Cache Memory Management via Content Life Cycle Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing cache replacement methods in communication systems do not consider the life cycle of content, leading to suboptimal cache management and reduced cache hit ratio.
Innovation Solution
A method for managing cache memory in a communication system that calculates the expected popularity of content based on its life cycle, using metrics like expected number of views, rising momentum, and recent views to prioritize and store content, thereby enhancing cache replacement efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If LRU or LFU cache replacement methods are used, then cache management is simple, but cache hit ratio is suboptimal because content life cycle is not considered
Solution Approach 1:
The patent changes the parameter used for cache replacement from simple time-based (LRU) or frequency-based (LFU) metrics to a composite parameter that incorporates content life cycle stage classification. By dividing content life into introduction, growth, maturity, and decline stages, the system adapts replacement decisions to the actual popularity trajectory of each content item, thereby improving cache hit ratio without significantly increasing operational complexity.
2Device complexity
If cache replacement does not consider content life cycle, then cache management is straightforward, but content popularity trends are not utilized for optimized replacement
Solution Approach 1:
The patent applies preliminary action by pre-classifying content into life cycle stages based on popularity trends before replacement decisions are needed. The system continuously monitors content popularity and assigns life cycle stages in advance, so that when cache replacement is required, the system can immediately consult the pre-determined life cycle classification rather than performing complex real-time analysis, thus improving replacement efficiency.
3Ease of manufacture
If traditional cache methods are used, then implementation is simple, but future content popularity cannot be predicted for optimal caching
Solution Approach 1:
The patent implements feedback by continuously monitoring content popularity metrics and using this information to update life cycle stage classifications. The system feeds back popularity trend data to the cache replacement algorithm, allowing it to adapt to changing content demands. This feedback mechanism enables the system to predict future popularity by analyzing recent trends, thereby optimizing cache retention decisions and reducing content delivery delays.
Data Source
Figure 1
Figure 2~3
Figure 4
AI summary
In the present invention, a base station determines from a communication system whether a first content, which is requested by a mobile terminal, is saved on a cache memory, attributes a predetermined priority ranking to the first content and saves same on the cache memory when the first content is not saved on the cache memory, and updates the priority ranking of the first content on the basis of a predicted popularity of the first content, wherein the predicted popularity is decided on the basis of change in the number of views of a content that corresponds to a category of the first content.