Adaptive Cache Replacement Strategy Based on Greedy Dual Size Frequency Incorporating DBScan for Internet Access Anomaly Detection

In content delivery networks (CDN) and edge caching, efficient cache replacement strategies are essential to reduce latency and enhance hit ratios.This study introduces adaptive similarCache (ASC), an adaptive caching framework based on greedy dual size frequency (GDSF) that includes DBSCAN for anomaly detection.ASC framework analyzes the IRcache dataset, which consists of proxy server access logs from four cities-Boulders, New York, Silicon Valey, and Urbana Campaign-and simulates dynamic adjustments to access patterns as they would occur in real-time, considerably boosting cache performance under fluctuating traffic situations when compared to classic approaches like LRU and LFU.The technique improves cache utilization, achieving a hit ratio of up to 45% and exhibiting lower latency in high-demand conditions, focusing specifically on data from these cities.The findings underscore the need for adaptive caching techniques, particularly in resource-constrained contexts, and suggest the possibility of future hybrid systems.