Abstract: Traditional method of mining frequent elements in distributed data stream tends to result in excessively communication within layers , and bandwidth is bottleneck. To minimize communication requirements, we propose a method of sampling from distributed da

Key words: Distributed data stream, Frequent items, Algorithm