A novel networking paradigm, known as Named Data Networking (NDN), is proposed as a replacement for the current IP-based internet. With its in-network content caching capabilities, NDN significantly enhances network performance by increasing content availability, lowering latency, and alleviating the load on central producers. This is attained by enabling consumers to retrieve data, based on its name rather than its IP address, from the closest router instead of the farthest producer. In this work, we apply the Window Least Recently Frequently Used (WLRFU) policy as a novel cache replacement strategy in NDN networks. This policy combines recency and frequency based on a sliding access window. Using the ccnSim simulator, we compare the performance of WLRFU with that of the LRFU (Least Recently Frequently Used) and FIFO (First In First Out) policies by employing a real-world network topology and varying several simulation parameters, such as cache size, consumer interest rate, MZipf \(\alpha \) , number of consumers, and the number of producers. Simulation results reveal that the implemented strategy outperforms others with higher cache hit rate and lower latency. Specifically, WLRFU surpasses LRFU, raising the cache hit rate by 4.70 %–14.87 % and reducing latency by 6.89 %–8.94 %.