Pseudorandomness

Abstract

Most physical processes expose some random behaviour. A good example of such random behaviour is noise in telecommunication channels. A great irony is that when there is a need for a source of random bits or numbers, then the ever-present randomness is in short supply. Generation of a large volume of random bits is usually very expensive and requires special hardware. Also the parameters of truly random generators can fluctuate, so they need to be calibrated and tested from time to time. The major drawback of truly random generators is the lack of reproducibility of the yielded bits and numbers. The reproducibility is crucial in simulations where there is a need to repeat the same experiments many times. It is also necessary in some cryptographic applications when, for instance, two communicating parties want to generate identical sequences from a shared secret (and short) key. From a cryptographic point of view, we are interested in deterministic algorithms that efficiently generate strings of bits and that cannot be distinguished from truly random ones. Readers interested in the subject are referred to Goldreich [205].