Mathematical Foundations of Computer Science 1999

Volume 1672 of the series Lecture Notes in Computer Science pp 262-272

Efficient Strongly Universal and Optimally Universal Hashing

Extended Abstract
  • Philipp WoelfelAffiliated withLehrstuhl Informatik 2, Universität Dortmund

* Final gross prices may vary according to local VAT.

Get Access


New hash families are analyzed, mainly consisting of the hash functions h a,b : {0,... , u − 1} → {0,... , r − 1}, x ↦ ((ax + b) mod(kr)) div k. Universal classes of such functions have already been investigated in [5, 6], and used in several applications, e.g. [3,9]. The new constructions which are introduced here, improve in several ways upon the former results. Some of them achieve a smaller universality parameter, i.e., two keys collide under a randomly chosen function with a smaller probability. In fact, an optimally universal hash class is presented, which means that the universality parameter achieves the minimum possible value. Furthermore, the bound of the universality parameter of a known, almost strongly universal hash family is improved, and it is shown how to reduce the size of a known class, retaining its properties. Finally, a new composition technique for constructing hash classes for longer keys is presented. Its application leads to efficient hash families which consist of linear functions over the ring of polynomials over ℤm.