Abstract
Sensing systems powered by energy harvesting have traditionally been designed to tolerate long periods without energy. As the Internet of Things (IoT) evolves toward a more transient and opportunistic execution paradigm, reducing energy storage costs will be key for its economic and ecologic viability. However, decreasing energy storage in harvesting systems introduces reliability issues. Transducers only produce intermittent energy at low voltage and current levels, making guaranteed task completion a challenge. Existing ad hoc methods overcome this by buffering enough energy either for single tasks, incurring large data-retention overheads, or for one full application cycle, requiring a large energy buffer. We present Julienning: an automated method for optimizing the total energy cost of batteryless applications. Using a custom specification model, developers can describe transient applications as a set of atomically executed kernels with explicit data dependencies. Our optimization flow can partition data- and energy-intensive applications into multiple execution cycles with bounded energy consumption. By leveraging interkernel data dependencies, these energy-bounded execution cycles minimize the number of system activations and nonvolatile data transfers, and thus the total energy overhead. We validate our methodology with two batteryless cameras running energy-intensive machine learning applications. Using a solar testbed, we replay real-world illuminance traces to experimentally demonstrate optimized batteryless execution with a transducer-to-application energy efficiency of 74.5%. Partitioning results demonstrate that compared to ad hoc solutions, our method can reduce the required energy storage by over 94% while only incurring a 0.12% energy overhead.
- [1] . 2013. Exploring convolutional neural network structures and optimization techniques for speech recognition. In Interspeech, Vol. 11. Citeseer, 73–75.Google Scholar
- [2] . 2020. Batteryless zero-maintenance embedded sensing at the Mithræum of circus maximus. In Proc. SenSys Conf.ACM, 368–381.Google Scholar
- [3] . 2020. On securing persistent state in intermittent computing. In Proc. ENSSys Workshop.Google Scholar
- [4] . 2015. Hibernus: Sustaining computation during intermittent supply for energy-harvesting systems. Embed. Syst. Lett. IEEE 7, 1 (2015). Google Scholar
Digital Library
- [5] . 2016. Efficient state retention for transiently-powered embedded sensing. In Int. Conf. Embed. Wireless Sys. Netw.137–148.Google Scholar
- [6] . 2016. Energy harvesting and wireless transfer in sensor network applications: Concepts and experiences. ACM Trans. Sensor Netw. (TOSN) 12, 3 (2016), 24.Google Scholar
Digital Library
- [7] . 2017. HarvOS: Efficient code instrumentation for transiently-powered embedded sensing. In Proc. IPSN Conf.ACM, New York, NY. Google Scholar
Digital Library
- [8] . 2014. Towards enabling uninterrupted long-term operation of solar energy harvesting embedded systems. In Proc. EWSN Conf. Springer, 66–83.Google Scholar
- [9] . 2011. Dewdrop: An energy-aware runtime for computational RFID. In Proc. USENIX NSDI. 197–210.Google Scholar
- [10] . 2016. Chain: Tasks and channels for reliable intermittent programs. In Proc. OOPSLA Conf.ACM. Google Scholar
Digital Library
- [11] . 2018. Termination checking and task decomposition for task-based intermittent programs. In Proc. Comp. Construct. Conf. ACM. Google Scholar
Digital Library
- [12] . 2018. A reconfigurable energy storage architecture for energy-harvesting devices. In Proc. ASPLOS Conf.ACM. Google Scholar
Digital Library
- [13] . 2020. Comparing NVM technologies through the lens of intermittent computation. In Proc. ENSSys Workshop. 77–78.Google Scholar
- [14] . 1959. A note on two problems in connexion with graphs. Numer. Math. 1, 1 (
Dec. 1959), 269–271. Google ScholarDigital Library
- [15] FLIR Systems, Inc. 2018. Lepton Engineering Datasheet. FLIR Systems, Inc.
