Penetration testing, or pentesting, refers to assessing network system security by trying to identify and exploit any existing vulnerabilities. Reinforcement Learning (RL) has recently become an effective method for creating autonomous pentesting agents. However, RL agents are typically trained in a simulated network environment. This can be challenging when deploying them in a real network infrastructure due to the lack of realism of the simulation-trained agents. In this paper, we present PenGym, a framework for training pentesting RL agents in realistic network environments. The most significant features of PenGym are its support for real pentesting actions, full automation of the network environment creation, and good execution performance. The results of our experiments demonstrated the advantages and effectiveness of using PenGym as a realistic training environment in comparison with a simulation approach (NASim). For the largest scenario, agents trained in the original NASim environment behaved poorly when tested in a real environment, having a high failure rate. In contrast, agents trained in PenGym successfully reached the pentesting goal in all our trials. Even after fixing logical modeling issues in simulation to create the revised version NASim(rev.), experiment results with the largest scenario indicated that agents trained in PenGym slightly outperformed, and were more stable, than those trained in NASim(rev.). Thus, the average number of steps required to reach the pentesting goal was 1.4 to 8 steps better for PenGym. Consequently, PenGym provides a reliable and realistic training environment for pentesting RL agents, eliminating the need to model agent actions via simulation.

In the fourth industrial revolution, securing the protection of supply chains has become an ever-growing concern. One such cyber threat is a hardware Trojan (HT), a malicious modification to an IC. HTs are often identified during the hardware manufacturing process but should be removed earlier in the design process. Machine learning-based HT detection in gate-level netlists is an efficient approach to identifying HTs at the early stage. However, feature-based modeling has limitations in terms of discovering an appropriate set of HT features. We thus proposeNHTD-GLin this paper, a novel node-wise HT detection method based on graph learning (GL). Given the formal analysis of the HT features obtained from domain knowledge,NHTD-GLbridges the gap between graph representation learning and feature-based HT detection. The experimental results demonstrate thatNHTD-GLachieves 0.998 detection accuracy and 0.921 F1-score and outperforms state-of-the-art node-wise HT detection methods.NHTD-GLextracts HT features without heuristic feature engineering.

As security risks to network systems have grown, red team assessment has emerged as a powerful methodology for discovering vulnerabilities.Such assessments are difficult to master because technical knowledge and experience are needed.Automating the vulnerability assessment of network systems is expected to help network system administrators conduct these assessments easily.The challenges for automating these assessments include accurately addressing many actions, observing network states, and generalizing agent models.In this paper, we propose a framework, called AutoRed, for the automation of red team assessment via strategic thinking using reinforcement learning (RL).Our framework addresses the following challenges: (1) facilitating action determination by adopting a hierarchical RL model via strategic thinking, (2) establishing a method to observe network systems using graph neural networks (GNNs), and (3) investigating the reusability and generalization ability of the proposed model through experiments.We further evaluate the proposed model in an emulated environment constructed on a virtual machine platform.The experimental results demonstrate that the proposed model trained on three scenarios simultaneously can be applied 10-40 times more efficiently to various scenarios, including unseen scenarios during training, than the state-of-the-art hierarchical model.

Large language models (LLMs) have recently been employed for a wide variety of purposes. Retrieval-augmented generation (RAG), in which an LLM generates a response based on context relevant to the prompt, is often used to enable the LLM to adapt to specialized domains. However, sentences generated by a generative LLM may contain incorrect information, known as “hallucinations.” The challenge in identifying hallucinations within the RAG framework involves evaluating the certainty of both context retrieval and LLM outputs. In this paper, we propose a metric called RAG certainty to quantify the certainty of LLM outputs within a RAG framework. The proposed metric is calculated based on certainty scores from both information retrieval and response generation. Experimental results demonstrate that the proposed metric effectively reflects the certainty of information retrieval in a RAG framework. We further validated the proposed metric through a case study that assesses the predicted Common Vulnerability Scoring Sys-tem (CVSS) scores for cybersecurity vulnerabilities and found that errors are mitigated according to the proposed metric.

Graph neural networks (GNNs) have been actively employed in hardware security and have demonstrated remarkable performance. In particular, GNN models for hardware Trojan (HT) detection significantly outperform existing machine learning-based detection methods. However, GNNs have a potential vulnerability to membership inference attack (MIA), which aims to determine whether a given sample is used in the training dataset. In this paper, we investigate the threat of MIAs for GNN-based HT detection models. First, the MIA scheme for GNN-based HT detection models is established based on the basic MIA settings. The experimental results demonstrate that MIA for GNN-based HT detection can leak information about the HTs included in the training dataset with a 0.945 attack AUC score in the worst-case scenario. Based on this observation, we propose a defense method against MIA utilizing a domain generalization technique. The proposed defense method successfully mitigated the vulnerability of MIA and degraded the attack AUC score to 0.536 for the netlist level while maintaining the original HT detection performance.

