Machine Learning-Based Methods for Enhancement of UAV-NOMA and D2D Cooperative Networks

. 2023 Mar 10;23(6):3014. doi: 10.3390/s23063014

Algorithm 1 2D UFBS placement process through the k-means algorithm

1:
input: The set of coordinates of all GMTs $U = {u_{1}, u_{2}, \dots, u_{N}}$ , and the number of UFBSs Y
2:
$ϵ = 10^{- 6}$
3:
t = 0
4:
Initialize $Y$ centroid points $C^{t} = \{p_{1}^{c}, p_{2}^{c}, . . ., p_{Y}^{c}\} \subseteq U$ , randomly
5:
repeat
6:
$S_{k}$ = $\emptyset, \forall_{k} = 1, 2, \dots, Y$
7:
for $i \leftarrow 1$ to N do
8:
$k^{*} = \underset{k = 1 \dots Y}{arg min} ∥ u_{i} - C_{k}^{t} ∥$
9:
$S_{k^{*}} = S_{k^{*}} \cup \{u_{i}\}$
10:
end for
11:
for $k \leftarrow 1$ to $Y$ do
12:
$C_{k}^{t} = \frac{1}{| S_{k} |} \sum_{u_{l} \in S_{k}} u_{l}$ .
13:
end for
14:
t = t +1
15:
until $C^{t} - C^{t - 1} \leq ϵ$
16:
output: A set of centroid points that the $Y$ UFBSs will be deployed $C^{t}$ .