Semantic Communication Enabling Robust Edge Intelligence for Time-Critical IoT Applications

Andrea Cavagna; Nan Li; Alexandros Iosifidis; Qi Zhang

doi:10.1109/ICCWorkshops57953.2023.10283786

Semantic Communication Enabling Robust Edge Intelligence for Time-Critical IoT Applications

Andrea Cavagna, Nan Li, Alexandros Iosifidis, Qi Zhang

Research output: Contribution to book/anthology/report/proceeding › Article in proceedings › Research › peer-review

Abstract

This paper aims to design robust Edge Intelligence using semantic communication for time-critical IoT applications. We systematically analyze the effect of image DCT coefficients on inference accuracy and propose the channel-agnostic effectiveness encoding for offloading by transmitting the most meaningful task data first. This scheme can well utilize all available communication resource and strike a balance between transmission latency and inference accuracy. Then, we design an effectiveness decoding by implementing a novel image augmentation process for convolutional neural network (CNN) training, through which an original CNN model is transformed into a Robust CNN model. We use the proposed training method to generate Robust MobileNet-v2 and Robust ResNet-50. The proposed Edge Intelligence framework consists of the proposed effectiveness encoding and effectiveness decoding. The experimental results show that the effectiveness decoding using the Robust CNN models perform consistently better under various image distortions caused by channel errors or limited communication resource. The proposed Edge Intelligence framework using semantic communication significantly outperforms the conventional approach under latency and data rate constraints, in particular, under ultra stringent deadlines and low data rate.

Original language	English
Title of host publication	2023 IEEE International Conference on Communications Workshops (ICC Workshops)
Number of pages	6
Publisher	IEEE
Publication date	Oct 2023
Pages	1617-1622
ISBN (Electronic)	979-8-3503-3307-7, 979-8-3503-3308-4
DOIs	https://doi.org/10.1109/ICCWorkshops57953.2023.10283786
Publication status	Published - Oct 2023
Event	2nd International Workshop on Semantic Communications - Rome, Italy Duration: 28 May 2023 → 1 Jun 2023

Conference

Conference	2nd International Workshop on Semantic Communications
Country/Territory	Italy
City	Rome
Period	28/05/2023 → 01/06/2023

Series	IEEE International Conference on Communications workshops
ISSN	2694-2941

Keywords

CNN inference
Edge Intelligence
Semantic communication
low latency
service reliability

Access to Document

10.1109/ICCWorkshops57953.2023.10283786

https://arxiv.org/pdf/2211.13787Licence: Unspecified

Cite this

@inproceedings{35a9681967f44f4ab5738d4eb6ad7dc1,

title = "Semantic Communication Enabling Robust Edge Intelligence for Time-Critical IoT Applications",

abstract = "This paper aims to design robust Edge Intelligence using semantic communication for time-critical IoT applications. We systematically analyze the effect of image DCT coefficients on inference accuracy and propose the channel-agnostic effectiveness encoding for offloading by transmitting the most meaningful task data first. This scheme can well utilize all available communication resource and strike a balance between transmission latency and inference accuracy. Then, we design an effectiveness decoding by implementing a novel image augmentation process for convolutional neural network (CNN) training, through which an original CNN model is transformed into a Robust CNN model. We use the proposed training method to generate Robust MobileNet-v2 and Robust ResNet-50. The proposed Edge Intelligence framework consists of the proposed effectiveness encoding and effectiveness decoding. The experimental results show that the effectiveness decoding using the Robust CNN models perform consistently better under various image distortions caused by channel errors or limited communication resource. The proposed Edge Intelligence framework using semantic communication significantly outperforms the conventional approach under latency and data rate constraints, in particular, under ultra stringent deadlines and low data rate.",

keywords = "CNN inference, Edge Intelligence, Semantic communication, low latency, service reliability",

author = "Andrea Cavagna and Nan Li and Alexandros Iosifidis and Qi Zhang",

year = "2023",

month = oct,

doi = "10.1109/ICCWorkshops57953.2023.10283786",

language = "English",

series = "IEEE International Conference on Communications workshops",

publisher = "IEEE",

pages = "1617--1622",

booktitle = "2023 IEEE International Conference on Communications Workshops (ICC Workshops)",

note = " 2nd International Workshop on Semantic Communications ; Conference date: 28-05-2023 Through 01-06-2023",