Presents the front cover, title page, cover page, or splash screen of the proceedings record.

Puncturable encryption (PE), introduced by Green and Miers at IEEE S$&$P 2015, allows recipients to update their decryption keys to revoke decryption capability for selected messages without communicating with senders. In general, it allows users to control which ciphertexts their keys may decrypt. The notion of PE has been found very useful in many applications, such as asynchronous messaging systems, group messaging systems, public-key watermarking schemes, secure cloud emails, and many more. In this paper, we introduce a new primitive called hierarchical identity-based puncturable encryption (HIBPE) that enhances the concept of PE by allowing more general key delegation and flexible key puncture. It enhances the capability of the data owner for multi-level encrypted data sharing within a group of users by delegating the decryption keys of the users in higher-levels to generate decryption keys for the users in lower-levels. Moreover, it allows users to puncture (update) their decryption keys on tags so that a decryption key punctured on a tag can no longer decrypt ciphertexts under this tag. In addition, to control access to the users' data, the higher-level users can further puncture the delegated keys (for lower-level users) with some tags such that the part of the owner's data is labeled by the punctured tags will no longer be accessible by the lower-level users. These features offer an efficient and flexible solution for encrypted data sharing as well as data-access control mechanisms in a hierarchical setting. We propose the formal definition and security model for HIBPE schemes and provide a concrete HIBPE scheme based on the hardness of the learning with errors problem in the standard model. Further, we provide a generic construction of forward secure hierarchical identity-based encryption (fs-HIBE) from HIBPE, which enables the first quantum-safe construction of fs-HIBE in the standard model. Moreover, this is the first fs-HIBE construction by exploring the concept of PE. The proposed fs-HIBE provides quantum-safe protection for secret keys from exposure in multi-level encrypted data sharing by evolving the keys with time.

Recently, with the increase in outsourcing of IC design and manufacturing, the possibility of inserting hardware Trojans, which are circuits with malicious functions, has been pointed out. To prevent this threat, a method to identify hardware Trojans using neural networks has been proposed. On the other hand, adversarial attacks have emerged that modify circuit design information to reduce the accuracy of hardware-Trojan classification by neural networks. Since the features designed by existing methods do not take the attacks into account, it is necessary to consider a new method for countermeasures. In this paper, out of 76 features that are strongly related to hardware-Trojan features, we investigate them from the viewpoint of the robustness against the adversarial attacks on circuit design information and newly propose 24 hardware-Trojan features. We compare the classifiers using the proposed 24 features with the classifiers using 11, 36, 51, and 76 existing features, respectively and confirm that the proposed ones are more robust in identifying hardware Trojans in circuits subjected to the adversarial attacks.

Residue number systems provide efficient techniques for speeding up calculations and/or protecting against side channel attacks when used in the context of cryptographic engineering. One of the interests of such systems is their scalability, as the existence of large bases for some specialized systems is often an open question. In this paper, we present highly optimized methods for generating large bases for residue number systems and, in some cases, the largest possible bases. We show their efficiency by demonstrating their improvement over the state-of-the-art bases reported in the literature. This work make it possible to address the problem of the scalability issue of finding new bases for a specific system that arises whenever a parameter changes, and possibly open new application avenues.

Abstract The lattice‐based encryption scheme has high efficiency and reliability, and it can be run on small devices with limited memory capacity and computational resources such as sensor nodes or smart cards. The first implementation is presented of the original ring–learning‐with‐errors‐based encryption scheme on a standard Java Card platform by combining the number theoretic transform with improved Montgomery modular multiplication. Without any cryptographic coprocessor support, the decryption running time is around 7 s, corresponding to the AES‐128 security level. Two efficient discrete Gaussian sampling approaches, known at the discrete Ziggurat sampling algorithm and Knuth–Yao algorithm, were implemented on the Java Card and resulted in a reduction in running times. More important, polynomial modular multiplication is shown to perform efficiently on a standard Java Card platform even when the big integers and floating‐point number operations are not supported. The results show the feasibility of implementing more lattice‐based cryptosystems on existing memory‐constrained Java Cards. A preliminary version of this paper appeared with the title ‘Memory‐constrained implementation of lattice‐based encryption scheme on standard Java Card’ in Proceedings of the 2017 IEEE International Symposium on Hardware Oriented Security and Trust (HOST) .

