1: \begin{abstract}

2: Over-the-air federated learning (OTA-FL) has emerged as an efficient mechanism that exploits the superposition property of the wireless medium and performs model aggregation for federated learning in the air.

3: OTA-FL is naturally sensitive to wireless channel fading, which could significantly diminish its learning accuracy.

4: To address this challenge, in this paper, we propose an OTA-FL algorithm called \alg (\ul{ch}annel-quality-\ul{a}ware ove\ul{r}-the-air \ul{l}ocal \ul{e}stimating and \ul{s}caling).

5: Our \alg algorithm performs channel state information (CSI) estimation and adaptive scaling to mitigate the impacts of wireless channel fading.

6: We establish the theoretical convergence rate performance of \alg and analyze the impacts of CSI error on the convergence of \algns. 

7: We show that the adaptive channel inversion scaling scheme in \alg is robust under imperfect CSI scenarios.

8: We also demonstrate through numerical results that \alg outperforms existing OTA-FL algorithms with heterogeneous data under imperfect CSI.

9: %We also demonstrate through numerical results that \alg achieves outperforms existing OTA-FL algorithms with heterogeneous data under imperfect CSI.

10: \end{abstract}

11: