摘要 Internet of Things (IoT) consists of devices, smart objects, and people that are given a unique identifier connected to the internet via wireless network. These devices are usually resources limited such as battery powered, memory and processing capability etc. The migration of these objects to the internet make them susceptible to various attacks related to external or internal unauthorized devices, which affect physical links, radio hops and routing discovery, give rise to malfunction and resources exhausting. The security mechanisms widely used on the Internet are too heavy to be integrated on small constrained objects. Therefore, we need secure protocols or add a security features to the existing ones with a minimum overhead. Routing also is one of the factors that may impact the security. In this paper, we will focus our work on the State of the art of existing protocols and study their vulnerabilities. Potential solutions based on the blockchain technology are also discussed in this paper.
摘要 A network traffic classifier (NTC) is an important part of current network monitoring systems, being its task to infer the network service that is currently used by a communication flow (e.g., HTTP and SIP). The detection is based on a number of features associated with the communication flow, for example, source and destination ports and bytes transmitted per packet. NTC is important, because much information about a current network flow can be learned and anticipated just by knowing its network service (required latency, traffic volume, and possible duration). This is of particular interest for the management and monitoring of Internet of Things (IoT) networks, where NTC will help to segregate traffic and behavior of heterogeneous devices and services. In this paper, we present a new technique for NTC based on a combination of deep learning models that can be used for IoT traffic. We show that a recurrent neural network (RNN) combined with a convolutional neural network (CNN) provides best detection results. The natural domain for a CNN, which is image processing, has been extended to NTC in an easy and natural way. We show that the proposed method provides better detection results than alternative algorithms without requiring any feature engineering, which is usual when applying other models. A complete study is presented on several architectures that integrate a CNN and an RNN, including the impact of the features chosen and the length of the network flows used for training.
摘要 Network traffic classification is used in various applications such as network traffic management, policy enforcement, and intrusion detection systems. Although most applications encrypt their network traffic and some of them dynamically change their port numbers, Machine Learning (ML) and especially Deep Learning (DL)-based classifiers have shown impressive performance in network traffic classification. In this article, we evaluate the robustness of DL-based network traffic classifiers against Adversarial Network Traffic (ANT). ANT causes DL-based network traffic classifiers to predict incorrectly using Universal Adversarial Perturbation (UAP) generating methods. Since there is no need to buffer network traffic before sending ANT, it is generated live. We partition the input space of the DL-based network traffic classification into three categories: packet classification, flow content classification, and flow time series classification. To generate ANT, we propose three new attacks injecting UAP into network traffic. AdvPad attack injects a UAP into the content of packets to evaluate the robustness of packet classifiers. AdvPay attack injects a UAP into the payload of a dummy packet to evaluate the robustness of flow content classifiers. AdvBurst attack injects a specific number of dummy packets with crafted statistical features based on a UAP into a selected burst of a flow to evaluate the robustness of flow time series classifiers. The results indicate injecting a little UAP into network traffic, highly decreases the performance of DL-based network traffic classifiers in all categories.
