Table of Content

20 February 2025, Volume 11 Issue 2

Previous Issue   

An Optimized Computation Method for Cipher Symbol Functions  Based on Homomorphic Encryption

2025, 11(2):  100. 

Asbtract ( )   PDF (1092KB) ( )  

References | Related Articles | Metrics

Fully homomorphic encryption extends encryption to computations, allowing ciphertext processing without decryption. Comparative operations, crucial in applications like deep learning, pose a challenge in homomorphic encryption environments restricted to addition and multiplication. Feng et al. (CNS 2023) proposed a comparison method using dynamic polynomial combinations. This paper enhances dynamic polynomial, allowing polynomial fluctuations within (-2,2). It introduces a novel equation system for solving dynamic polynomials and utilizes finite third and fifthdegree polynomials to construct more precise composite polynomials for approximating the sign function. It analyzes the method’s optimality in depth consumption and computational complexity, achieving a 32% reduction in runtime compared to the optimal method in a previous study (CNS 2023). The homomorphic comparison algorithm in this paper, for ε=2-20，α=20 requires only 0.69ms in amortized runtime.

Indoor Localization Security Scheme Based on Geographic  Indistinguishability and Flexible WiFi Deployment

2025, 11(2):  107. 

Asbtract ( )   PDF (1351KB) ( )  

References | Related Articles | Metrics

In indoor localization services, WiFi fingerprinting technology has received widespread attention due to its extensive coverage  and high localization accuracy. However, for the online phase of location query, the user’s personal sensitive information is vulnerable to malicious attacks resulting in location privacy leakage. Existing WiFi fingerprintbased indoor positioning technologies primarily focus on single flat surfaces within indoor environments, which restricts the flexibility of WiFi deployment. When WiFi is deployed in multidimensional scenarios, addressing spatial location privacy issues becomes imperative. In this paper, a WiFi fingerprinting indoor localization privacy protection scheme based on geographic indistinguishability is proposed, in which the user generates a new received signal strength vector by using his own received signal strength and sends the obtained data to the location service provider through noise obfuscation, and introduces a digital signature technique to ensure that the client’s identity is not forged before obfuscating the position to be sent to the location service provider to achieve localization. Experimental results based on the simulation experimental platform show that the new scheme supports flexible deployment of WiFi, and is able to realize highprecision localization for the first time in the case of flexible deployment of 12 WiFi access points with guaranteed localization error of less than 1m while protecting location privacy.

Edge Cloud Collaborative Attributebased Signcryption Scheme  Based on State Secret SM9

2025, 11(2):  115. 

Asbtract ( )   PDF (1474KB) ( )  

References | Related Articles | Metrics

In order to improve the security and efficiency of data interaction in edge cloud collaborative mode, an edge cloud collaborative attributebased signcryption scheme based on state secret SM9 is proposed. This scheme integrates the state secret SM9 algorithm with attributebased signcryption algorithm, constructs a mixed key and ciphertext policy access control mechanism with a linear secret sharing scheme, and implements partial outsourcing decryption through an edge cloud collaborative network. The experimental analysis results demonstrate that the proposed scheme provides flexible access control while achieving efficient and reliable security protection in the edge cloud collaborative mode, making it suitable for dynamic and complex cloud application scenarios.

Research on Deep Learningbased Spatiotemporal Feature Fusion  Network Intrusion Detection Model

2025, 11(2):  122. 

Asbtract ( )   PDF (1944KB) ( )  

References | Related Articles | Metrics

As the number of network attacks increases, network intrusion detection systems are becoming increasingly important in maintaining network security. Most studies have used deep learning approaches for network intrusion detection but have not fully utilized the features of traffic from multiple perspectives. Additionally, these studies often suffer from the use of outdated experimental datasets. In this paper, a parallelstructured DSCInceptionBiLSTM network is proposed to evaluate the designed network model using stateoftheart datasets. The model consists of two branches, network traffic image, and text anomaly traffic detection. Spatial and temporal features of traffic are extracted by improved convolutional neural networks and recurrent neural networks, respectively. Finally, network intrusion detection is achieved by fusing spatiotemporal features. The experimental results show that our model achieves 99.96%, 99.19%, and 99.95% accuracy on the three datasets of CICIDS 2017, CSECICIDS 2018 and CICDDoS 2019, respectively, effectively classifying the anomalous traffic with high precision and meeting the requirements of intrusion detection system.

