Random Number Generation
[Framework]

This module generates random numbers. More...

Collaboration diagram for Random Number Generation:


Files
file	fip_rand.c
	Random Number Generation.
Functions
void	fip_srand (uint32_t seed)
	Seed the random number with a certain seed.
uint8_t	fip_rand8Limited (uint8_t max)
	Get a random number between uniformly distributed between 0 to max.
uint8_t	fip_rand8 (void)
	Get a random number between uniformly distributed between 0 to UINT8_MAX.
uint16_t	fip_rand16Limited (uint16_t max)
	Get a random number between uniformly distributed between 0 to max.
uint16_t	fip_rand16 (void)
	Get a random number between uniformly distributed between 0 to UINT16_MAX.
uint32_t	fip_rand32Limited (uint32_t max)
	Get a random number between uniformly distributed between 0 to max.
uint32_t	fip_rand32 (void)
	Get a random number between uniformly distributed between 0 to UINT32_MAX.
Variables
static uint32_t	fip_random_number = 1u
	This number stores the 32-bit random number in use on the system.
static uint32_t *const	fip_random_number32 = (uint32_t *)&fip_random_number
	Access the 32-bit random number as a 32-bit number (for read only).
static uint16_t *const	fip_random_number16 = (uint16_t *)&fip_random_number
	Access the 32-bit random number as a 16-bit number (for read only).
static uint8_t *const	fip_random_number8 = (uint8_t *)&fip_random_number
	Access the 32-bit random number as a 8-bit number (for read only).

Detailed Description

This module generates random numbers.

More importantly it does a good job of generating random numbers between user-defined spans, which is often required in network protocols, but is mostly done wrong. For example to scale a random number between 0 - 20000, you might just use something like:

randNumber = rand() % 20000;

Unfortunately this tends to drastically destroy the distribution of the numbers, which can have unexpected results when used with network protocols. For example the following graph shows the result of taking an otherwise perfectly flat distribution between 0-32768, then subjecting it to % 20000:

Notice how much more likely smaller numbers are (almost twice as likely). Results change depending on the value of X in rand() % X.

Using the functions in this module instead result in the expected distribution. Various functions are provided to provide a whole range of random number sizes in both limited (where the MAX value is user-defined) and normal (where the MAX value is defined based on the maximum value that can be represented in an integer of a certain bit-size).