摘要 In this paper, we attempt to improve industrial safety from the perspective of communication security. We leverage the protocol fuzzing technology to reveal errors and vulnerabilities inside implementations of industrial network protocols(INPs). Traditionally, to effectively conduct protocol fuzzing, the test data has to be generated under the guidance of protocol grammar, which is built either by interpreting the protocol specifications or reverse engineering from network traces. In this study, we propose an automated test case generation method, in which the protocol grammar is learned by deep learning. Generative adversarial network(GAN) is employed to train a generative model over real-world protocol messages to enable us to learn the protocol grammar. Then we can use the trained generative model to produce fake but plausible messages, which are promising test cases. Based on this approach, we present an automatical and intelligent fuzzing framework(GANFuzz) for testing implementations of INPs. Compared to prior work, GANFuzz offers a new way for this problem. Moreover, GANFuzz does not rely on protocol specification, so that it can be applied to both public and proprietary protocols, which outperforms many previous frameworks. We use GANFuzz to test several simulators of the Modbus-TCP protocol and find some errors and vulnerabilities
Network traffic classification is an essential tool in cyber security for the recognition and interception of cyber insider threats. Traffic classification is the first step to distinguish various applications and protocols that are available in the network. The core component of network intrusion detection is a network traffic analysis that investigates the network behavior based on traffic characterization. Network traffic classification is the centre segment of the network traffic analysis and particularly for filtering traffic in order to identify any malicious activities within the network. Numerous techniques are proposed by researchers to date varying from traditional Port-based to Machine/Deep Learning techniques. Machine Learning emerges as a prominent solution for encrypted and real-time traffic classification; exploiting the statistical properties of the network flow. In this paper various Machine/Deep Learning techniques for network traffic classification are critically evaluated. The main purpose is to investigate different traffic classification approaches (supervised, semi-supervised, unsupervised) and provide summarization for a set of trends followed by various researchers for classifying network traffic
Reverse engineering of unknown network protocols based on recorded traffic traces enables security analyses and debugging of undocumented network services. In particular for binary protocols, existing approaches (1) lack comprehensive methods to classify or determine the data type of a discovered segment in a message, e.,g., a number, timestamp, or network address, that would allow for a semantic interpretation and (2) have strong assumptions that prevent analysis of lower-layer protocols often found in IoT or mobile systems. In this paper, we propose the first generic method for analyzing unknown messages from binary protocols to reveal the data types in message fields. To this end, we split messages into segments of bytes and use their vector interpretation to calculate similarities. These can be used to create clusters of segments with the same type and, moreover, to recognize specific data types based on the clusters' characteristics. Our extensive evaluation shows that our method provides precise classification in most cases and a data-type-recognition precision of up to 100% at reasonable recall, improving the state-of-the-art by a factor between 1.3 and 3.7 in realistic scenarios. We open-source our implementation to facilitate follow-up works.
Internet of Things (IoT) consists of devices, smart objects, and people that are given a unique identifier connected to the internet via wireless network. These devices are usually resources limited such as battery powered, memory and processing capability etc. The migration of these objects to the internet make them susceptible to various attacks related to external or internal unauthorized devices, which affect physical links, radio hops and routing discovery, give rise to malfunction and resources exhausting. The security mechanisms widely used on the Internet are too heavy to be integrated on small constrained objects. Therefore, we need secure protocols or add a security features to the existing ones with a minimum overhead. Routing also is one of the factors that may impact the security. In this paper, we will focus our work on the State of the art of existing protocols and study their vulnerabilities. Potential solutions based on the blockchain technology are also discussed in this paper.
Internet of Things (IoT) consists of devices, smart objects, and people that are given a unique identifier connected to the internet via wireless network. These devices are usually resources limited such as battery powered, memory and processing capability etc. The migration of these objects to the internet make them susceptible to various attacks related to external or internal unauthorized devices, which affect physical links, radio hops and routing discovery, give rise to malfunction and resources exhausting. The security mechanisms widely used on the Internet are too heavy to be integrated on small constrained objects. Therefore, we need secure protocols or add a security features to the existing ones with a minimum overhead. Routing also is one of the factors that may impact the security. In this paper, we will focus our work on the State of the art of existing protocols and study their vulnerabilities. Potential solutions based on the blockchain technology are also discussed in this paper.