A Malicious TLS Traffic Detection Method with Multimodal Features

2025, 11(2):  130. 

Asbtract ( )   PDF (3159KB) ( )  

References | Related Articles | Metrics

The malicious TLS traffic detection aims to identify network traffic that involves malicious activities transmitted through the TLS protocol. Due to the encryption properties of the TLS protocol, traditional textbased traffic analysis methods have limited effectiveness when dealing with encrypted traffic. To address this issue, a malicious TLS traffic detection method called MultiModal Feature Fusion for TLS Traffic Detection (MTBRL) has been proposed. This method extracts and fuses features from different modalities to detect malicious TLS traffic. Firstly, expert knowledge is employed for feature engineering, extracting key features from encrypted traffic, including protocol versions, encryption suites, and certificate information. These features are processed and transformed into twodimensional image representations. Then, ResNet is utilized to encode these images and extract their features. Simultaneously, an encrypted traffic pretrained BERT model is used to encode TLS flows, allowing the learning of contextual and semantic features of the TLS traffic. Additionally, an LSTM model is employed to encode the sequence of packet length distributions of the encrypted traffic, capturing temporal characteristics. Finally, through feature fusion techniques, the different modality features are integrated, and the model’s weight parameters are automatically learned and optimized using the backpropagation algorithm to accurately predict malicious TLS traffic. Experimental results demonstrate that this method achieves accuracy, precision, recall, and F1score of 94.94%, 94.85%, 94.15%, and 94.45%, on the DataCon2020 dataset. This performance is significantly superior to traditional machine learning and deep learning methods. 

Group Key Management Mechanism for Internet of Vehicles

2025, 11(2):  139. 

Asbtract ( )   PDF (873KB) ( )  

References | Related Articles | Metrics

Based on the characteristics of the Internet of vehicles(IoV), a treebased lightweight group key management mechanism (Lightweight Tree Group Key Management Mechanism, LTGKM) is proposed to realize the security of multicast and broadcast communications in the IoV. LTGKM adopts a hierarchical approach to generate, distribute and update the group keys. The management node of various layers generates the corresponding group key using the HMAC function as the key derivation function, and distributes the group key to the child node based on the encrypted certification algorithm; When a new node joins, the parent node generates a new group key and distribute it to the new node, and the remaining nodes update the group key by themselves; when the user leaves, the nonleaf node updates the group key by themselves, and the new group key is distributed to the leaf node by its father node. Security analysis shows that LTGKM realizes the randomness, forward security, and backward security during the group key generation and update, and the confidentiality, integrity and uniformity during key distribution. Performance analysis shows that LTGKM has obvious advantages in storage, computing and communication.

Research on Crosschain Privacy Sharing Based on Improved #br# Notary Mechanism#br#

2025, 11(2):  146. 

Asbtract ( )   PDF (1405KB) ( )  

References | Related Articles | Metrics

In recent years, blockchain technology has been continuously developed, but with the development, problems have also emerged. For example, the problem of privacy data leakage on the chain and the problem of crosschain interaction of private data. In response to the above problems, a crosschain privacy sharing model based on an improved notary mechanism is proposed. First, the model designs a crosschain data sharing mechanism based on threshold proxy reencryption. This mechanism stipulates that notary nodes need to pledge a certain amount of tokens, crosschain initiators need to provide a certain amount of crosschain rewards, and honest notary nodes obtain tokens. Coin reward, the malicious notary node deducts a certain amount of deposit and gives it to the initiator. This mechanism effectively reduces the possibility that the notary node is a malicious node. Then, design a token locking mechanism based on hash locks and Merkle trees to ensure that cross chain initiators and notary nodes can achieve data trustworthiness across chains without mutual trust finally, experimental results and theoretical research prove that the proposed model has a low probability of malicious attack on the notary node and a high cost of malicious attack, and the algorithm used is better than other comparison schemes in terms of encryption and decryption efficiency.

