Imitating K-Means to Enhance Data Selection
In this study, a new approach that utilizes availability, security and time as selection criteria between different replicas is proposed. However, selecting the replica in accordance with the three factors simultaneously is complicated; therefore, concepts from the K-means clustering algorithm were adopted to create a balanced (best) solution. Numerical simulations were carried out to assess the proposed technique. The results show that the proposed system outperforms random algorithm by 17% and outperforms the round robin algorithm by 11%.
Cited References Fulltext