Rongxing Lu’s research while affiliated with University of New Brunswick and other places

Highlights What are the main findings? Simplification of several complex theoretical attacks on DNP3 shows they can be executed with simpler methods, such as bypassing IP allow-list restrictions using ARP spoofing and dynamic NAT table modifications. Practical implementation of man-in-the-middle attacks demonstrated the ability to bypass IP allow-list restrictions on IEDs in a controlled environment, addressing challenges like sequence numbers and checksum adjustments. Real-world defense mechanisms, including protocol hardening, encryption, and anomaly detection, effectively mitigate or prevent DNP3 attacks when implemented correctly. What is the implication of the main finding? Recognizing that simpler methods can replicate complex attacks highlights the importance of robust basic security measures, such as IP allow-listing and protocol monitoring. The feasibility of bypassing standard configurations necessitates the integration of advanced authentication and encryption protocols in OT systems to strengthen security. Defense-in-depth strategies, supported by regulatory standards like NERC CIP, underscore the importance of operational measures, including continuous monitoring and SOC practices, for effective cybersecurity in critical infrastructure. Abstract OT (operational technology) protocols such as DNP3/TCP, commonly used in the electrical utility sector, have become a focal point for security researchers. We assess the applicability of attacks previously published from theoretical and practical points of view. From the theoretical point of view, previous work strongly focuses on transcribing protocol details (e.g., list fields at the link, transport, and application layer) without providing the rationale behind protocol features or how the features are used. This has led to confusion about the impact of many theoretical DNP3 attacks. After a detailed analysis around which protocol features are used and how, a review of the configuration capabilities for several IEDs (Intelligent Electrical Devices), and some testing with real devices, we conclude that similar results to several complex theoretical attacks can be achieved with considerably less effort. From a more practical point of view, there is existing work on DNP3 man-in-the-middle attacks; however, research still needs to discuss how to overcome a primary hardening effect: IEDs can be configured to allow for communication with specific IP addresses (allow list). For purely scientific purposes, we implemented a DNP3 man-in-the-middle attack capable of overcoming the IP allow-list restriction. We tested the attack using real IEDs and network equipment ruggedized for electrical environments. Even though the man-in-the-middle attack can be successful in a lab environment, we also explain the defense-in-depth mechanisms provided by industry in real life that mitigate the attack. These mechanisms are based on standard specifications, capabilities of the OT hardware, and regulations applicable to some electrical utilities.

With the pervasiveness of GPS-enabled devices, mobile users can directly visit the best travel routes matching their interests and obtain a better user experience via location-based travel recommendation services. As the number of queries grows, the travel agency for location-based travel recommendations tends to outsource its recommendation services to the cloud server. Since the travel agency’s popular travel routes and raw trajectory data from mobile users contain sensitive information, privacy protection should be guaranteed. Although some schemes have been proposed to solve the privacy problems, no previous works related to the location-based recommendation are proposed over mobile users’ raw trajectories. To solve this problem, we propose a privacy-preserving travel recommendation scheme based on stay points over the raw encrypted trajectory data. Specifically, we first propose an adapted longest common subsequence computation algorithm to measure the similarity of two trajectories. Second, to support some computations under ciphertext, we design several secure two-party computation (S2PC) primitives (e.g., secure division, secure mean coordinate, and secure comparison) based on the Paillier cryptosystem. Third, we implement secure stay points extraction and adapted longest common subsequence computation protocols via these secure computation primitives. Finally, we analyze the security of our proposed scheme in the semi-honest model and show that the privacy of mobile users’ trajectories, query results, and the travel agency’s popular travel routes are well protected. Meanwhile, we evaluate the performance of each secure computation primitive and conduct extensive experiments on synthetic datasets, and the experimental results show that our scheme is practical in the real applications.

Convolutional Neural Networks (CNNs) possess extensive applicability across diverse domains, particularly in the realm of image recognition. In light of the advent of machine learning as a service, the utilization of a well-trained CNN model by servers to execute image classification based on user queries has become a significant service, catering to a wide array of applications. Nevertheless, this convenience is accompanied by the inherent risk of data privacy and model privacy disclosure, which can have severe ramifications, particularly in the context of specialized scenarios like medical images and location images. Hence, how to perform classification for CNN with privacy protection emerges as a crucial research concern. Furthermore, the nonlinearity of CNN's activation function renders it unsuitable for homomorphic cryptosystems. In order to address these challenges, we put forth a privacy-preserving CNN classification scheme employing a distributed two trapdoors public-key cryptosystem (DT-PKC). Initially, we introduce a security protocol toolkit encompassing protocols for secure multiplication, secure activation function computing, and average pooling. In addition, we propose a novel continuous and derivative Tanhplus function as an approximation of the Relu function, aiming to enhance the accuracy of classification results. The secure activation function computing protocol utilizes the aforementioned Tanhplus function in conjunction with the proposed homogenization algorithm to compute the activation function. This protocol guarantees more precise and accurate output in the activation function calculation of CNN when operating under ciphertext. Furthermore, the adoption of the DT-PKC cryptosystem not only ensures privacy protection for CNN classification but also provides support for lightweight users and multiple keys. Finally, security analysis and performance evaluations demonstrate that the proposed scheme is secure, practicable, and efficient with high accuracy.

