A sub-nJ CMOS ECG classifier for wireless smart sensor

Area of Science:

• Biomedical Engineering
• Wearable Technology
• Signal Processing

Background:

• Body area sensor networks offer efficient remote patient monitoring.
• Continuous electrocardiogram (ECG) signal transmission is energy-intensive for wireless sensors.
• Reducing sensor energy consumption is critical for prolonged monitoring.

Purpose of the Study:

• To develop an energy-efficient analog classifier for ECG signals.
• To embed this classifier into a wireless sensor for real-time analysis.
• To reduce the power demands of physiological monitoring systems.

Main Methods:

• Implementation of a sparse neural associative memory classifier in 65 nm CMOS technology.
• Analog circuit design for low-power ECG signal classification.
• Performance evaluation of the classifier in terms of accuracy and energy consumption.

Main Results:

• Achieved 93.6% classification accuracy for ECG signals.
• Required only 234 pJ per classification, a six-order-of-magnitude reduction compared to digital accelerators.
• Resulting sensor lifespan is 191 times longer than sensors transmitting raw data.

Conclusions:

• Analog implantation of ECG classifiers offers substantial energy savings.
• This approach significantly extends the operational life of wireless medical sensors.
• Enables more sustainable and cost-effective continuous physiological monitoring.