Considering the poor storage and query performances of nave counting Bloom filter (NCBF)
a data structure called geometric Bloom filter (GBF) is presented.In order to achieve space-efficient storage and fast query
the structure introduces the idea of hash fingerprints
partitions Bloom filter twice and stores elements with buckets.Based on theory of differential equation and probability
analytical expressions of GBF are deduced.The relational expressions between error probability and space complexity are also established.Furthermore
the inner characteristic of GBF taking on geometric distribution is proved.Simulated results indicate that GBF can achieve lower error probability and computational complexity without sacrificing accuracy compared with NCBF.