I have 24 cluster heads and 250 nodes. The population is randomly generate between 0 to 1( if value>0.6 then 1 else 0). If the value is 1cluster head is activated else not activated. The activated cluster heads form clusters based on distance( The minimum more than 17 clusters are activated because the coverage is 1600 and uniform distribution of nodes). The fitness of network calculated by using summation each cluster performance( how many nodes connected to cluster head divided by max allowed like 18/30) is divided by a number of clusters activated. The similar scenario figure attached below.

Here update the velocity by Velocity(i+1)=w*Velocity(i)+c1*rand()*(pBest(i)-Position(i))+c2*rand()*(gBest(i)-Position(i)); c1,c2=2.

s=1/(1+exp(-Velocity(i+1))); if rand()