DL has been merged with fast and extremely power efficient neuromorphic processors for cyber security. In this implementation, the data has been numerical encoded to train with un-supervised deep learning techniques called Auto Encoder (AE) in the training phase. The generated weights of AE are used as initial weights for the supervised training phase using neural networks. The final weights are converted to discrete values using Discrete Vector Factorization (DVF) for generating crossbar weight, synaptic weights, and thresholds for neurons. Finally, the generated crossbar weights, synaptic weights, threshold, and leak values are mapped to crossbars and neurons. In the testing phase, the encoded test samples are converted to spiking form by using hybrid encoding technique. The model has been deployed and tested on the IBM Neurosynaptic Core Simulator (NSCS) and on actual IBM TrueNorth neurosynaptic chip. The experimental results show around 90.12% accuracy for network intrusion detection for cyber security on the physical neuromorphic chip. Furthermore, we have investigated the proposed system not only for detection of malicious packets but also for classifying specific types of attacks and achieved 81.31% recognition accuracy. The neuromorphic implementation provides incredible detection and classification accuracy for network intrusion detection with extremely low power.

The Internet of Things (IoT) revolution has brought millions of small, low-cost, connected devices into our homes, cities, infrastructure, and more. However, these devices are often plagued by security vulnerabilities that pose threats to user privacy or can threaten the Internet architecture as a whole. Home networks can be particularly vulnerable to these threats as they typically have no network administrator and often contain unpatched or otherwise vulnerable devices. In this paper, we argue that the unique security challenges of home networks require a new network-layer architecture to both protect against external threats and mitigate attacks from compromised devices. We present initial findings based on traffic analysis from a small-scale IoT testbed toward identifying predictable patterns in IoT traffic that may allow construction of a policy-based framework to restrict malicious traffic. Based on our observations, we discuss key features for the design of this architecture to promote future developments in network-layer security in smart home networks.

This work concerns distributed consensus algorithms and application to a network intrusion detection system (NIDS) [21]. We consider the problem of defending the system against multiple data falsification attacks (Byzantine attacks), a vulnerability of distributed peer-to-peer consensus algorithms that has not been widely addressed in its practicality. We consider both naive (independent) and colluding attackers. We test three defense strategy implementations, two classified as outlier detection methods and one reputation-based method. We have narrowed our attention to outlier and reputation-based methods because they are relatively light computationally speaking. We have left out control theoretic methods which are likely the most effective methods, however their computational cost increase rapidly with the number of attackers. We compare the efficiency of these three implementations for their computational cost, detection performance, convergence behavior and possible impacts on the intrusion detection accuracy of the NIDS. Tests are performed based on simulations of distributed denial of service attacks using the KSL-KDD data set.

With the rapid application of the network based communication in industries, the security related problems appear to be inevitable for automation networks. The integration of internet into the automation plant benefited companies and engineers a lot and on the other side paved ways to number of threats. An attack on such control critical infrastructure may endangers people's health and safety, damage industrial facilities and produce financial loss. One of the approach to secure the network in automation is the development of an efficient Network based Intrusion Detection System (NIDS). Despite several techniques available for intrusion detection, they still lag in identifying the possible attacks or novel attacks on network efficiently. In this paper, we evaluate the performance of detection mechanism by combining the deep learning techniques with the machine learning techniques for the development of Intrusion Detection System (IDS). The performance metrics such as precession, recall and F-Measure were measured.

Link quality protocols employ link quality estimators to collect statistics on the wireless link either independently or cooperatively among the sensor nodes. Furthermore, link quality routing protocols for wireless sensor networks may modify an estimator to meet their needs. Link quality estimators are vulnerable against malicious attacks that can exploit them. A malicious node may share false information with its neighboring sensor nodes to affect the computations of their estimation. Consequently, malicious node may behave maliciously such that its neighbors gather incorrect statistics about their wireless links. This paper aims to detect malicious nodes that manipulate the link quality estimator of the routing protocol. In order to accomplish this task, MINTROUTE and CTP routing protocols are selected and updated with intrusion detection schemes (IDSs) for further investigations with other factors. It is proved that these two routing protocols under scrutiny possess inherent susceptibilities, that are capable of interrupting the link quality calculations. Malicious nodes that abuse such vulnerabilities can be registered through operational detection mechanisms. The overall performance of the new LQR protocol with IDSs features is experimented, validated and represented via the detection rates and false alarm rates.

As the frequency, severity, and sophistication of cyber attacks increase, along with our dependence on reliable computing infrastructure, the role of Intrusion Detection Systems (IDS) gaining importance. One of the challenges in deploying an IDS stems from selecting a combination of detectors that are relevant and accurate for the environment where security is being considered. In this work, we propose a new measurement approach to address two key obstacles: the base-rate fallacy, and the unit of analysis problem. Our key contribution is to utilize the notion of a `signal', an indicator of an event that is observable to an IDS, as the measurement target, and apply the multiple instance paradigm (from machine learning) to enable cross-comparable measures regardless of the unit of analysis. To support our approach, we present a detailed case study and provide empirical examples of the effectiveness of both the model and measure by demonstrating the automated construction, optimization, and correlation of signals from different domains of observation (e.g. network based, host based, application based) and using different IDS techniques (signature based, anomaly based).

