A GNN-based Approach to Optimize Cache Hit Ratio in NDN Networks

Named data networking (NDN) is an emerging network architecture that has in-network caching functionality. Optimizing caching in NDN can reduce the traffic workload and improve network efficiency. This paper represents a Graph Neural Network (GNN) based caching strategy to improve caching in NDN. Firstly, we utilize the convolutional neural network to extract time-series features for each node. Secondly, we apply GNN to make node-wise content caching probability predictions. Finally, we make cache replacement decisions according to the content caching probability ranking of each node. Experimentation shows that our caching strategy achieves around 30% higher cache hit ratio and 5 milliseconds lower latency than the state-of-the-art caching strategy.