This paper analyzes the main challenges associated with noninvasive, continuous, wearable, and long-term breathing monitoring. The characteristics of an acoustic breathing signal from a miniature sensor are studied in the presence of sources of noise and interference artifacts that affect the signal. Based on these results, an algorithm has been devised to detect breathing. It is possible to implement the algorithm on a single integrated circuit, making it suitable for a miniature sensor device. The algorithm is tested in the presence of noise sources on five subjects and shows an average success rate of 91.3% (combined true positives and true negatives).

This paper presents a nanopower programmable bandpass filter suitable to process biomedical signals. The filter proves to be very robust to mismatch and process variations even when it has been implemented using MOS transistors biased in the weak inversion region. The paper analyses design issues associated to matching and process variations for the chosen filter topology and constituent transconductor block. The design equations justify the choice of both when the main constraints are robustness and power. The sixth order, bandpass filter prototype consumes 70 nW of power, with a dynamic range greater than 47 dB and operates at 1-V power supply. The filter was designed as part of a wearable breathing detector but its wide programmability range makes it suitable for many other biomedical sensor interfaces that require steep low frequency rejection band as well as ultralow power and low voltage operation.

Abstract-This brief presents a first order low pass filter topology capable of providing cut-off frequencies down to 2 mHz with a power consumption of 5 nW. The circuit is intended for signal conditioning applications, particularly for use with very low frequency physiological signals in low power portable medical equipment. To achieve the low frequency cut-off the filter is based around the use of a clocked transconductor which provides a low transconductance while using a relatively high bias current level. The circuit is implemented in a 0.35 µm technology with a 1 V supply and has a measured 32 µV RMS of noise and 64 dB dynamic range. In terms of power consumption and cut-off frequency the reported filter outperforms previous filters from the literature.

No abstract

No abstract