Research on Multimodal Cyberbullying Detection Model for #br# Social Networking Platforms#br#

2025, 11(2):  154. 

Asbtract ( )   PDF (2099KB) ( )  

References | Related Articles | Metrics

With the rapid development of social networking platforms, the issue of cyberbullying has become increasingly prominent. The diverse forms of online expression that combine text and images have increased the difficulty of detecting and managing cyberbullying. This paper constructs a Chinese multimodal cyberbullying dataset that includes both text and images. By integrating the BERT(bidirectional encoder representations from transformers) model with the ResNet50 model, we extract singlemodal features from text and images, respectively, and perform decisionlevel fusion. The fused features are then detected, achieving accurate identification of text and images as either cyberbullying or noncyberbullying. Experimental results indicate that the multimodal cyberbullying detection model proposed in this paper can effectively identify social media posts or comments that contain cyberbullying characteristics in both text and images. It enhances the practicality, accuracy, and efficiency of detecting multimodal cyberbullying, providing a new approach and method for the detection and management of cyberbullying on social networking platforms. This contributes to the creation of a healthier and more civilized online environment.

Overview of Regulation of Crossborder Data Flow

2025, 11(2):  164. 

Asbtract ( )   PDF (1274KB) ( )  

References | Related Articles | Metrics

The development of the digital economy has made crossborder data flow an inevitable trend, and while bringing economic benefits, the security of crossborder data flow cannot be ignored. Due to the complexity of the subjects and scenes involved in the process of crossborder data flow, and the uncontrollability of the process, how to regulate the possible security problems in the process of crossborder data flow has become the focus of the world. So far, there is no unified governance rule system for crossborder data flow in the world, and at the same time, there are huge differences in legislation on crossborder data flow in different countries, which results in the complex situation of legislation on crossborder data flow in the world. This paper describes the current situation of crossborder data flow from the perspectives of laws and regulations, bilateral agreements and standards, and in this way develops horizontal comparisons, sorts out the existing regulatory differences, analyzes the challenges and opportunities China faces under the current trend, and gives reasonable countermeasures.

Exploring Effective Factors Leading to Data Leakage in Pretrained #br# Language Models#br#

#br#

2025, 11(2):  181. 

Asbtract ( )   PDF (1261KB) ( )  

References | Related Articles | Metrics

Currently, pretrained language models are widely used to learn general language representations from massive training corpora. The performance of downstream tasks in the field of natural language processing has been significantly improved after using the pretrained language model, but the overfitting phenomenon of the deep neural network makes the pretrained language model may have the risk of leaking the privacy of the training corpus. This paper selects T5, GPT, OPT and other widely used pretrained language models as research objects, and uses model inversion attacks to explore the factors that affect the data leakage of pretrained language models. During the experiment, the pretrained language model was used to generate a large number of samples, and the samples most likely to cause data leakage risk were selected for verification by indicators such as perplexity. It proved that different models such as T5 have different degrees of data leakage problems. For the same model,  the larger size of the model, the scale, the greater the possibility of data leakage; adding a specific prefix makes it easier to obtain leaked data. The future data leakage problem and its defense methods are prospected.


Research and Application of Trusted Data Security Management #br# Technology Based on Chameleon Hash#br#

2025, 11(2):  189. 

Asbtract ( )   PDF (2544KB) ( )  

References | Related Articles | Metrics

To simultaneously address the demands for data updates and data security management in the field of data circulation, this paper investigates a trusted data security management scheme based on chameleon hash. Initially, the mathematical foundations of chameleon hash are analyzed and three construction methods are compared. A data security management approach that integrates chameleon hash with homomorphic encryption is summarized and applied to digital rights protection. This method not only permits the updating and modification of submitted data but also ensures data security and userfriendliness. Finally, the efficiency of the proposed method is evaluated through experimental simulations. The results demonstrate that the data security management and update method proposed in this paper is suitable for environments requiring frequent data updates and certain security needs. This method effectively reduces the overall cost of data updates, providing an efficient and secure solution for the circulation of data elements.