{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,7]],"date-time":"2026-07-07T15:51:10Z","timestamp":1783439470259,"version":"3.54.6"},"reference-count":42,"publisher":"Association for Computing Machinery (ACM)","issue":"5s","license":[{"start":{"date-parts":[[2021,9,17]],"date-time":"2021-09-17T00:00:00Z","timestamp":1631836800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"National Science Foundation","award":["CMMI-1739503"],"award-info":[{"award-number":["CMMI-1739503"]}]},{"name":"Graduate Assistance in Areas of National Need","award":["P200A180052"],"award-info":[{"award-number":["P200A180052"]}]}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":["ACM Trans. Embed. Comput. Syst."],"published-print":{"date-parts":[[2021,10,31]]},"abstract":"<jats:p>\n            Autonomous vehicles (AV) are expected to revolutionize transportation and improve road safety significantly. However, these benefits do not come without cost; AVs require large Deep-Learning (DL) models and powerful hardware platforms to operate reliably in real-time, requiring between several hundred watts to one kilowatt of power. This power consumption can dramatically reduce vehicles\u2019 driving range and affect emissions. To address this problem, we propose SAGE: a methodology for selectively offloading the key energy-consuming modules of DL architectures to the cloud to optimize edge, energy usage while meeting real-time latency constraints. Furthermore, we leverage Head Network Distillation (HND) to introduce efficient\n            <jats:italic>bottlenecks<\/jats:italic>\n            within the DL architecture in order to minimize the network overhead costs of offloading with almost no degradation in the model\u2019s performance. We evaluate SAGE using an Nvidia Jetson TX2 and an industry-standard Nvidia Drive PX2 as the AV edge, devices and demonstrate that our offloading strategy is practical for a wide range of DL models and internet connection bandwidths on 3G, 4G LTE, and WiFi technologies. Compared to edge-only computation, SAGE reduces energy consumption by an average of\n            <jats:bold>36.13%<\/jats:bold>\n            ,\n            <jats:bold>47.07%<\/jats:bold>\n            , and\n            <jats:bold>55.66%<\/jats:bold>\n            for an AV with one low-resolution camera, one high-resolution camera, and three high-resolution cameras, respectively. SAGE also reduces upload data size by up to\n            <jats:bold>98.40%<\/jats:bold>\n            compared to direct camera offloading.\n          <\/jats:p>","DOI":"10.1145\/3477006","type":"journal-article","created":{"date-parts":[[2021,9,17]],"date-time":"2021-09-17T18:36:51Z","timestamp":1631903811000},"page":"1-22","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":23,"title":["SAGE: A Split-Architecture Methodology for Efficient End-to-End Autonomous Vehicle Control"],"prefix":"10.1145","volume":"20","author":[{"given":"Arnav","family":"Malawade","sequence":"first","affiliation":[{"name":"University of California Irvine, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Mohanad","family":"Odema","sequence":"additional","affiliation":[{"name":"University of California Irvine, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sebastien","family":"Lajeunesse-degroot","sequence":"additional","affiliation":[{"name":"University of California