Skip to Main Content
A Bloom filter is a simple space-efficient randomised data structure allowing membership queries over data sets. It is widely utilised in peer-to-peer network, traffic measurement and distributed systems. Aiming at the deficiencies of the naïve counting Bloom filters (NCBFs), a novel data structure called balanced counting Bloom filters (BCBFs) is presented. In order to achieve space-efficient storage and effective query, the BCBF adopts the following methods: introducing hash fingerprints, partitioning bucket vectors into equally sized segments and storing elements with the least load bucket. Analytical expressions are deduced in detail based on the theory of differential equations and probability. Besides, simulations are conducted based on the data produced by computer and real network trace. The results demonstrate that the BCBF cannot only serve the same functionality as the NCBF using much less space, but also becomes a valuable tool in hardware to scale the high-speed link.