On the application of agglomerative hierarchical clustering for cache-assisted D2D networks

With rapid increase in the use of traffic-intensive applications, approaches that improve users experience by reducing delay are recently receiving enormous attention. Caching in Device-to-Device (D2D) networks, in particular, is considered as an effective technique to improve the service quality of the network. In this paper, an agglomerative hierarchical clustering algorithm is proposed for a cache-assisted D2D communication network. The algorithm considers users preferences and groups them into the same cluster based on the similarity of their requested content. An optimal caching strategy has been applied and the cache hit probability has further being optimized within each cluster. Performance of the algorithm has been examined in different clusters, considering both sparse and dense user environments. Simulation results show that the cache hit probability within each cluster is higher for higher Zipf parameter, denser domains, and larger number of participating devices. The D2D cache hit probability has also been examined with changing number of clusters under a base station. In this scenario, the results show that the clustering based D2D cache hit probability is higher than the non-clustered case, and the cache hit probability increases with increasing number of clusters