Current through Register Vol. 54, No. 49, December 7, 2024
Section 810a.5 - Random number generator standards(a) The random number generator must be cryptographically strong at the time of submission for approval. When more than one instance of a random number generator is used in an interactive gaming system, each instance must be separately evaluated and certified. When each instance is identical but involves a different implementation within a game/application, each implementation shall also be separately evaluated and certified. Any outcomes from the random number generator used for game symbol selection/game outcome determination must be shown, by data analysis and a source code read, to: (1) Be statistically independent, unless the submission has been approved for a persistent-state outcome determination. (2) Be fairly distributed (within statistically expected bounds) over their range. (3) Pass various recognized statistical tests. (4) Be cryptographically strong. (b) Random number generators must adhere to standards in § 461a.7 (relating to slot machine minimum design standards). (c) The gaming laboratory may employ the use of various recognized tests to determine whether or not the random values produced by the random number generator pass the desired confidence level of 95%. These tests include the following: (2) Equi-distribution (frequency) test. (6) Coupon collectors test. (8) Kolmogorov-Smirnov test. (9) Adjacency criterion tests. (10) Order statistic test. (11) Runs tests (patterns of occurrences should not be recurrent). (12) Interplay correlation test. (13) Serial correlation test potency and degree of serial correlation (outcomes should be independent of the previous game, unless the submission has been approved for a persistent-state outcome determination). (14) Tests on subsequences. (15) Poisson distribution. (d) The scaling method may not compromise the cryptographic strength of the random number generator. The scaling method must preserve the distribution of the scaled values. For example, if a 32-bit random number generator with a range of the set of integers in the closed interval (0, 232-1) were to be scaled to the range of the set of integers in the closed interval (1, 6) so that the scaled values can be used to simulate the roll of a standard six-sided die, then each integer in the scaled range should theoretically appear with equal frequency. In the example given, if the theoretical frequency for each value is not equal, then the scaling method is considered to have a bias. Thus, a compliant scaling method must have bias equal to zero. (e) If the interactive gaming system utilizes hard-based random number generators, there must be dynamic/active, real-time monitoring of the output with a sample size large enough to allow for reasonably high statistically powerful testing so that game play is disabled when an output testing failure is detected. (f) If the interactive gaming system utilizes a software-based random number generator, it must adhere to all of the following: (1) The period of the random number generator, in conjunction with the methods of implementing the random number generator outcomes, must be sufficiently large to ensure that all game independent outcome combinations/permutations are possible for the given game/application, unless the submission has been approved for a persistent-state outcome determination. (2) The methods of seeding/reseeding must ensure that all seed values are determined in a manner that does not compromise the cryptographic security of the random number generator. (3) To ensure that random number generator outcomes cannot be predicted, adequate background cycling/activity must be implemented in between games. Whenever a game outcome is made up of multiple mapped random number generator values, background cycling/activity must be implemented during the game (that is, in between the selection of each mapped random number generator value) to ensure that the game outcome is not comprised of sequential mapped random number generator outcomes. The rate of background cycling/activity must be sufficiently random in and of itself to prevent prediction.