The security and typical attack behavior of Modbus/TCP industrial network communication protocol are analyzed. The data feature of traffic flow is extracted through the operation mode of the depth analysis abnormal behavior, and the intrusion detection method based on the support vector machine (SVM) is designed. The method analyzes the data characteristics of abnormal communication behavior, and constructs the feature input structure and detection system based on SVM algorithm by using the direct behavior feature selection and abnormal behavior pattern feature construction. The experimental results show that the method can effectively improve the detection rate of abnormal behavior, and enhance the safety protection function of industrial network.

A honeypot is a deception tool for enticing attackers to make efforts to compromise the electronic information systems of an organization. A honeypot can serve as an advanced security surveillance tool for use in minimizing the risks of attacks on information technology systems and networks. Honeypots are useful for providing valuable insights into potential system security loopholes. The current research investigated the effectiveness of the use of centralized system management technologies called Puppet and Virtual Machines in the implementation automated honeypots for intrusion detection, correction and prevention. A centralized logging system was used to collect information of the source address, country and timestamp of intrusions by attackers. The unique contributions of this research include: a demonstration how open source technologies is used to dynamically add or modify hacking incidences in a high-interaction honeynet system; a presentation of strategies for making honeypots more attractive for hackers to spend more time to provide hacking evidences; and an exhibition of algorithms for system and network intrusion prevention.

Consensus algorithms provide strategies to solve problems in a distributed system with the added constraint that data can only be shared between adjacent computing nodes. We find these algorithms in applications for wireless and sensor networks, spectrum sensing for cognitive radio, even for some IoT services. However, consensus-based applications are not resilient to compromised nodes sending falsified data to their neighbors, i.e. they can be the target of Byzantine attacks. Several solutions have been proposed in the literature inspired from reputation based systems, outlier detection or model-based fault detection techniques in process control. We have reviewed some of these solutions, and propose two mitigation techniques to protect the consensus-based Network Intrusion Detection System in [1]. We analyze several implementation issues such as computational overhead, fine tuning of the solution parameters, impacts on the convergence of the consensus phase, accuracy of the intrusion detection system.

SCADA security is an increasingly important research area as these systems, used for process control and automation, are being exposed to the Internet due to their use of TCP/IP protocols as a transport mechanism for control messages. Most of the existing research work on SCADA systems has focused on addressing SCADA security by monitoring attacks or anomalies at the network level. The main issue affecting these systems today is that by focusing our attention on network-level monitoring needs, security practitioners may remain unaware of process level constraints. The proposed framework helps ensure that a mechanism is in place to help map process level constraints, as described by process engineers, to network level monitoring needs. Existing solutions have tried to address this problem but have not been able to fully bridge the gap between the process and the network. The goal of this research is to provide a solution that (i) leverages the knowledge process engineers have about the system (to help strengthen cyber security) and that has the ability to (ii) seamlessly monitors process constraints at the network level using standard network security tools. A prototype system for the Modbus TCP protocol and the Bro IDS has been built to validate the approach.

Intrusion detection systems need to be both accurate and fast. Speed is important especially when operating at the network level. Additionally, many intrusion detection systems rely on signature based detection approaches. However, machine learning can also be helpful for intrusion detection. One key challenge when using machine learning, aside from the detection accuracy, is using machine learning algorithms that are fast. In this paper, several processing architectures are considered for use in machine learning based intrusion detection systems. These architectures include standard CPUs, GPUs, and cognitive processors. Results of their processing speeds are compared and discussed.

Traffic of Industrial Control System (ICS) between the Human Machine Interface (HMI) and the Programmable Logic Controller (PLC) is highly periodic. However, it is sometimes multiplexed, due to multi-threaded scheduling. In previous work we introduced a Statechart model which includes multiple Deterministic Finite Automata (DFA), one per cyclic pattern. We demonstrated that Statechart-based anomaly detection is highly effective on multiplexed cyclic traffic when the individual cyclic patterns are known. The challenge is to construct the Statechart, by unsupervised learning, from a captured trace of the multiplexed traffic, especially when the same symbols (ICS messages) can appear in multiple cycles, or multiple times in a cycle. Previously we suggested a combinatorial approach for the Statechart construction, based on Euler cycles in the Discrete Time Markov Chain (DTMC) graph of the trace. This combinatorial approach worked well in simple scenarios, but produced a false-alarm rate that was excessive on more complex multiplexed traffic. In this paper we suggest a new Statechart construction method, based on spectral analysis. We use the Fourier transform to identify the dominant periods in the trace. Our algorithm then associates a set of symbols with each dominant period, identifies the order of the symbols within each period, and creates the cyclic DFAs and the Statechart. We evaluated our solution on long traces from two production ICS: one using the Siemens S7-0x72 protocol and the other using Modbus. We also stress-tested our algorithms on a collection of synthetically-generated traces that simulate multiplexed ICS traces with varying levels of symbol uniqueness and time overlap. The resulting Statecharts model the traces with an overall median false-alarm rate as low as 0.16% on the synthetic datasets, and with zero false-alarms on production S7-0x72 traffic. Moreover, the spectral analysis Statecharts consistently out-performed the previous combinatorial Statecharts, exhibiting significantly lower false alarm rates and more compact model sizes.