Irvine, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5390-0497","authenticated-orcid":false,"given":"Mohammad Abdullah","family":"Al Faruque","sequence":"additional","affiliation":[{"name":"University of California Irvine, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"320","published-online":{"date-parts":[[2021,9,17]]},"reference":[{"key":"e_1_2_1_1_1","unstructured":"2016. All new Teslas are equipped with NVIDIA\u2019s new Drive PX 2 AI platform for self-driving-Electrek. https:\/\/electrek.co\/2016\/10\/21\/all-new-teslas-are-equ ipped-with-nvidias-new-drive-px-2-ai-platform-for-self-driving. (Oct 2016). [Online; accessed 9. Nov. 2020].  2016. All new Teslas are equipped with NVIDIA\u2019s new Drive PX 2 AI platform for self-driving-Electrek. https:\/\/electrek.co\/2016\/10\/21\/all-new-teslas-are-equ ipped-with-nvidias-new-drive-px-2-ai-platform-for-self-driving. (Oct 2016). [Online; accessed 9. Nov. 2020]."},{"key":"e_1_2_1_2_1","volume-title":"Nvidia cranks up and turns down its drive AGX orin computers. Forbes (Jun","author":"Abuelsamid Sam","year":"2020","unstructured":"Sam Abuelsamid . 2020. Nvidia cranks up and turns down its drive AGX orin computers. Forbes (Jun 2020 ). https:\/\/www.forbes.com\/sites\/samabuelsamid\/2020\/05\/14\/nvidia-cranks-up-an d-turns-down-its-drive-agx-orin-computers. Sam Abuelsamid. 2020. Nvidia cranks up and turns down its drive AGX orin computers. Forbes (Jun 2020). https:\/\/www.forbes.com\/sites\/samabuelsamid\/2020\/05\/14\/nvidia-cranks-up-an d-turns-down-its-drive-agx-orin-computers."},{"key":"e_1_2_1_3_1","volume-title":"Energy management-as-a-service over fog computing platform","author":"Al Faruque Mohammad Abdullah","year":"2015","unstructured":"Mohammad Abdullah Al Faruque and Korosh Vatanparvar . 2015. Energy management-as-a-service over fog computing platform . IEEE internet of things journal 3, 2 ( 2015 ), 161\u2013169. Mohammad Abdullah Al Faruque and Korosh Vatanparvar. 2015. Energy management-as-a-service over fog computing platform. IEEE internet of things journal 3, 2 (2015), 161\u2013169."},{"key":"e_1_2_1_4_1","volume-title":"Retrieved","author":"ARM.","year":"2021","unstructured":"ARM. arm npu Ethos-77 . Retrieved April , 2021 from https:\/\/www.arm.com\/products\/silicon-ip-cpu\/ethos\/ethos-n77. ARM. arm npu Ethos-77. Retrieved April, 2021 from https:\/\/www.arm.com\/products\/silicon-ip-cpu\/ethos\/ethos-n77."},{"key":"e_1_2_1_5_1","doi-asserted-by":"publisher","DOI":"10.5555\/2969033.2969123"},{"key":"e_1_2_1_6_1","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV.2019.00489"},{"key":"e_1_2_1_7_1","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-01267-0_15"},{"key":"e_1_2_1_8_1","doi-asserted-by":"publisher","DOI":"10.1109\/ICRA.2018.8460487"},{"key":"e_1_2_1_9_1","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV.2019.00942"},{"key":"e_1_2_1_10_1","doi-asserted-by":"publisher","DOI":"10.1109\/JIOT.2020.3001218"},{"key":"e_1_2_1_11_1","volume-title":"Conference on robot learning. PMLR, 1\u201316","author":"Dosovitskiy Alexey","year":"2017","unstructured":"Alexey Dosovitskiy , German Ros , Felipe Codevilla , Antonio Lopez , and Vladlen Koltun . 2017 . CARLA: An open urban driving simulator . In Conference on robot learning. PMLR, 1\u201316 . Alexey Dosovitskiy, German Ros, Felipe Codevilla, Antonio Lopez, and Vladlen Koltun. 