Content popularity prediction and edge caching method based on deep learning

Abstract

The invention discloses a content popularity prediction and edge caching method based on deep learning, comprising the following steps: (1) preprocessing the content popularity data; (2) training a neural network offline to obtain a popularity prediction model and Content classifier; (3) Get the category of the content by the content classifier, and use the prediction model of the corresponding category to predict the popularity of the content online; (4) Compare the popularity prediction values ​​of all the contents and make corresponding cache decisions. The present invention can enable edge nodes to predict the popularity of different contents online and track their popularity changes in real time, and make corresponding caching decisions based on the predicted popularity of contents, thereby ensuring Edge nodes continue to cache hot content, and obtain a cache hit rate that is asymptotically close to the ideal caching method.

Description

technical field [0001] The invention relates to a content popularity prediction and edge caching method based on deep learning, and belongs to the technical field of edge caching in mobile communications. Background technique [0002] With the continuous and rapid increase of various smart devices and advanced mobile application services, wireless networks are under unprecedented data transmission pressure in recent years. The ever-increasing amount of data transmission puts enormous pressure on the backhaul links with limited capacity, especially during the peak hours of business transmission. Edge caching technology can effectively reduce the backhaul load by placing the most popular content on the node closer to the requesting user, and has received extensive attention from the industry and academia in recent years. Due to the limited storage space of nodes and content popularity changes with time and space, edge caching technology faces various challenges, such as, in o...

AI Technical Summary

Problems solved by technology

[0003] Traditional caching strategies such as first-in-first-out caching strategy, least recently used caching strategy, least recently used caching strategy, and variants of these algorithms have been widely used in wired networks, however, due to the coverage and storage space of edge nodes in wireless networks Limited, the above traditional caching strategies cannot directly predict content popularity in advance, these traditional caching strategies may suffer severe performance degradation in wireless networks, therefore, these traditional caching strategies are no longer suitable for wireless networks

The current research work is turning to edge caching strategies based on popularity, but the popularity trends of different contents are different, the prediction of content popularity is more complex and cannot track the changes of popularity in real time, and the improvement of cache performance is limited. Moreover, there is less information for predicting popularity in the mobile network, and using too much feature data for prediction obviously does not meet the requirements of edge caching

Images