Modeling bipolar junction effects for high entropy generators

Abstract

The analog noise produced by semiconductor junctions is considered to be the fundamentally unpredictable sum of astronomical numbers of independent quantum actions. In this way, the unpredictability part of this makes noise attractive for cryptography. On the other hand, real noise often exhibits measurable and predictable correlations which contradict the simple model. Modeling bipolar junction effects can provide some quantitative and qualitative information on the entropy generated by such devices. The paper proposes a flowchart for modeling noise related bipolar junction effects along with an algorithm for building noise models for semiconductor components.