A Time Pattern-Based Intelligent Cache Optimization Policy on Korea Advanced Research Network

Data is growing quickly due to a significant increase in social media applications. Today, billions of people use an enormous amount data access the Internet. The backbone network experiences substantial load as result users. Users same region or company frequently ask for similar material, especially on platforms. subsequent request content can be satisfied from edge if stored proximity user. Applications that require relatively low latency Content Delivery Network (CDN) technology meet their requirements. An and center constitute CDN architecture. To fulfill requests minimize impact network, requested buffered closer user device. Which should kept primary concern. cache policy has been optimized using various conventional unconventional methods, but they have yet include timestamp beside video request. 24-h pattern was obtained publicly available datasets. popularity influenced by time day, shown time-based profile. We present optimization method based requests. problem described hit ratio maximization emphasizing relevance score machine learning model accuracy. A predicts cached next stamp, identifies removed cache. Afterwards, we gather logs generate extracted pattern. These are pre-processed create dataset divided into three-time slots per day. Long short-term memory (LSTM) trained this forecast at interval. proposed caching evaluated our architecture deployed Korean Advanced Research (KOREN) infrastructure. Our findings demonstrate how adding patterns impacts system increasing rate. show effectiveness model, compare results with state-of-the-art techniques.