Dynamic Temperature Management of Near-Sensor Processing for Energy-Efficient High-Fidelity Imaging
Vision processing on traditional architectures is inefficient due to energy-expensive off-chip data movement. Many researchers advocate pushing processing close to the sensor to substantially reduce data movement. However, continuous near-sensor processing raises sensor temperature, impairing imagin...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2021-01-01
|
Series: | Sensors |
Subjects: | |
Online Access: | https://www.mdpi.com/1424-8220/21/3/926 |
_version_ | 1797405860019830784 |
---|---|
author | Venkatesh Kodukula Saad Katrawala Britton Jones Carole-Jean Wu Robert LiKamWa |
author_facet | Venkatesh Kodukula Saad Katrawala Britton Jones Carole-Jean Wu Robert LiKamWa |
author_sort | Venkatesh Kodukula |
collection | DOAJ |
description | Vision processing on traditional architectures is inefficient due to energy-expensive off-chip data movement. Many researchers advocate pushing processing close to the sensor to substantially reduce data movement. However, continuous near-sensor processing raises sensor temperature, impairing imaging/vision fidelity. We characterize the thermal implications of using 3D stacked image sensors with near-sensor vision processing units. Our characterization reveals that near-sensor processing reduces system power but degrades image quality. For reasonable image fidelity, the sensor temperature needs to stay below a threshold, situationally determined by application needs. Fortunately, our characterization also identifies opportunities—unique to the needs of near-sensor processing—to regulate temperature based on dynamic visual task requirements and rapidly increase capture quality on demand. Based on our characterization, we propose and investigate two thermal management strategies—stop-capture-go and seasonal migration—for imaging-aware thermal management. For our evaluated tasks, our policies save up to 53% of system power with negligible performance impact and sustained image fidelity. |
first_indexed | 2024-03-09T03:16:21Z |
format | Article |
id | doaj.art-d015654ecdb34fe7a20a4a27ef8c3c88 |
institution | Directory Open Access Journal |
issn | 1424-8220 |
language | English |
last_indexed | 2024-03-09T03:16:21Z |
publishDate | 2021-01-01 |
publisher | MDPI AG |
record_format | Article |
series | Sensors |
spelling | doaj.art-d015654ecdb34fe7a20a4a27ef8c3c882023-12-03T15:20:33ZengMDPI AGSensors1424-82202021-01-0121392610.3390/s21030926Dynamic Temperature Management of Near-Sensor Processing for Energy-Efficient High-Fidelity ImagingVenkatesh Kodukula0Saad Katrawala1Britton Jones2Carole-Jean Wu3Robert LiKamWa4School of Electrical, Computer and Energy Engineering, Arizona State University, Tempe, AZ 85281, USASchool of Electrical, Computer and Energy Engineering, Arizona State University, Tempe, AZ 85281, USASchool of Electrical, Computer and Energy Engineering, Arizona State University, Tempe, AZ 85281, USASchool of Electrical, Computer and Energy Engineering, Arizona State University, Tempe, AZ 85281, USASchool of Electrical, Computer and Energy Engineering, Arizona State University, Tempe, AZ 85281, USAVision processing on traditional architectures is inefficient due to energy-expensive off-chip data movement. Many researchers advocate pushing processing close to the sensor to substantially reduce data movement. However, continuous near-sensor processing raises sensor temperature, impairing imaging/vision fidelity. We characterize the thermal implications of using 3D stacked image sensors with near-sensor vision processing units. Our characterization reveals that near-sensor processing reduces system power but degrades image quality. For reasonable image fidelity, the sensor temperature needs to stay below a threshold, situationally determined by application needs. Fortunately, our characterization also identifies opportunities—unique to the needs of near-sensor processing—to regulate temperature based on dynamic visual task requirements and rapidly increase capture quality on demand. Based on our characterization, we propose and investigate two thermal management strategies—stop-capture-go and seasonal migration—for imaging-aware thermal management. For our evaluated tasks, our policies save up to 53% of system power with negligible performance impact and sustained image fidelity.https://www.mdpi.com/1424-8220/21/3/926thermal managementimage sensorsfidelitycontinuous mobile vision |
spellingShingle | Venkatesh Kodukula Saad Katrawala Britton Jones Carole-Jean Wu Robert LiKamWa Dynamic Temperature Management of Near-Sensor Processing for Energy-Efficient High-Fidelity Imaging Sensors thermal management image sensors fidelity continuous mobile vision |
title | Dynamic Temperature Management of Near-Sensor Processing for Energy-Efficient High-Fidelity Imaging |
title_full | Dynamic Temperature Management of Near-Sensor Processing for Energy-Efficient High-Fidelity Imaging |
title_fullStr | Dynamic Temperature Management of Near-Sensor Processing for Energy-Efficient High-Fidelity Imaging |
title_full_unstemmed | Dynamic Temperature Management of Near-Sensor Processing for Energy-Efficient High-Fidelity Imaging |
title_short | Dynamic Temperature Management of Near-Sensor Processing for Energy-Efficient High-Fidelity Imaging |
title_sort | dynamic temperature management of near sensor processing for energy efficient high fidelity imaging |
topic | thermal management image sensors fidelity continuous mobile vision |
url | https://www.mdpi.com/1424-8220/21/3/926 |
work_keys_str_mv | AT venkateshkodukula dynamictemperaturemanagementofnearsensorprocessingforenergyefficienthighfidelityimaging AT saadkatrawala dynamictemperaturemanagementofnearsensorprocessingforenergyefficienthighfidelityimaging AT brittonjones dynamictemperaturemanagementofnearsensorprocessingforenergyefficienthighfidelityimaging AT carolejeanwu dynamictemperaturemanagementofnearsensorprocessingforenergyefficienthighfidelityimaging AT robertlikamwa dynamictemperaturemanagementofnearsensorprocessingforenergyefficienthighfidelityimaging |