Rev. 200. Google Scholar - [16] . 2018. MRAM as embedded non-volatile memory solution for 22FFL FinFET technology. In 2018 IEEE International Electron Devices Meeting (IEDM). IEEE, 18–1.Google Scholar
- [17] . 2020. On-demand communication with the batteryless mirocard: Demo abstract. In Proc. 18th Conf. Embed. Netw. Sens. Syst. 629–630.Google Scholar
- [18] . 2015. Reducing energy consumption in microcontroller-based platforms with low design margin co-processors. In Proc. DATE Conf.Google Scholar
- [19] . 2018. Thermal image-based CNN’s for ultra-low power people recognition. In Proc. Comput. Frontiers Conf. ACM.Google Scholar
- [20] . 2016. Dynamic energy burst scaling for transiently powered systems. In Proc. DATE Conf. EDA Consortium, 349–354.Google Scholar
- [21] . 2017. Wearable, energy-opportunistic vision sensing for walking speed estimation. In Proc. SAS Symp. IEEE, 1–6.Google Scholar
- [22] . 2018. Efficient, long-term logging of rich data sensors using transient sensor nodes. ACM Trans. Embed. Comput. Syst. 17, 1 (2018). Google Scholar
Digital Library
- [23] . 2017. A scan-chain based state retention methodology for IoT processors operating on intermittent energy. In Proc. DATE Conf. EDA Consortium.Google Scholar
- [24] . 2015. Tragedy of the Coulombs: Federating energy storage for tiny, intermittently-powered sensors. In Proc. SenSys Conf. ACM.Google Scholar
- [25] . 2017. Timely execution on intermittently powered batteryless sensors. In Proc. SenSys Conf.Google Scholar
- [26] . 2013. Write activity reduction on non-volatile main memories for embedded chip multiprocessors. ACM Trans. Embed. Comput. Syst. 12, 3 (
April 2013), Article77 , 27 pages. Google ScholarDigital Library
- [27] . 2019. Capacity over capacitance for reliable energy harvesting sensors. In Proc. 18th Int. Conf. Inf. Process. Sensor Netw. 193–204.Google Scholar
- [28] . 2014. QUICKRECALL: A low overhead HW/SW approach for enabling computations across power cycles in transiently powered computers. Proc. Int. Conf. VLSI Design (2014). Google Scholar
Digital Library
- [29] . 2014. An FRAM-based nonvolatile logic MCU SoC exhibiting 100% digital state retention at VDD=0 V achieving zero leakage with \(\lt\)400-ns wakup time for ULP applications. IEEE J. Solid-State Circuits 49, 1 (Jan. 2014). Google Scholar
Cross Ref
- [30] . 2015. Recurrent convolutional neural network for object recognition. In Proc. IEEE Conf. Comput. Vis, Pattern Recogn. 3367–3375.Google Scholar
- [31] . 2016. A 65nm ReRAM-enabled nonvolatile processor with 6\(\times\) reduction in restore time and 4\(\times\) higher clock frequency using adaptive data retention and self-write-termination nonvolatile logic. In Proc. ISSCC Conf.IEEE. Google Scholar
Cross Ref
- [32] . 2020. Dynamic task-based intermittent execution for energy-harvesting devices. ACM Trans. Sens. Netw. (TOSN) 16, 1 (2020), 1–24.Google Scholar
Digital Library
- [33] . 2016. Powering Long Range Wireless Nodes with Harvested Energy.
Technical Report , November. Zürcher Hochschule für Angewandte Wissenschaften (ZHAW).Google Scholar - [34] . 2019. Event-triggered natural hazard monitoring with convolutional neural networks on the edge. In Proc. IPSN Conf. IEEE, 73–84.Google Scholar
- [35] . 2010. Adaptive power management for environmentally powered system. IEEE Trans. Comput. 59, 4 (2010), 478–491.Google Scholar
Digital Library
- [36] . 2015. WISPCam: A battery-free RFID camera. In Proc. IEEE RFID.Google Scholar
- [37] OmniVision 2006. OV7670/OV7171 CMSO VGA CameraChip Sensor Preliminary Datasheet. OmniVision.
