Hibernus++: A Self-Calibrating and Adaptive System for Transiently-Powered Embedded Devices

TLDR: This paper proposes Hibernus++ to intelligently adapt the hibernate and restore thresholds in response to source dynamics and system load properties, which provides an average 16% reduction in energy consumption and an improvement of 17% in application execution time over state-of-the-art approaches.

Abstract: Energy harvesters are being used to power autonomous systems, but their output power is variable and intermittent. To sustain computation, these systems integrate batteries or supercapacitors to smooth out rapid changes in harvester output. Energy storage devices require time for charging and increase the size, mass, and cost of systems. The field of transient computing moves away from this approach, by powering the system directly from the harvester output. To prevent an application from having to restart computation after a power outage, approaches such as Hibernus allow these systems to hibernate when supply failure is imminent. When the supply reaches the operating threshold, the last saved state is restored and the operation is continued from the point it was interrupted. This paper proposes Hibernus++ to intelligently adapt the hibernate and restore thresholds in response to source dynamics and system load properties. Specifically, capabilities are built into the system to autonomously characterize the hardware platform and its performance during hibernation in order to set the hibernation threshold at a point which minimizes wasted energy and maximizes computation time. Similarly, the system auto-calibrates the restore threshold depending on the balance of energy supply and consumption in order to maximize computation time. Hibernus++ is validated both theoretically and experimentally on microcontroller hardware using both synthesized and real energy harvesters. Results show that Hibernus++ provides an average 16% reduction in energy consumption and an improvement of 17% in application execution time over state-of-the-art approaches.

Figures

Fig. 5. Self-calibration of the hibernation voltage threshold, VH

Fig. 6. Hibernation self-calibration circuit

Fig. 7. Calculated voltage drop (Vcal - Vmeas) with different Vcal

Fig. 19. Energy overhead comparison between Hibernus++ and QuickRecall

TABLE IV EXPERIMENTALLY MEASURED PARAMETERS

Fig. 18. Three synthetic traces of real harvesters used in Section V-B

Frequently asked questions

Q1. What contributions have the authors mentioned in the paper "Hibernus++: a self-calibrating and adaptive system for transiently-powered embedded devices" ?

This work proposes Hibernus++ to intelligently adapt the hibernate and restore thresholds in response to source dynamics and system load properties. 

Q2. What is the number of snapshots with Mementos?

The number of snapshots with Mementos depends on the checkpoint placement, the value of Vmin and the supply interruption frequency; while for Hibernus++ it only depends on the supply interruption frequency. 

Q3. What is the way to determine the optimal time for a snapshot?

if only one nonvolatile memory block is used for snapshot, a power loss during a snapshots is likely to result in the loss of all data up to that point. 

Q4. What is the value of the voltage at the end of the hibernation process?

The drop in supply voltage due to hibernation (the process of storing the snapshot) is given by Vcal − Vmeas, where Vmeas is the voltage measured at the end of the hibernation process. 

Q5. What is the new paradigm of c mputing?

A new paradigm, which addresses computing challenges with transient power sources such as energy harvesting, is of ‘transiently- owered c mputing’ [14]. 

Q6. What is the drawback of ch ckpoi ting?

a drawback of ch ckpoi ting is that it is impossible to predict the xact time of failures, so computation time will be wasted by (1) taking unnecessary checkpoints, and (2) rolling back to the last checkpoint if power failure occurs towards the end of a checkpoint interval. 

Q7. Why did the power analyser capture the traces?

Due to limitations of the power analyser (which captures power traces and allows them to be replayed as a synthesized source), the authors could only collect 20 s of indoor PV behaviour during which lights are turned on and off twice. 

Q8. What are the operating parameters of the system?

The operating parameters of the system are shown: hibernate and restore operations, the calibrate and classify operations, and the time for the FFT execution and the system in ON mode. 

Q9. What is the performance of Mementos with higher Vm?

Mementos is more stable with higher Vm, but the performance decreases due to the large number of snapshots, as shown in Table I. 

Q10. What is the concept of checkpointing used in large-scale computing?

This typically borrows from the concept of checkpointing, which has been used in large-scale computing for decad s to provide robustness against errors or hardware failure [15].