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. The method comprises the following steps: (1) preprocessing content popularity data; (2) offlinetraining the neural network to obtain a popularity prediction model and a content classifier; (3) obtaining the category of the content by a content classifier, and predicting the popularity of the content online by using a prediction model of the corresponding category; and (4) comparing the popularity predicted values of all the contents, and making a corresponding cache decision. According tothe invention, under the condition that only the characteristic of content request frequency is used; the edge nodes are enabled to predict popularity of different contents online and track popularitychanges of the contents in real time, and a corresponding caching decision is made based on the predicted content popularity, so that the edge nodes are ensured to continuously cache hot contents, and a caching hit rate which is asymptotically close to an ideal caching method is obtained.

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 continued rapid proliferation of various smart devices and advanced mobile application services, wireless networks are experiencing unprecedented data transmission pressures in recent years. The ever-increasing amount of data transmission has brought enormous pressure on backhaul links with limited capacity, especially at peak traffic times. Edge caching technology can effectively reduce the backhaul load by placing the most popular content on nodes closer to the requesting user, and has received extensive attention in the industry and academia in recent years. Due to the limited storage space of nodes and the variation of content popularity with time and space, edge caching technology faces various challenges, such as, in order to ma...

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