Require: num_federated_rounds T

1: procedure AGGREGATING

2: Initialise global model: W(0)

3: for t ← 1 ⋯ T do

4: for client k ← 1 ⋯ K do  ⊳ Run in parallel

5: Send W(t−1) to client k

6: Receive model updates and number of local training iterations $\left(\mathrm{\Delta }{W}_k^{\left(t-1\right)},N_k\right)$ from client’s local training with ${\mathcal{L}}_k\left(X_k;W^{\left(t-1\right)}\right)$

7: end for

8: $W^{\left(t\right)}\leftarrow W^{\left(t-1\right)}+\frac{1}{{\sum }_k{N}_k}{\sum }_k\left(N_k\cdot W_k^{\left(t-1\right)}\right)$

9: end for

10: return W(t)

11: end procedure