QRNGs are used to improve cryptographic protocols but have been too large, power hungry and costly to include in client devices. The research team at University of Geneva has exploited the fact that modern CMOS image sensors are sensitive to light at the level of a few photons and photon emission from light sources is an inherently quantum process. In addition CMOS image sensors have millions of pixels and high-speed readouts allowing for millions of random bits to be generated per second.
Quantum random number generation on a mobile phone was authored by Bruno Sanguinetti, Anthony Martin, Hugo Zbinden and Nicolas Gisin and submitted to the arxiv.org online journal. The authors used a commercial astronomy monochrome CCD camera (ATIK 383L), and the green pixels on a CMOS sensor in a Nokia N9 mobile phone.
Full details of the method for extracting RNGs can be found in the paper
Related links and articles:
Quantum random number generation on a mobile phone
News articles:
Sony 13Mpixel sensor offers better video
CMOS image sensor market on 10% CAGR
Top ten CMOS image sensor vendors ranked
Sony curves image sensors; TSMC stacks them