State-of-the-art performance for many emerging edge applications is achieved by deep neural networks (DNNs). Often, the employed DNNs are location- and time-dependent, and the parameters of a specific DNN must be delivered from an edge server to the edge device rapidly and efficiently to carry out time-sensitive inference tasks. This can be considered as a joint source-channel coding (JSCC) problem, in which the goal is not to recover the DNN coefficients with the minimal distortion, but in a manner that provides the highest accuracy in the downstream task. For this purpose we introduce AirNet, a novel training and analog transmission method to deliver DNNs over the air. We first train the DNN with noise injection to counter the wireless channel noise. We also employ pruning to identify the most significant DNN parameters that can be delivered within the available channel bandwidth, knowledge distillation, and nonlinear bandwidth expansion to provide better error protection for the most important network parameters. We show that AirNet achieves significantly higher test accuracy compared to the separation-based alternative, and exhibits graceful degradation with channel quality.

AirNet: Neural Network Transmission over the Air / Jankowski, M.; Gunduz, D.; Mikolajczyk, K.. - 2022-:(2022), pp. 2451-2456. (Intervento presentato al convegno 2022 IEEE International Symposium on Information Theory, ISIT 2022 tenutosi a fin nel 2022) [10.1109/ISIT50566.2022.9834372].

AirNet: Neural Network Transmission over the Air

Gunduz D.;
2022

Abstract

State-of-the-art performance for many emerging edge applications is achieved by deep neural networks (DNNs). Often, the employed DNNs are location- and time-dependent, and the parameters of a specific DNN must be delivered from an edge server to the edge device rapidly and efficiently to carry out time-sensitive inference tasks. This can be considered as a joint source-channel coding (JSCC) problem, in which the goal is not to recover the DNN coefficients with the minimal distortion, but in a manner that provides the highest accuracy in the downstream task. For this purpose we introduce AirNet, a novel training and analog transmission method to deliver DNNs over the air. We first train the DNN with noise injection to counter the wireless channel noise. We also employ pruning to identify the most significant DNN parameters that can be delivered within the available channel bandwidth, knowledge distillation, and nonlinear bandwidth expansion to provide better error protection for the most important network parameters. We show that AirNet achieves significantly higher test accuracy compared to the separation-based alternative, and exhibits graceful degradation with channel quality.
2022
2022 IEEE International Symposium on Information Theory, ISIT 2022
fin
2022
2022-
2451
2456
Jankowski, M.; Gunduz, D.; Mikolajczyk, K.
AirNet: Neural Network Transmission over the Air / Jankowski, M.; Gunduz, D.; Mikolajczyk, K.. - 2022-:(2022), pp. 2451-2456. (Intervento presentato al convegno 2022 IEEE International Symposium on Information Theory, ISIT 2022 tenutosi a fin nel 2022) [10.1109/ISIT50566.2022.9834372].
File in questo prodotto:
File Dimensione Formato  
AirNet_Neural_Network_Transmission_over_the_Air.pdf

Accesso riservato

Tipologia: Versione pubblicata dall'editore
Dimensione 1.27 MB
Formato Adobe PDF
1.27 MB Adobe PDF   Visualizza/Apri   Richiedi una copia
2105.11166.pdf

Open access

Tipologia: Versione originale dell'autore proposta per la pubblicazione
Dimensione 555.73 kB
Formato Adobe PDF
555.73 kB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

Licenza Creative Commons
I metadati presenti in IRIS UNIMORE sono rilasciati con licenza Creative Commons CC0 1.0 Universal, mentre i file delle pubblicazioni sono rilasciati con licenza Attribuzione 4.0 Internazionale (CC BY 4.0), salvo diversa indicazione.
In caso di violazione di copyright, contattare Supporto Iris

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11380/1286029
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 13
  • ???jsp.display-item.citation.isi??? ND
social impact