2017. CARLA: An open urban driving simulator. In Conference on robot learning. PMLR, 1\u201316."},{"key":"e_1_2_1_12_1","doi-asserted-by":"publisher","DOI":"10.1109\/VETECF.2007.461"},{"key":"e_1_2_1_13_1","volume-title":"Mobile edge computing for the internet of vehicles: Offloading framework and job scheduling","author":"Feng Jingyun","year":"2018","unstructured":"Jingyun Feng , Zhi Liu , Celimuge Wu , and Yusheng Ji. 2018. Mobile edge computing for the internet of vehicles: Offloading framework and job scheduling . IEEE vehicular technology magazine 14, 1 ( 2018 ), 28\u201336. Jingyun Feng, Zhi Liu, Celimuge Wu, and Yusheng Ji. 2018. Mobile edge computing for the internet of vehicles: Offloading framework and job scheduling. IEEE vehicular technology magazine 14, 1 (2018), 28\u201336."},{"key":"e_1_2_1_14_1","volume-title":"L\u00f3pez","author":"Gonz\u00e1lez Alejandro","year":"2016","unstructured":"Alejandro Gonz\u00e1lez , Zhijie Fang , Yainuvis Socarras , Joan Serrat , David V\u00e1zquez , Jiaolong Xu , and Antonio M . L\u00f3pez . 2016 . Pedestrian detection at day\/night time with visible and FIR cameras: A comparison. Sensors 16, 6 (2016). Alejandro Gonz\u00e1lez, Zhijie Fang, Yainuvis Socarras, Joan Serrat, David V\u00e1zquez, Jiaolong Xu, and Antonio M. L\u00f3pez. 2016. Pedestrian detection at day\/night time with visible and FIR cameras: A comparison. Sensors 16, 6 (2016)."},{"key":"e_1_2_1_15_1","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.90"},{"key":"e_1_2_1_16_1","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-01234-2_27"},{"key":"e_1_2_1_17_1","volume-title":"NIPS Deep Learning and Representation Learning Workshop. http:\/\/arxiv.org\/abs\/1503","author":"Hinton Geoffrey","year":"2015","unstructured":"Geoffrey Hinton , Oriol Vinyals , and Jeffrey Dean . 2015 . Distilling the knowledge in a neural network . In NIPS Deep Learning and Representation Learning Workshop. http:\/\/arxiv.org\/abs\/1503 .02531. Geoffrey Hinton, Oriol Vinyals, and Jeffrey Dean. 2015. Distilling the knowledge in a neural network. In NIPS Deep Learning and Representation Learning Workshop. http:\/\/arxiv.org\/abs\/1503.02531."},{"key":"e_1_2_1_18_1","volume-title":"Weinberger","author":"Huang Gao","year":"2016","unstructured":"Gao Huang , Zhuang Liu , and Kilian Q . Weinberger . 2016 . Densely connected convolutional networks. CoRR abs\/1608. 06993 (2016). Gao Huang, Zhuang Liu, and Kilian Q. Weinberger. 2016. Densely connected convolutional networks. CoRR abs\/1608. 06993 (2016)."},{"key":"e_1_2_1_19_1","doi-asserted-by":"publisher","DOI":"10.1145\/2307636.2307658"},{"key":"e_1_2_1_20_1","volume-title":"Baidu unveils its camera-based Apollo Lite self-driving suite. Roadshow (Jun","author":"Hyatt Kyle","year":"2019","unstructured":"Kyle Hyatt . 2019. Baidu unveils its camera-based Apollo Lite self-driving suite. Roadshow (Jun 2019 ). https:\/\/www.cnet.com\/roadshow\/news\/baidu-apollo-lite-camera-based-self-dr iving. Kyle Hyatt. 2019. Baidu unveils its camera-based Apollo Lite self-driving suite. Roadshow (Jun 2019). https:\/\/www.cnet.com\/roadshow\/news\/baidu-apollo-lite-camera-based-self-dr iving."},{"key":"e_1_2_1_21_1","volume-title":"Argo gives its self-driving vehicle hardware a big upgrade. Roadshow (Jan","author":"Hyatt Kyle","year":"2021","unstructured":"Kyle Hyatt . 2021. Argo gives its self-driving vehicle hardware a big upgrade. Roadshow (Jan 2021 ). https:\/\/www.cnet.com\/roadshow\/news\/argo-self-driving-car-hardware-upgrade. Kyle Hyatt. 