Research And Implementation Of Fingerprint-based Web Access Identification Method

The security of web services is crucial for information security in modern society,and browser fingerprinting technology has become one of the important techniques to enhance web access identification.However,achieving continuous user access identification requires not only improving the accuracy of identifying normal variations in fingerprints but also enhancing the robustness under the interference of fake fingerprints.In this context,this paper proposes a dynamic fingerprint linking algorithm for web access identification based on Convolutional Neural Networks(CNN)and the Linformer model.The specific research work mainly includes the following three aspects:(1)To address the issue of constantly changing browser fingerprints,a dynamic fingerprint linking algorithm based on the CNN-Linformer model is designed.This algorithm utilizes the multi-head linear attention mechanism in the Linformer model to weight the input data from different positions,thereby focusing on more important fingerprint features and improving the accuracy of access identification.Experimental results demonstrate that the proposed algorithm achieves an average maximum tracking duration of 44 days for user browsers,enabling long-term and persistent identification of users.Furthermore,as the fingerprint collecting frequency increases,the proposed algorithm maintains a high average ownership ratio while exhibiting a lower number of ids per user compared to Long Short-Term Memory(LSTM)and Bidirectional Gated Recurrent Unit(BiGRU)based linking algorithm.(2)To solve the problem of users or attackers deliberately altering browser fingerprints using fake fingerprinting techniques to evade access identification,this paper analyzes common fake fingerprinting techniques based on the principle of fingerprint feature consistency.On this basis,the consistency features of the operating system,browser,and Canvas are incorporated as part of the fingerprint features to improve the original dynamic fingerprint linking algorithm.Experimental results demonstrate that the optimized algorithm achieves a Matthews Correlation Coefficient(MCC)of 98.89%,indicating that the algorithm has excellent classification performance.In access identification scenarios with interference from fake fingerprints,it not only exhibits high accuracy but also demonstrates robustness.(3)This paper designs and implements a fingerprint-based web access identification system that integrates browser fingerprint extraction,multifactor authentication,and dynamic fingerprint linking technologies.The system is capable of performing access identification when users log in and continuously identifying users throughout their interaction with web services.The system demonstrates strong practicality and operability,making it highly valuable for real-world applications.