}

Cavagna, A, Li, N, Iosifidis, A & Zhang, Q 2023, Semantic Communication Enabling Robust Edge Intelligence for Time-Critical IoT Applications. in 2023 IEEE International Conference on Communications Workshops (ICC Workshops). IEEE, IEEE International Conference on Communications workshops, pp. 1617-1622, 2nd International Workshop on Semantic Communications, Rome, Italy, 28/05/2023. https://doi.org/10.1109/ICCWorkshops57953.2023.10283786

Semantic Communication Enabling Robust Edge Intelligence for Time-Critical IoT Applications. / Cavagna, Andrea; Li, Nan; Iosifidis, Alexandros et al.
2023 IEEE International Conference on Communications Workshops (ICC Workshops). IEEE, 2023. p. 1617-1622 (IEEE International Conference on Communications workshops).

Research output: Contribution to book/anthology/report/proceeding › Article in proceedings › Research › peer-review

TY - GEN

T1 - Semantic Communication Enabling Robust Edge Intelligence for Time-Critical IoT Applications

AU - Cavagna, Andrea

AU - Li, Nan

AU - Iosifidis, Alexandros

AU - Zhang, Qi

PY - 2023/10

Y1 - 2023/10

N2 - This paper aims to design robust Edge Intelligence using semantic communication for time-critical IoT applications. We systematically analyze the effect of image DCT coefficients on inference accuracy and propose the channel-agnostic effectiveness encoding for offloading by transmitting the most meaningful task data first. This scheme can well utilize all available communication resource and strike a balance between transmission latency and inference accuracy. Then, we design an effectiveness decoding by implementing a novel image augmentation process for convolutional neural network (CNN) training, through which an original CNN model is transformed into a Robust CNN model. We use the proposed training method to generate Robust MobileNet-v2 and Robust ResNet-50. The proposed Edge Intelligence framework consists of the proposed effectiveness encoding and effectiveness decoding. The experimental results show that the effectiveness decoding using the Robust CNN models perform consistently better under various image distortions caused by channel errors or limited communication resource. The proposed Edge Intelligence framework using semantic communication significantly outperforms the conventional approach under latency and data rate constraints, in particular, under ultra stringent deadlines and low data rate.

AB - This paper aims to design robust Edge Intelligence using semantic communication for time-critical IoT applications. We systematically analyze the effect of image DCT coefficients on inference accuracy and propose the channel-agnostic effectiveness encoding for offloading by transmitting the most meaningful task data first. This scheme can well utilize all available communication resource and strike a balance between transmission latency and inference accuracy. Then, we design an effectiveness decoding by implementing a novel image augmentation process for convolutional neural network (CNN) training, through which an original CNN model is transformed into a Robust CNN model. We use the proposed training method to generate Robust MobileNet-v2 and Robust ResNet-50. The proposed Edge Intelligence framework consists of the proposed effectiveness encoding and effectiveness decoding. The experimental results show that the effectiveness decoding using the Robust CNN models perform consistently better under various image distortions caused by channel errors or limited communication resource. The proposed Edge Intelligence framework using semantic communication significantly outperforms the conventional approach under latency and data rate constraints, in particular, under ultra stringent deadlines and low data rate.

KW - CNN inference

KW - Edge Intelligence

KW - Semantic communication

KW - low latency

KW - service reliability

UR - http://www.scopus.com/inward/record.url?scp=85177830978&partnerID=8YFLogxK

U2 - 10.1109/ICCWorkshops57953.2023.10283786

DO - 10.1109/ICCWorkshops57953.2023.10283786

M3 - Article in proceedings

T3 - IEEE International Conference on Communications workshops

SP - 1617

EP - 1622

BT - 2023 IEEE International Conference on Communications Workshops (ICC Workshops)

PB - IEEE

T2 - 2nd International Workshop on Semantic Communications

Y2 - 28 May 2023 through 1 June 2023

ER -

Semantic Communication Enabling Robust Edge Intelligence for Time-Critical IoT Applications

Abstract

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this