A new technique for proving the adaptive indistinguishability of two systems, each composed of some component systems, is presented, using only the fact that corresponding component systems are non-adaptively indistinguishable. The main tool is the definition of a special monotone condition for a random system F, relative to another random system G, whose probability of occurring for a given distinguisher D is closely related to the distinguishing advantage ε of D for F and G, namely it is lower and upper bounded by ε and ε(1 + In 1/ε), respectively. A concrete instantiation of this result shows that the cascade of two random permutations (with the second one inverted) is indistinguishable from a uniform random permutation by adaptive distinguishers which may query the system from both sides, assuming the components' security only against non-adaptive one-sided distinguishers. As applications we provide some results in various fields as almost k-wise independent probability spaces, decorrelation theory and computational indistinguishability (i.e., pseudo-randomness). © Springer-Verlag 2004.
CITATION STYLE
Maurer, U., & Pietrzak, K. (2004). Composition of random systems: When two weak make one strong. Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2951, 410–427. https://doi.org/10.1007/978-3-540-24638-1_23
Mendeley helps you to discover research relevant for your work.