CTJ Dodson, UMIST, Manchester
Information Geometry: Security and Stochastic Clustering
Designers of secure software systems need to monitor and quantify event clustering in order to minimize information leakage to probes by an attacker, perhaps through introduction of obscuring procedures in a restricted memory device such as a smartcard. An ideal situation would be to have scheduling that to an attacker resembles closely a random sequence of events. The basic “random” model for stochastic events is the Poisson process; for events on a line this results in an exponential distribution of intervals between events. Here we discuss the differential geometry of manifolds of gamma distributions, which contain exponential distributions as a special case; this gives a means of quantifying departures from randomness.