The Internet of Things or “IoT” defines a highly interconnected network of heterogeneous devices where all kinds of communications seem to be possible, even unauthorized ones. As a result, the security requirement for such network becomes critical whilst common standard Internet security protocols are recognized as unusable in this type of networks, particularly due to some classes of IoT devices with constrained resources. The document discusses the applicability and limitations of existing IP-based Internet security protocols and other security protocols used in wireless sensor networks, which are potentially suitable in the context of IoT. The analysis of these protocols is discussed based on a taxonomy focusing on the key distribution mechanism
Encrypted traffic classification requires discriminative and robust traffic representation captured from content-invisible and imbalanced traffic data for accurate classification, which is challenging but indispensable to achieve network security and network management. The major limitation of existing solutions is that they highly rely on the deep features, which are overly dependent on data size and hard to generalize on unseen data. How to leverage the open-domain unlabeled traffic data to learn representation with strong generalization ability remains a key challenge. In this paper,we propose a new traffic representation model called Encrypted Traffic Bidirectional Encoder Representations from Transformer (ET-BERT), which pre-trains deep contextualized datagram-level representation from large-scale unlabeled data. The pre-trained model can be fine-tuned on a small number of task-specific labeled data and achieves state-of-the-art performance across five encrypted traffic classification tasks, remarkably pushing the F1 of ISCX-Tor to 99.2% (4.4% absolute improvement), ISCX-VPN-Service to 98.9% (5.2% absolute improvement), Cross-Platform (Android) to 92.5% (5.4% absolute improvement), CSTNET-TLS 1.3 to 97.4% (10.0% absolute improvement). Notably, we provide explanation of the empirically powerful pre-training model by analyzing the randomness of ciphers. It gives us insights in understanding the boundary of classification ability over encrypted traffic. The code is available at: this https URL
A network traffic classifier (NTC) is an important part of current network monitoring systems, being its task to infer the network service that is currently used by a communication flow (e.g., HTTP and SIP). The detection is based on a number of features associated with the communication flow, for example, source and destination ports and bytes transmitted per packet. NTC is important, because much information about a current network flow can be learned and anticipated just by knowing its network service (required latency, traffic volume, and possible duration). This is of particular interest for the management and monitoring of Internet of Things (IoT) networks, where NTC will help to segregate traffic and behavior of heterogeneous devices and services. In this paper, we present a new technique for NTC based on a combination of deep learning models that can be used for IoT traffic. We show that a recurrent neural network (RNN) combined with a convolutional neural network (CNN) provides best detection results. The natural domain for a CNN, which is image processing, has been extended to NTC in an easy and natural way. We show that the proposed method provides better detection results than alternative algorithms without requiring any feature engineering, which is usual when applying other models. A complete study is presented on several architectures that integrate a CNN and an RNN, including the impact of the features chosen and the length of the network flows used for training.
Monitoring network traffic to identify content, services, and applications is an active research topic in network traffic control systems. While modern firewalls provide the capability to decrypt packets, this is not appealing for privacy advocates. Hence, identifying any information from encrypted traffic is a challenging task. Nonetheless, previous work has identified machine learning methods that may enable application and service identification. The process involves high level feature extraction from network packet data then training a robust machine learning classifier for traffic identification. We propose a classification technique using an ensemble of deep learning architectures on packet, payload, and inter-arrival time sequences. To our knowledge, this is the first time such deep learning architectures have been applied to the Server Name Indication (SNI) classification problem. Our ensemble model beats the state of the art machine learning methods and our up-to-date model can be found on github: \url{this https URL}
Internet traffic classification has become more important with rapid growth of current Internet network and online applications. There have been numerous studies on this topic which have led to many different approaches. Most of these approaches use predefined features extracted by an expert in order to classify network traffic. In contrast, in this study, we propose a \emph{deep learning} based approach which integrates both feature extraction and classification phases into one system. Our proposed scheme, called "Deep Packet," can handle both \emph{traffic characterization} in which the network traffic is categorized into major classes (\eg, FTP and P2P) and application identification in which end-user applications (\eg, BitTorrent and Skype) identification is desired. Contrary to most of the current methods, Deep Packet can identify encrypted traffic and also distinguishes between VPN and non-VPN network traffic. After an initial pre-processing phase on data, packets are fed into Deep Packet framework that embeds stacked autoencoder and convolution neural network in order to classify network traffic. Deep packet with CNN as its classification model achieved recall of 0.98 in application identification task and 0.94 in traffic categorization task. To the best of our knowledge, Deep Packet outperforms all of the proposed classification methods on UNB ISCX VPN-nonVPN dataset.