Supply chain management is critical in industrial efforts to reduce costs and time, stabilize product supply, and improve profitability. Surveillance cameras play a significant role in supply chain management, as they can record work activities to achieve appropriate monitoring of company operations. A solution for ensuring the reliability of the video taken by a surveillance camera is to achieve data integrity using a cryptographic algorithm. Another important solution is the identification of a surveillance camera to ensure the traceability of video. This paper proposes a novel approach for video-based fingerprint extraction and key generation, that can be used for camera PUF construction. Our experiment shows that a 256-bit key can be extracted from 50 frames of a normal video with a resolution of 800 X 600 pixels, and we theoretically prove that our methodology satisfies the randomness, uniqueness, steadiness, and unpredictability requirements. Our PUF construction can thus be used to identify a surveillance camera from a video.

The Internet of things (IoT) has come into widespread use, and data protection and integrity are critical for connected IoT devices in order to maintain security and privacy. Low-power wide-area (LPWA) technologies for IoT wireless communication achieve data protection and integrity by using encryption and message authentication. However, side-channel analysis techniques exist that have the capacity to recover secret information from a device. In this paper, we apply a side-channel analysis technique to the payload encryption process and message authentication code generation process on a real LoRaWAN end-device. The entire AES-128 key for the payload encryption can be recovered with 260 electromagnetic(EM)-leakage traces and 12 bytes of the key for message authentication code generation can be recovered with 140 EM-leakage traces.

Quantum computers have the potential to solve some difficult mathematical problems efficiently and thus will inevitably exert a more significant impact on the traditional asymmetric cryptography. The National Institute of Standards and Technology (NIST) has opened a formal call for the submission of proposals of quantum-resistant public-key cryptographic algorithms to set the next-generation cryptography standards. Compared to powerful machines with ample amount of hardware resources such as racks of servers and IoT devices, including the massive number of microcontrollers, smart terminals, and sensor nodes with limited computing capacity, should also have some postquantum cryptography features for security and privacy. To ensure the correct execution of encryption algorithms on any platforms, the portability of implementation becomes more important. As distinguished from C/C++, JavaScript is a popular cross-platform language that can be used for the web applications and some hardware platforms directly, and it could be one of the solutions of portability. Therefore, we investigate and implement several recent lattice-based encryption schemes and public-key exchange protocols including Lizard, ring-Lizard, Kyber, Frodo, and NewHope in JavaScript, which are the active candidates of postquantum cryptography due to their applicabilities and efficiencies. We show and compare the performance of our JavaScript implementation on web browsers, embedded device Tessel2, Android phone, and several JavaScript-enabled platforms on PC and Mac. Our work shows that implementing lattice-based cryptography on JavaScript-enabled platforms is achievable and results in desirable portability.

The Physically Unclonable Function (PUF) is a technique that generates unique device identifiers based on variations in the manufacturing process. The Internet of Things (IoT) has become widespread, and various kinds of devices are now available. Device authentication and key management are essential to provide a secure service to these devices. We can use the unforgeable identifier generated by the PUF as a key for encryption and authentication. However, the existing PUFs require a dedicated hardware or low-level software, i.e., driver. Thus, they are impractical to use on smartphones or IoT devices due to the severe limitations of production cost and power consumption. In this paper, we propose a sensor-based PUF that utilizes the accelerometer and gyroscope, which are widely available on smartphones and IoT devices. We implement the proposed PUF on a smartwatch and show that accelerometer-based PUF achieves good usability, extreme robustness, and a high entropy of 91.66 bits.

Copyright protection is a major issue in online content distribution services and many key-management schemes have been proposed for protecting content. Tree-based schemes aim at reducing the load on the server, and do not give consideration to that on clients. This scheme is not fully suitable for devices with low computational capacity. On the other hand, the load on clients is low in a star-based scheme. However, the load on the server becomes large in proportion to the number of clients. This structure is far from scalable. We propose a key-management scheme that is the intermediate scheme of a star-based scheme and a tree-based scheme. Then, we evaluate both the load on the server and clients in key-management scheme using the three structures. We find that the load on server and that on clients in our scheme have a relation of trade-off. We can construct optimal key-management structure satisfying system requirements using our scheme. Furthermore, the loads on both the server and clients in our scheme are lower than tree-based structure scheme under a certain condition.