Low-Power Communication Method using On-Device Deep Neural Network for Low-Power Image Recognition System 


Vol. 44,  No. 8, pp. 1588-1596, Aug.  2019
10.7840/kics.2019.44.8.1588


PDF
  Abstract

In recent years, low-power image recognition system has been actively studied with the evolution of IoT technology. In order to manipulate a deep neural network, which has high accuracy for image recognition, the device only captures images and transmits the image to the server and then performs image recognition on the server. However, there is a problem in that power consumption is large due to communication for continuous image transmission. In this paper, we propose low-power communication method using on-device deep neural network to solve the communication power consumption problem of image recognition systems. Using on-device deep neural network from the constrained resource device, we optimized the power consumption by transmitting the image only when the image recognition result is the image of interest. Also, we propose a power consumption model for the proposed method. And, we evaluate the power consumption and lifetime of the proposed method compared to baseline method. Experimental results show that the proposed method reduces power consumption by 54.4% compared to the baseline method.

  Statistics
Cumulative Counts from November, 2022
Multiple requests among the same browser session are counted as one view. If you mouse over a chart, the values of data points will be shown.


  Cite this article

[IEEE Style]

J. Lim and Y. Baek, "Low-Power Communication Method using On-Device Deep Neural Network for Low-Power Image Recognition System," The Journal of Korean Institute of Communications and Information Sciences, vol. 44, no. 8, pp. 1588-1596, 2019. DOI: 10.7840/kics.2019.44.8.1588.

[ACM Style]

Jaebong Lim and Yunju Baek. 2019. Low-Power Communication Method using On-Device Deep Neural Network for Low-Power Image Recognition System. The Journal of Korean Institute of Communications and Information Sciences, 44, 8, (2019), 1588-1596. DOI: 10.7840/kics.2019.44.8.1588.

[KICS Style]

Jaebong Lim and Yunju Baek, "Low-Power Communication Method using On-Device Deep Neural Network for Low-Power Image Recognition System," The Journal of Korean Institute of Communications and Information Sciences, vol. 44, no. 8, pp. 1588-1596, 8. 2019. (https://doi.org/10.7840/kics.2019.44.8.1588)

Search
(IN TITLE, AUTHOR, ABSTRACT,KEYWORDS)

Advanced Search ▼

POPULAR KEYWORDS
(TOP 10 KEYWORDS)

Recent Publications
(LAST 3 YEARS)






pISSN : 1226 - 4717
eISSN : 2287 - 3880