As a critical component of the security architecture of 5G network, the authentication protocol plays a role of the first safeguard in ensuring the communication security, such as the confidentiality of user data. EAP-TLS is one of such protocols being defined in the 5G standards to provide key services in the specific IoT circumstances. This protocol is currently under the process of standardization, and it is vital to guarantee that the standardized protocol is free from any design flaws, which may result in severe vulnerabilities and serious consequences when implemented in real systems. However, it is still unclear whether the proposed 5G EAP-TLS authentication protocol provides the claimed security guarantees. To fill this gap, we present in this work a comprehensive formal analysis of the security related properties of the 5G EAP-TLS authentication protocol based on the symbolic model checking approach. Specifically, we build the first formal model of the 5G EAP-TLS authentication protocol in the applied pi calculus, and perform an automated security analysis of the formal protocol model by using the ProVerif model checker. Our analysis results show that there are some subtle flaws in the current protocol design that may compromise the claimed security objectives. To this end, we also propose and verify a possible fix that is able to mitigate these flaws. To the best of our knowledge, this is the first thorough formal analysis of the 5G EAP-TLS authentication protocol.

Unmanned aerial vehicles (UAVs) are vulnerable to network attacks. Designing an effective intrusion detection system (IDS) for UAVs is crucial. However, UAVs have limited computing resources and need to deal with massive amounts of network data, which further increases the difficulty of detection. Moreover, most existing IDSs have large parameters. In this study, we develop a tiny machine learning-based IDS to solve the above issue. We first establish an improved fuzzy rough set (FRS) model based on adaptive neighborhoods. Then, using the proposed FRS model, we employ a feature selection (FS) method to select optimal features and reduce the overall computational cost of the IDS. Furthermore, we proposed a tiny intrusion detection model that attains high-precision detection via shallow deep learning. Additionally, the proposed method can address intrusion detection problems in scenarios with partial data missing. According to the evaluations, the proposed method can effectively address intrusion detection in UAVs.

We generate correlated photon pairs in an all-pass silicon microring resonator using continuous-wave pumped spontaneous four-wave mixing and we characterize in detail the optical bistability in the nonclassical single-photon regime. Special attention is given to the resonance-locked method by balancing the pump-induced heating and the active off-chip cooling, leading to stable operation of the photon-pair source. The maximal coincidence rate of $20.53\pm 0.34\mathrm{Hz}$, maximal coincidence-to-accidental ratio of $654\pm 125$, and minimal zero-delayed heralded second-order correlation of $0.14\pm 0.09$ are achieved. The wavelengths of photon pairs remain unchanged at different pump power, indicating that our source is more compatible with the common communication systems than those using conventional on-resonance strategies, that tunes the pump wavelength. This work not only experimentally validates the nonclassical bistability behaviors, but also puts forward a simple method of pump-resonance matching for high-quality performance.

Software-Defined Networking (SDN) enhances the flexibility and programmability of networks by separating control plane and data plane. The logically centralized control mechanism makes the control plane vulnerable in both single and multiple controller scenarios. Malicious third parties can exploit vulnerabilities of reactive forwarding mode to launch distributed denial-of-service (DDoS) attacks against SDN controllers. Unfortunately, existing DoS/DDoS solutions under single controller can not afford effective performance under multiple controllers due to the absence of cooperative detection and mitigation. To solve the above problem, we propose a blockchain-based SDN-targeted DDoS defense framework (BSD-Guard) that can provide cooperative detection and mitigation mechanism to protect SDN controllers. BSD-Guard introduces a blockchain-based secure middle plane between control plane and data plane. The secure middle plane calculates the suspect rate of new flows based on the collected packets’ information and reports suspect lists to blockchain for immutably storing and sharing. Besides, the smart contract deployed on blockchain in advance constitutes collaborative defense strategies based on the suspect lists reported from multiple SDN domains. When receiving defense strategies, the secure middle plane converts them to specific flow table actions and installs actions into relevant switches. The experimental results indicate that BSD-Guard can efficiently detect DoS/DDoS attacks in multiple controllers scenario and issue precise defensive strategies near the source of attack by identifying the attack path.

ChunLiu ChunLiu, Hongguang Zhang, Kainan Zhao, Xinghai Ju, Lin Yang. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Intrusion detection systems are a crucial technique for securing the Internet of Things (IoT) from malicious attacks. Additionally, due to the continuous emergence of new vulnerabilities and unknown attack types, only a small number of attack samples in the IoT environments can be captured for analysis. In this work, we introduce an anchor graph clustering (AGC) method for intrusion detection to address the challenge of limited labeled samples in the IoT environments. AGC initially transforms the raw data into the embedding space to obtain more representative anchors. Then, AGC unifies anchor graph construction, anchor graph learning, and graph clustering into a unified framework, solving the resulting optimization problem through an iterative solution algorithm. Finally, AGC leverages the powerful analytical capabilities of graph learning to achieve fine-grained classification of low-quality labels. Experimental results on both real and synthetic datasets confirm that AGC can identify intrusions with high precision, while also being time-efficient in detection.