Fuzzing (Fuzz testing) can effectively identify security vulnerabilities in software by providing a large amount of unexpected input to the target program. An important part of fuzzing test is the fuzzing data generation. Numerous traditional methods to generate fuzzing data have been developed, such as model-based fuzzing data generation and random fuzzing data generation. These techniques require the specification of the input data format or analyze the input data format by manual reverse engineering. In this paper, we introduce an approach using Wasserstein generative adversarial networks (WGANs), a deep adversarial learning method, to generate fuzzing data. This method does not require defining the input data format. To the best of our knowledge, this study is the first to use a WGAN-based method to generate fuzzing data. Industrial security has been an important and pressing issue globally. Network protocol fuzzing plays a significant role in ensuring the safety and reliability of industrial control systems (ICSs). Thus, the proposed method is significant for ICS testing. In the experiment, we use an industrial control protocol such as the Modbus-TCP protocol and EtherCAT protocol as our test target. Results indicate that this approach is more intelligent and capable than the methods used in previous studies. In addition, owing to its design, this model can be trained within a short time, which is computationally light and practical.
Network traffic classification is used in various applications such as network traffic management, policy enforcement, and intrusion detection systems. Although most applications encrypt their network traffic and some of them dynamically change their port numbers, Machine Learning (ML) and especially Deep Learning (DL)-based classifiers have shown impressive performance in network traffic classification. In this article, we evaluate the robustness of DL-based network traffic classifiers against Adversarial Network Traffic (ANT). ANT causes DL-based network traffic classifiers to predict incorrectly using Universal Adversarial Perturbation (UAP) generating methods. Since there is no need to buffer network traffic before sending ANT, it is generated live. We partition the input space of the DL-based network traffic classification into three categories: packet classification, flow content classification, and flow time series classification. To generate ANT, we propose three new attacks injecting UAP into network traffic. AdvPad attack injects a UAP into the content of packets to evaluate the robustness of packet classifiers. AdvPay attack injects a UAP into the payload of a dummy packet to evaluate the robustness of flow content classifiers. AdvBurst attack injects a specific number of dummy packets with crafted statistical features based on a UAP into a selected burst of a flow to evaluate the robustness of flow time series classifiers. The results indicate injecting a little UAP into network traffic, highly decreases the performance of DL-based network traffic classifiers in all categories.
摘要 Although the reliability of OCSP responders is extremely important for certificate revocation checks in PKIs, no standardized penetration testing solution for them exists yet. Therefore, the focus of this thesis is the search for OCSP's design and implementation related vulnerabilities. For this purpose a custom black-box fuzzer based on boofuzz is developed. It enables automatic testing of OCSP responders to reveal potential exploitable vulnerabilities using its main generation-based input creation model or its additional mutation-based model. With the fuzzer's monitoring tools the fuzzing run can be monitored and logs are produced for later analysis. This is successfully tested in a fuzzing campaign, which proofs the applicability of the custom fuzzer on arbitrary OCSP responders. The fuzzing campaign also shows, that the generation-based model is better applicable for a complex protocol like OCSP and it reveals potential exploitable implementation related security vulnerabilities within the audits created for the different targets.
摘要 Server fuzzing is difficult. Unlike simple command-line tools, servers feature a massive state space that can be traversed effectively only with well-defined sequences of input messages. Valid sequences are specified in a protocol. In this paper, we present AFLNET, the first greybox fuzzer for protocol implementations. Unlike existing protocol fuzzers, AFLNET takes a mutational approach and uses state-feedback to guide the fuzzing process. AFLNET is seeded with a corpus of recorded message exchanges between the server and an actual client. No protocol specification or message grammars are required. AFLNET acts as a client and replays variations of the original sequence of messages sent to the server and retains those variations that were effective at increasing the coverage of the code or state space. To identify the server states that are exercised by a message sequence, AFLNET uses the server's response codes. From this feedback, AFLNET identifies progressive regions in the state space, and systematically steers towards such regions. The case studies with AFLNET on two popular protocol implementations demonstrate a substantial performance boost over the state-of the-art. AFLNET discovered two new CVEs which are classified as critical (CVSS score CRITICAL 9.8).