Bloom Filters & Their Applications

A Bloom Filter (BF) is a data structure suitable for performing set membership queries very efficiently. A Standard Bloom Filter representing a set of n elements is generated by an array of m bits and uses k independent hash functions. Bloom Filters have some attractive properties including low storage requirement, fast membership checking and no false negatives. False positives are possible but their probability may be controlled and significantly lowered depending upon the application requirements. There are many variants of the standard Bloom Filter – counting BF, variable increment BF, compressed BF, scalable BF, generalized BF, stable BF and Bloomier Filter. Bloom Filters are increasingly finding applications in fast and approximate search, encrypted search in the cloud, routing and controlling of network traffic, network intrusion detection and differential database and file updating. This paper explores the typical properties of Bloom Filters, their variants and their suitability for use in present day applications.