2021. Argo gives its self-driving vehicle hardware a big upgrade. Roadshow (Jan 2021). https:\/\/www.cnet.com\/roadshow\/news\/argo-self-driving-car-hardware-upgrade."},{"key":"e_1_2_1_22_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.dib.2020.105304"},{"key":"e_1_2_1_23_1","doi-asserted-by":"publisher","DOI":"10.1145\/3093337.3037698"},{"key":"e_1_2_1_24_1","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2020.3011705"},{"key":"e_1_2_1_25_1","volume-title":"On the deployment and noise filtering of vehicular radar application for detection enhancement in roads and tunnels. Sensors 18, 3","author":"Kim Young-Duk","year":"2018","unstructured":"Young-Duk Kim , Guk-Jin Son , Chan-Ho Song , and Hee-Kang Kim . 2018. On the deployment and noise filtering of vehicular radar application for detection enhancement in roads and tunnels. Sensors 18, 3 ( 2018 ). Young-Duk Kim, Guk-Jin Son, Chan-Ho Song, and Hee-Kang Kim. 2018. On the deployment and noise filtering of vehicular radar application for detection enhancement in roads and tunnels. Sensors 18, 3 (2018)."},{"key":"e_1_2_1_26_1","doi-asserted-by":"publisher","DOI":"10.1109\/JSYST.2019.2959703"},{"key":"e_1_2_1_27_1","volume-title":"Influence of waveform characteristics on LiDAR Ranging Accuracy and Precision. Sensors 18, 4","author":"Li Xiaolu","year":"2018","unstructured":"Xiaolu Li , Bingwei Yang , Xinhao Xie , Duan Li , and Lijun Xu. 2018. Influence of waveform characteristics on LiDAR Ranging Accuracy and Precision. Sensors 18, 4 ( 2018 ). Xiaolu Li, Bingwei Yang, Xinhao Xie, Duan Li, and Lijun Xu. 2018. Influence of waveform characteristics on LiDAR Ranging Accuracy and Precision. Sensors 18, 4 (2018)."},{"key":"e_1_2_1_28_1","doi-asserted-by":"publisher","DOI":"10.1145\/3296957.3173191"},{"key":"e_1_2_1_29_1","doi-asserted-by":"publisher","DOI":"10.1145\/3349614.3356022"},{"key":"e_1_2_1_30_1","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2020.3039714"},{"key":"e_1_2_1_31_1","doi-asserted-by":"crossref","unstructured":"Yoshitomo Matsubara and Marco Levorato. 2020. Neural Compression and Filtering for Edge-assisted Real-time Object Detection in Challenged Networks. (2020). arxiv:cs.CV\/2007.15818  Yoshitomo Matsubara and Marco Levorato. 2020. Neural Compression and Filtering for Edge-assisted Real-time Object Detection in Challenged Networks. (2020). arxiv:cs.CV\/2007.15818","DOI":"10.1109\/ICPR48806.2021.9412388"},{"key":"e_1_2_1_32_1","doi-asserted-by":"publisher","DOI":"10.1109\/DAC18074.2021.9586259"},{"key":"e_1_2_1_33_1","volume-title":"Feat. Post-Volta GPUs. AnandTech (Oct","author":"Nate Oh.","year":"2017","unstructured":"Nate Oh. 2017. NVIDIA Announces Drive PX Pegasus at GTC Europe 2017: Level 5 Self-Driving Hardware , Feat. Post-Volta GPUs. AnandTech (Oct 2017 ). https:\/\/www.anandtech.com\/show\/11913\/nvidia-announces-drive-px-pegasus-at -gtc-europe-2017-feat-nextgen-gpus. Nate Oh. 2017. NVIDIA Announces Drive PX Pegasus at GTC Europe 2017: Level 5 Self-Driving Hardware, Feat. Post-Volta GPUs. AnandTech (Oct 2017). https:\/\/www.anandtech.com\/show\/11913\/nvidia-announces-drive-px-pegasus-at -gtc-europe-2017-feat-nextgen-gpus."},{"key":"e_1_2_1_34_1","volume-title":"Cellular V2X as the essential enabler of superior global connected transportation services","author":"Papathanassiou Apostolos","year":"2017","unstructured":"Apostolos Papathanassiou and Alexey Khoryaev . 