Using MOPSO for Optimizing Randomized Response Schemes in Privacy Computing

Algorithm 2

Mutation for discrete MOPSO.

Input: ARC and mutation percentage P

Output: New particles.

Step  1. Randomized mutation

Select particles from ARC according to percentage P;

Each selected particle is re-randomized;

columns in each particle are replaced by newly generated columns which still

satisfy constraints as in Algorithm 1;

Calculate fitness for each newly generated particle.

Step  2. Guided mutation

Select the remaining particles from ARC;

Each particle is mutated by the guidance of GBEST;

columns in each particle are replaced by corresponding columns in

GBEST which still satisfy constraints as in Algorithm 1;

Calculate fitness for each newly generated particle.