Algorithm 1: Genetic algorithm-based device association strategy

1:Input: Device set $\mathcal{I}$, UAV set $\mathcal{U}$, ground 5G base station set $\mathcal{B}$ and channel state information. 2:Output: Optimal device association strategy $\{{\mathbf{x}}_{\mathrm{u}}\left(t\right),{\mathbf{x}}_{\mathrm{b}}\left(t\right)\}$. 3:Initialize: Population size $N_{number}$, crossover probability $p_{Crossover}$, mutation probability $p_{Mutation}$, maximum number of iterations $N_{iteration}$ and $f=1$. 4:Random generate initial population $J_f$. 5:repeat 6:for each population individual in $J_f$ do 7:   Calculate the fitness function using (12). 8:end for 9:Selective manipulation of individuals using elitist retention strategies and roulette wheel strategies. 10:Replication of elite individuals into the next generation of populations. 11:For all parental chromosomes that are selected, a set of two is used, and each set is crossed over according to the crossover probability $p_{Crossover}$ to generate new offspring to be added to the next generation population. 12:According to $p_{Mutation}$, genes in the chromosomes of non-elite individuals of the parent are mutated, and the mutated individuals are inserted into the next generation population. 13:Combining lines 8–11 yields the new generation of populations $J_{f+1}$. 14:$f=f+1$. 15:until $f>N_{iteration}$. 16:Return $\{{\mathbf{x}}_{\mathrm{u}}\left(t\right),{\mathbf{x}}_{\mathrm{b}}\left(t\right)\}$.