2017. Cellular V2X as the essential enabler of superior global connected transportation services . IEEE 5G Tech Focus 1, 2 ( 2017 ), 1\u20132. Apostolos Papathanassiou and Alexey Khoryaev. 2017. Cellular V2X as the essential enabler of superior global connected transportation services. IEEE 5G Tech Focus 1, 2 (2017), 1\u20132."},{"key":"e_1_2_1_35_1","doi-asserted-by":"publisher","DOI":"10.1109\/JETCAS.2020.3015889"},{"key":"e_1_2_1_36_1","doi-asserted-by":"publisher","DOI":"10.1109\/SICE.2016.7749210"},{"key":"e_1_2_1_37_1","volume-title":"A survey of end-to-end driving: Architectures and training methods","author":"Tampuu Ardi","year":"2020","unstructured":"Ardi Tampuu , Tambet Matiisen , Maksym Semikin , Dmytro Fishman , and Naveed Muhammad . 2020. A survey of end-to-end driving: Architectures and training methods . IEEE Transactions on Neural Networks and Learning Systems ( 2020 ). Ardi Tampuu, Tambet Matiisen, Maksym Semikin, Dmytro Fishman, and Naveed Muhammad. 2020. A survey of end-to-end driving: Architectures and training methods. IEEE Transactions on Neural Networks and Learning Systems (2020)."},{"key":"e_1_2_1_38_1","volume-title":"Ozlem Aslan, Shengjie Wang, Rich Caruana, Abdelrahman Mohamed, Matthai Philipose, and Matt Richardson.","author":"Urban Gregor","year":"2017","unstructured":"Gregor Urban , Krzysztof J. Geras , Samira Ebrahimi Kahou , Ozlem Aslan, Shengjie Wang, Rich Caruana, Abdelrahman Mohamed, Matthai Philipose, and Matt Richardson. 2017 . Do Deep Convolutional Nets Really Need to be Deep and Convolutional ? (2017). arxiv:stat.ML\/1603.05691 Gregor Urban, Krzysztof J. Geras, Samira Ebrahimi Kahou, Ozlem Aslan, Shengjie Wang, Rich Caruana, Abdelrahman Mohamed, Matthai Philipose, and Matt Richardson. 2017. Do Deep Convolutional Nets Really Need to be Deep and Convolutional? (2017). arxiv:stat.ML\/1603.05691"},{"key":"e_1_2_1_39_1","doi-asserted-by":"publisher","DOI":"10.1145\/3203408"},{"key":"e_1_2_1_40_1","doi-asserted-by":"publisher","DOI":"10.1145\/3300061.3345448"},{"key":"e_1_2_1_41_1","doi-asserted-by":"publisher","DOI":"10.1109\/ICC.2017.7997360"},{"key":"e_1_2_1_42_1","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2016.2597169"}],"container-title":["ACM Transactions on Embedded Computing Systems"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3477006","content-type":"unspecified","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3477006","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,6,17]],"date-time":"2025-06-17T19:30:46Z","timestamp":1750188646000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3477006"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2021,9,17]]},"references-count":42,"journal-issue":{"issue":"5s","published-print":{"date-parts":[[2021,10,31]]}},"alternative-id":["10.1145\/3477006"],"URL":"https:\/\/doi.org\/10.1145\/3477006","relation":{},"ISSN":["1539-9087","1558-3465"],"issn-type":[{"value":"1539-9087","type":"print"},{"value":"1558-3465","type":"electronic"}],"subject":[],"published":{"date-parts":[[2021,9,17]]},"assertion":[{"value":"2021-04-01","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2021-07-01","order":1,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2021-09-17","order":2,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}