A dynamic structure of counting bloom filter

Abstract

Counting Bloom Filter based on the counter-array structure has the shortcoming of counter overflow and less space-efficient. To address these shortcomings, we propose a dynamic structure for Counting Bloom Filter which dynamically changes the counter size according to the number of inserted elements. Hence it not only makes a better use of memory space but also eliminates counter overflow. We put up with the methods of addition and subtraction bit by bit while inserting and deleting elements to effectively reduce the times of memory access. In this way, an effective tradeoff can be achieved between counter access speed and space efficiency. Besides, to reduce excessive memory allocation/deallocation cost caused by consecutively changing counter size, we propose a configurable delayed shrinking algorithm which can appropriately delay the counter size shrinking based on user's configuration. The experiment results show that our dynamic structure could meet the needs of most application scenarios. © 2012 Springer-Verlag GmbH.