Split federated learning (SFL) is a recently proposed distributed collaborative learning architecture that integrates federated learning (FL) with split learning (SL), offering an ingenious solution for safeguarding privacy in resource-limited environments. Despite the compelling potential of SFL and its appealing attributes, its robustness remains uncharted territory. In this paper, we investigate the security and robustness of SFL, with a specific focus on its susceptibility to malicious client-driven poisoning attacks. Specifically, we study the weaknesses of SFL against the well-known poisoning attacks designed for FL, like dataset poisoning, weight poisoning, and label poisoning. We also introduce a novel type of poisoning attacks tailored for SFL, named smash poisoning, and evaluate the robustness against smash poisoning attacks and advanced hybrid attacks (DatasetSmash, LabelSmash, and WeightSmash) that amalgamate smash poisoning with the other three methods for FL. By simulating these attacks across diverse domains over four datasets, we find that most of these attacks (including weight, WeightSmash, and LabelSmash poisoning) can disrupt the converged models with straightforward poisoning actions or have persistent negative influence on the model accuracy even after the termination of the attacks. Furthermore, our findings reveal that the robustness of SFL can be augmented by strategically adjusting the system parameters, such as client quantity, bottleneck size, or split type. Finally, we verify the effectiveness of the typical defense mechanisms of poisoning attacks intended for FL and design a new defense strategy that filters out malicious smashed data to improve the robustness of SFL. We observe that the adoption of properly chosen defense mechanisms is beneficial in decreasing the security risks of SFL, but entirely eliminating the impacts of poisoning attacks in SFL is still challenging.

This paper proposes EdgeCrypt Tracker, an object-tracking algorithm combined with differential encryption to provide better accuracy and runtime efficiency for battery-operated Internet of Unmanned Aerial Vehicles (IoUAV). Specifically, EdgeCrypt Tracker operates directly on High-Efficiency Video Coding (HEVC) and has three stages: pre-processing, object tracking, and differential encryption. The pre-processing stage separates intra frames and removes artificial camera motion caused by camera movement from inter frames. Next, the object tracking stage utilizes a hybrid neural network, combining a Single Shot Multibox Detector (SSD) network with a MobileNetV3 backbone that processes intra coded blocks and a Fast Gated Recurrent Neural Network (FastGRNN) network that processes inter coded blocks. Finally, the tracked information is passed to the differential encryption stage, which encrypts all syntax elements within moving objects and alternate syntax elements related to the background. Experimental results demonstrate that EdgeCrypt Tracker achieves an average object tracking accuracy of 92%, real-time inference with a 35% lower encryption overhead compared to state-of-the-art methods. This work demonstrates the potential of integrating object tracking and encryption within video compression for secure, efficient UAV-based surveillance.

With the ubiquity of Internet of Things, location-based service (LBS) providers have collected huge volumes of individuals’ trajectories, which are valuable for some applications, e.g., store location choosing for merchants. However, directly publishing raw trajectories to applications may violate individuals’ data privacy and lead to unexpected loss. Although many trajectory synthesis methods under differential privacy have been proposed to privately publish trajectories data, they cannot sufficiently preserve the semantic information of trajectories. Aiming at this issue, in this paper, we introduce a semantic-preserving scheme to synthesize trajectories for publishing under differential privacy. Specifically, we first design a hierarchical graphical model (HGM) to capture the semantic feature of trajectories. Then, we propose a metric, named the correlation score, to measure the relationship between two locations, which can well capture the geographic feature of trajectories. After that, we propose a private trajectory synthesis algorithm by first adding Laplace noises to the extracted features and then synthesizing trajectories based on the noisy features and the Markov chain theory. Privacy analysis demonstrates that our scheme can protect the privacy of trajectories. In addition, performance evaluation illustrates that our synthetic trajectories maintain good utility semantically and geographically.

As an electronic form of traditional voting, electronic voting is becoming more and more popular in today’s information society. Most of the existing electronic voting protocols need a trusted center to calculate the voting result, but the requirement of a trusted center is often unrealistic and prone to single point of failure. In this regard, the decentralized electronic voting protocols based on blockchain have been proposed. Unfortunately, most existing blockchain-based voting protocols fail to ensure anonymity, legitimacy, and correctness of counting. Besides, they do not satisfy robustness, i.e., the voting result cannot be counted in the event of voter abstention. To address the above challenges, we propose a novel blockchain-based self-tallying voting protocol, where the group signature and zero-knowledge proof are utilized in a way that the voter can securely distribute anonymous and unlinkable electronic ballots, thereby guaranteeing complete anonymity and legitimacy. Meanwhile, a novel signcryption algorithm is designed by combining distributed ElGamal encryption and Paillier encryption algorithms, which enhances the computational efficiency of voting results while supporting robustness. The security proof shows that our protocol ensures the confidentiality of ballots, complete anonymity, legitimacy, fairness, dispute-freeness and resistance against multi-voting. In addition, our protocol satisfies robustness, i.e., voting result can be correctly calculated and verified even if some voters abstain from voting. Finally, extensive experiments show that our protocol greatly reduces the computational cost and communication overhead, and is more practical than existing self-tallying voting protocols.