Self-Powered Wearable IoT Devices for Health and Activity Monitoring

Bhat, Ganapati; Gupta, Ujjwal; Tuncel, Yiğit; Karabacak, Fatih; Ozev, Sule; Ogras, Ümit Y.

doi:10.1561/1000000056

Cited by 11 publications

(8 citation statements)

References 1 publication

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To find the rest of the level hypervectors 𝐿 2 1 to 𝐿 10 1 and 𝐿 2 2 to 𝐿 10 2 , we first calculate the number of bits to flip at each consecutive level as 𝑏 = 𝐷/2(𝑀 − 1) = 55. Then, 𝐿 2 1 is found by 𝐿 2 1 = 𝑔(𝐿 1 1 , 55) and the rest of 𝐿 𝑚 1 and 𝐿 𝑚 2 are calculated through the same procedure. As a result, we obtain 20 different level hypervectors that represent the quantized 𝑓 1 and 𝑓 2 .…”

Section: Background On Hdc 31 Hdc Trainingmentioning

confidence: 99%

“…For example, assistive devices for Parkinson's Disease patients need to provide precisely timed audio cues to cope with gait disturbances [10,12]. Despite the intensity of sensor data, these devices must operate at a tight energy budget (∼ 𝜇W) due to limited battery capacity [2,9]. Offloading the data to the cloud is not an attractive solution since it increases the communication energy [35] and raises privacy concerns [25].…”

mentioning

confidence: 99%

“…Likewise, deep neural networks and other sophisticated algorithms are not viable due to the limited computation and energy capacity of wearable edge devices. Therefore, there is a strong need for computationally light learning algorithms that can provide high accuracy, real-time inference, and robustness to noisy sensor data [2].…”

mentioning

confidence: 99%

See 2 more Smart Citations

Hypervector Design for Efficient Hyperdimensional Computing on Edge Devices

Basaklar,

Tuncel,

Narayana

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Hyperdimensional computing (HDC) has emerged as a new lightweight learning algorithm with smaller computation and energy requirements compared to conventional techniques. In HDC, data points are represented by high-dimensional vectors (hypervectors), which are mapped to high-dimensional space (hyperspace). Typically, a large hypervector dimension (≥ 1000) is required to achieve accuracies comparable to conventional alternatives. However, unnecessarily large hypervectors increase hardware and energy costs, which can undermine their benefits. This paper presents a technique to minimize the hypervector dimension while maintaining the accuracy and improving the robustness of the classifier. To this end, we formulate the hypervector design as a multi-objective optimization problem for the first time in the literature. The proposed approach decreases the hypervector dimension by more than 32× while maintaining or increasing the accuracy achieved by conventional HDC. Experiments on a commercial hardware platform show that the proposed approach achieves more than one order of magnitude reduction in model size, inference time, and energy consumption. We also demonstrate the trade-off between accuracy and robustness to noise and provide Pareto front solutions as a design parameter in our hypervector design. CCS CONCEPTS• Computing methodologies → Machine learning; • Hardware → Power and energy.

show abstract

Section: Background On Hdc 31 Hdc Trainingmentioning

confidence: 99%

mentioning

confidence: 99%

See 1 more Smart Citation

Hypervector Design for Efficient Hyperdimensional Computing on Edge Devices

Basaklar,

Tuncel,

Narayana

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…This growth has been enabled by advances in sensor technologies, low-power processing, and communication technologies. IoT devices have applications in a wide range of fields such as wearable health monitoring, environmental monitoring, digital agriculture, and robotics [9]- [13]. IoT devices are deployed either as a network of multiple devices or single nodes, depending on the application requirements.…”

Section: Related Workmentioning

confidence: 99%

“…Studies have shown that ambient light has the highest potential for EH with power levels of >1 mW and <100 µW in outdoor and indoor conditions, respectively. Similarly, piezoelectric human motion EH can yield about 15 µW [16], while body heat and RF EH have lower power levels of <5 µW, respectively [13].…”

Section: Related Workmentioning

confidence: 99%

ECO: Enabling Energy-Neutral IoT Devices through Runtime Allocation of Harvested Energy

Tuncel¹,

Bhat²,

Park³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Energy harvesting offers an attractive and promising mechanism to power low-energy devices. However, it alone is insufficient to enable an energy-neutral operation, which can eliminate tedious battery charging and replacement requirements. Achieving an energy-neutral operation is challenging since the uncertainties in harvested energy undermine the quality of service requirements. To address this challenge, we present a rollout-based runtime energy-allocation framework that optimizes the utility of the target device under energy constraints. The proposed framework uses an efficient iterative algorithm to compute initial energy allocations at the beginning of a day. The initial allocations are then corrected at every interval to compensate for the deviations from the expected energy harvesting pattern. We evaluate this framework using solar and motion energy harvesting modalities and American Time Use Survey data from 4772 different users. Compared to state-of-theart techniques, the proposed framework achieves 34.6% higher utility even under energy-limited scenarios. Moreover, measurements on a wearable device prototype show that the proposed framework has less than 0.1% energy overhead compared to iterative approaches with a negligible loss in utility.

show abstract