Thermodynamic aspects of confidentiality

We analyse secure computation as a physical process and connect it to recent advances in security, namely Quantitative Information Flow. Using a classic thermodynamic argument involving the second principle and reversibility we show that any deterministic computation, where the final state of the system is observable, must dissipate at least WkBT ln 2. Here W is the information theoretic notion of security as defined in Quantitative Information Flow, kB the Boltzmann constant and T the temperature of the environment. Such minimum dissipation is also an upper bound on another probabilistic quantification of confidentiality introduced by Smith. We then explore the thermodynamics of timing channels in Brownian computers. Here the low energies involved lead to the emergence of new timing channels arising directly from the entropy variations related to computation.