Skip to Main Content
This paper presents an efficient distributed clustering technique capable of identifying embedded clusters over very large spatial datasets. The technique is based upon a client server approach, where the huge dataset stored in the server are partitioned into almost k equal partitions which are used by k clients to identify the embedded clusters in parallel for each partition sent by the server. Finally, the embedded clusters obtained from the k clients are merged at the Server for the ultimate results. Experimental results establish the superiority of the technique in terms of scale-up, speedup as well as cluster quality, in comparison to its other counterparts (, ).