Development of a machine learning-based in-network caching strategy for NDN

Abstract

Cache management is an important component in any network, and more is that it’s essential in the NDN architecture. In this work first, we will talk about the difference between IP (Internet Protocol), NDN (Named Data Network), the cache management, its types, and the importance of that in the NDN architecture. Including ML (Machine Learning), using it at the cache level is the focus of our work. The Apriori algorithm is supervised learning that is suited for our work as we have the data for this learning approach, after that we find the association rules, to recommend the next requested data. A low hit ratio leads to the increase of requesting content from the content provider, for the client that means more access latency and pressure on the server. The result obtained determines that the performance of the network and its influence on the cache increased by 3.2% in the two content store sizes of 20 and 40, the size of 80 shows an increase of the cache hit ratio for both curves as identical with both reach 90%.