Detection of previously unknown attacks and malicious messages is a challenging problem faced by modern network intrusion detection systems. Anomaly-based solutions, despite being able to detect unknown attacks, have not been used often in practice due to their high false positive rate, and because they provide little actionable information to the security officer in case of an alert. In this paper we focus on intrusion detection in industrial control systems networks and we propose an innovative, practical and semantics-aware framework for anomaly detection. The network communication model and alerts generated by our framework are userunderstandable, making them much easier to manage. At the same time the framework exhibits an excellent tradeoff between detection rate and false positive rate, which we show by comparing it with two existing payload-based anomaly detection methods on several ICS datasets.

With cyber-physical systems opening to the outside world, security can no longer be considered a secondary issue. One of the key aspects in security of cyber-phyiscal systems is to deal with intrusions. In this paper, we highlight the several unique properties of control applications in cyber-physical systems. Using these unique properties, we propose a systematic intrusion-damage assessment and mitigation mechanism for the class of observable and controllable attacks. On the one hand, in cyber-physical systems, the plants follow certain laws of physics and this can be utilized to address the intrusion-damage assessment problem. That is, the states of the controlled plant should follow those expected according to the physics of the system and any major discrepancy is potentially an indication of intrusion. Here, we use a machine learning algorithm to capture the normal behavior of the system according to its dynamics. On the other hand, the control performance strongly depends on the amount of allocated resources and this can be used to address the intrusion-damage mitigation problem. That is, the intrusion-damage mitigation is based on the idea of allocating more resources to the control application under attack. This is done using a feedback-based approach including a convex optimization.

This paper introduces a novel graph-analytic approach for detecting anomalies in network flow data called GraphPrints. Building on foundational network-mining techniques, our method represents time slices of traffic as a graph, then counts graphlets–-small induced subgraphs that describe local topology. By performing outlier detection on the sequence of graphlet counts, anomalous intervals of traffic are identified, and furthermore, individual IPs experiencing abnormal behavior are singled-out. Initial testing of GraphPrints is performed on real network data with an implanted anomaly. Evaluation shows false positive rates bounded by 2.84% at the time-interval level, and 0.05% at the IP-level with 100% true positive rates at both.

The emergence of new network applications, such as the network intrusion detection system and packet-level accounting, requires packet classification to report all matched rules instead of only the best matched rule. Although several schemes have been proposed recently to address the multimatch packet classification problem, most of them require either huge memory or expensive ternary content addressable memory (TCAM) to store the intermediate data structure, or they suffer from steep performance degradation under certain types of classifiers. In this paper, we decompose the operation of multimatch packet classification from the complicated multidimensional search to several single-dimensional searches, and present an asynchronous pipeline architecture based on a signature tree structure to combine the intermediate results returned from single-dimensional searches. By spreading edges of the signature tree across multiple hash tables at different stages, the pipeline can achieve a high throughput via the interstage parallel access to hash tables. To exploit further intrastage parallelism, two edge-grouping algorithms are designed to evenly divide the edges associated with each stage into multiple work-conserving hash tables. To avoid collisions involved in hash table lookup, a hybrid perfect hash table construction scheme is proposed. Extensive simulation using realistic classifiers and traffic traces shows that the proposed pipeline architecture outperforms HyperCuts and B2PC schemes in classification speed by at least one order of magnitude, while having a similar storage requirement. Particularly, with different types of classifiers of 4K rules, the proposed pipeline architecture is able to achieve a throughput between 26.8 and 93.1 Gb/s using perfect hash tables.

Network traffic is a rich source of information for security monitoring. However the increasing volume of data to treat raises issues, rendering holistic analysis of network traffic difficult. In this paper we propose a solution to cope with the tremendous amount of data to analyse for security monitoring perspectives. We introduce an architecture dedicated to security monitoring of local enterprise networks. The application domain of such a system is mainly network intrusion detection and prevention, but can be used as well for forensic analysis. This architecture integrates two systems, one dedicated to scalable distributed data storage and management and the other dedicated to data exploitation. DNS data, NetFlow records, HTTP traffic and honeypot data are mined and correlated in a distributed system that leverages state of the art big data solution. Data correlation schemes are proposed and their performance are evaluated against several well-known big data framework including Hadoop and Spark.