Markov decision process-based cache replacement for efficient content retrieval in named data networking
Sushil Kumar Bagi, Neeraj Kumar · 2025
Named Data Networking (NDN) is a unique networking approach based on the data retrieval concept rather than host addresses, i.e., IP address-based, which improves the efficiency of information dissemination. Timely and accurate caching of popular content between the path of producer and consumer will be necessary for NDN. However, many algorithms have been developed so far to achieve better caching but still face limitations in efficiency and adaptability. Despite advancements, limited research has explored the application of AI/ML in NDN caching to enhance adaptive decision-making and the results affected when we applied AI/ML. Caching is a phenomenon of decision-making by which NDN decides which content should be replaced from the content store at a time when there is no space at the router in NDN. The development of artificial intelligence plays a vital role in research, where researchers use artificial intelligence and machine learning to automate the process by applying various algorithms to decision-making. In this paper, we applied one of the machine learning algorithms, i.e. a Markov Decision Process framework in a caching system. We discussed the result about FIFO, LFU and MDP based caching. The significance of this research paper is that it uses MDP-based machine learning in cache replacement and tries to show that both MDP-based caching and LFU achieve a higher cache hit ratio than FIFO.