Version 1.4. Google Scholar - [38] . 2015. An introduction to convolutional neural networks. arXiv:1511.08458.Google Scholar
- [39] . 2014. A 3.4-pJ FeRAM-enabled D flip-flop in 0.13-\(\mu\)m CMOS for nonvolatile processing in digital systems. IEEE J. Solid-State Circuits 49, 1 (
Jan. 2014), 202–211. Google ScholarCross Ref
- [40] . 2005. Design considerations for solar energy harvesting wireless embedded systems. In Proc. IPSN Conf. IEEE Press, 64.Google Scholar
- [41] . 2011. Mementos: System support for long-running computation on RFID-scale devices. SIGARCH Comput. Archit. News 46, 3 (
March 2011), 12. Google ScholarDigital Library
- [42] . 2018. RESTOP: Retaining external peripheral state in intermittently-powered sensor systems. Sensors 18, 1 (2018).Google Scholar
- [43] . 1988. Global value numbers and redundant computations. In Proc. 15th ACM SIGPLAN-SIGACT Symp. Principles Program. Lang. ACM, 12–27.Google Scholar
- [44] . 2020. Harvesting-aware optimal communication scheme for infrastructure-less sensing. ACM Trans. Internet Things 1, 4 (2020), 1–26.Google Scholar
Digital Library
- [45] . 2021. Environment and application testbed for low-power energy harvesting system design. IEEE Trans. Indust. Electron. 68, 11 (2021), 11146–11156. Google Scholar
Cross Ref
- [46] . 2017. Measurement and validation of energy harvesting IoT devices. In Proc. DATE Conf. IEEE, 1159–1164.Google Scholar
- [47] . 2019. Dataset: Tracing indoor solar harvesting. In Proc. 2nd Worksh. Data Acquis. Anal. (
DATA’19 ). ACM. Google ScholarDigital Library
- [48] . 2015. Handbook of Energy Harvesting Power Supplies and Applications. CRC Press.Google Scholar
Cross Ref
- [49] . 2020. Analysing and improving robustness of predictive energy harvesting systems. In Proc. ENSsys Worksh. Association for Computing Machinery, New York, NY.Google Scholar
- [50] . 2018. On Efficient Data Exchange in Multicore Architectures. Ph.D. Dissertation. ETH Zurich.Google Scholar
- [51] . 2018. Selective policies for efficient state retention in transiently-powered embedded systems: Exploiting properties of NVM technologies. Sustain. Comput. Inform. Syst. 22 (2018), 167–178. Google Scholar
Cross Ref
- [52] . 2013. A survey of multi-source energy harvesting systems. In Proc. DATE Conf.Google Scholar
Cross Ref
- [53] . 2008. Sliding-windows for rapid object class localization: A parallel technique. In Joint Pattern Recognition Symposium. Springer, 71–81.Google Scholar
- [54] . 2010. Phase change memory. Proc. IEEE 98, 12 (2010), 2201–2227.Google Scholar
Cross Ref
Index Terms
Dataflow Driven Partitioning of Machine Learning Applications for Optimal Energy Use in Batteryless Systems
Recommendations
Energy-aware system design for batteryless LPWAN devices in IoT applications
AbstractInterconnected LPWAN devices, which create the IoT, are usually powered by batteries that significantly limit their operational lifetime. The main disadvantage of batteries is that they must be periodically/manually replaced with new ...
Applications of Energy-Driven Computing: A Transiently-Powered Wireless Cycle Computer
ENSsys'17: Proceedings of the Fifth ACM International Workshop on Energy Harvesting and Energy-Neutral Sensing SystemsThere has been a dramatic increase in recent years in the number of battery-powered embedded electronic devices. However, the lifetime of these devices is limited by battery capacity. Energy harvesting is an efficient solution to overcome this ...
Integrated solar energy harvesting and storage
To explore integrated solar energy harvesting as a power source for low power systems, an array of energy scavenging photodiodes based on a passive-pixel architecture for CMOS imagers has been fabricated together with storage capacitors implemented ...






Comments