A Effective Algorithm for Counting Active Flows Based on Loop Bloom Filter

Network traffic analysis based on flows plays an important role in network monitoring and P2P identification. They are all grounded in flow counting and differ in how to count. Timeout detecting is a major strategy to identify termination of flows and can influence the validity of flow counting. The algorithms for counting flow based on Hash method can´t deal with timeout problems perfectly. In this paper, a novel algorithm based on loop bloom filter is proposed. It can estimate active flow with timeout detection on high speed links more accurately. Experiment results show that this algorithm can ensure accuracy while only consuming minimum time and memories.