Soliton Distribution

This small Java project implements a random number generator according to the soliton distribution. The distribution, introduced by Michael Luby in his paper "LT Codes", has two forms, one known as the ideal soliton distribution and the other being the robust soliton distribution. Both distributions are implemented in this project.

Ideal Soliton Distribution

The ideal soliton distribution has only one parameter, namely the number of blocks of a file. The following small piece of Java code shows how to set up a random number generator according to this distribution.

int nrBlocks = 1000;
SolitonGenerator srng = new IdealSolitonGenerator(nrBlocks);
srng.next();

Robust Soliton Distribution

This distribution is implemented by the class RobustSolitonGenerator and has two ways of creating random numbers. In the original definition the robust soliton distribution has three parameters:

• number of blocks
• the admissible failure probability
• a strictly positive constant c

The constant c influences the position of a spike in the distribution. Altough any value for c is possible, it is recommended to use values smaller than 1. The admissible failure probability ranges from 0 to 1, usually smaller values like 1% or 5% make sense. It follows a small example on how to use the distribution in Java:

int nrBlocks = 1000;
double failureProbability = 0.01;
double c = 0.2;
SolitonGenerator srng = new RobustSolitonGenerator(nrBlocks, c, failureProbability);
srng.next();

The other definition yields an equivalent distribution, however the parameters differ slightly:

• number of blocks
• the admissible failure probability
• the position of the spike

Instead of passing the constant c one can use the more intuitive parameter spike which describes the position of the spike in the distribution.

int nrBlocks = 1000;
double failureProbability = 0.01;
int spike = 40;
SolitonGenerator srng = new RobustSolitonGenerator(nrBlocks, spike, failureProbability);
srng.next();