| abstract |
The invention relates to a web page fingerprint monitoring method for anonymous network multi-tab browsing, which is suitable for the real scene of multiple tag web browsing, and is compatible with the special scene of single-tag web browsing. The present invention utilizes webpage fingerprint identification technology, firstly performs optimal block-based segmentation on multi-label webpage traffic, and then performs accurate webpage identification on the segmented label webpage traffic, so as to monitor the browsing of webpages through an anonymous network in a multi-label scenario the behavior of. The invention uses data blocks instead of data packets as the segmentation granularity to divide the web page traffic of different label web pages, improves the efficiency of segmentation and the accuracy of web page identification, and also reflects certain robustness. The present invention uses the same CNN classifier to determine multi-label browsing behavior and segment the traffic of each labeled web page, and then uses two CNN classifiers to identify multiple labeled web pages; during the implementation of the present invention, there is no need to manually adjust parameters, and the CNN classifier can be trained quickly convergent and robust. |