Analysis and design of true random number generators

In the thesis, random number generators which can be used in many fields, especially cryptographic applications, were studied and new generators are proposed. First of all, the structure of random number generators and the parameters to be considered when designing these structures were examined. The existing literature on these generators was analayzed and many designs related to the thesis were discussed. In addition, the parameters and tests used in the analysis of the generators were examined in detail. Then safe and efficient random numbers were generated with proposed two different approaches. The reliability of the random number sequences generated with the generators was demonstrated with analysis and tests. The first proposed approach is a hybrid random number generator consisting of two parts. Keccak was reorganized and used in the algorithmic method which is the first part of the generator. In the second part, additional inputs was produced by performing the ring oscillator in FPGA environment. The random numbers generated by this generator were shown to meet the security requirements for cryptographic applications. Furthermore, it was shown that the produced numbers did not contain any statistical weaknesses with the successful results from the NIST 800-22 test set and autocorrelation tests. Efficiency is another important advantage of the generator, and the received actual random number bit sequence is doubled. The second proposed approach is consists of two parts. Firstly, entropy source was created by using chaotic systems performed in FPGA environment and real random numbers were generated. Then, using the postprocessing, the possible problems found in the generator were solved. Keccak was edited and used for the postprocessing. The numbers generated by this approach were shown to meet the security requirements for cryptographic applications. Furthermore, it was shown that the NIST 800-22 test set and autocorrelation tests were successful. An important advantage of the proposed approach is that it operates at 100% efficiency.