Dinko Oletic

Low-Power Wearable Respiratory Sound Sensing

Building upon the findings from the field of automated recognition of respiratory sound patterns, we propose a wearable wireless sensor implementing on-board respiratory sound acquisition and classification, to enable continuous monitoring of symptoms, such as asthmatic wheezing. Low-power consumption of such a sensor is required in order to achieve long autonomy. Considering that the power consumption of its radio is kept minimal if transmitting only upon (rare) occurrences of wheezing, we focus on optimizing the power consumption of the digital signal processor (DSP). Based on a comprehensive review of asthmatic wheeze detection algorithms, we analyze the computational complexity of common features drawn from short-time Fourier transform (STFT) and decision tree classification. Four algorithms were implemented on a low-power TMS320C5505 DSP. Their classification accuracies were evaluated on a dataset of prerecorded respiratory sounds in two operating scenarios of different detection fidelities. The execution times of all algorithms were measured. The best classification accuracy of over 92%, while occupying only 2.6% of the DSPs processing time, is obtained for the algorithm featuring the time-frequency tracking of shapes of crests originating from wheezing, with spectral features modeled using energy.

Urban crowd sensing demonstrator: Sense the Zagreb Air

We demonstrate an urban crowd sensing application for monitoring air quality by use of specially-designed wearable sensors and mobile phones. The application is built upon the OpenIoT platform1 with the goal to support context-aware and energy-efficient acquisition and filtering of sensor data in mobile environments while ensuring adequate sensing coverage. We demonstrate how sensors and mobile devices jointly collect and share data of interest to measure air quality. In particular, we outline the main features of our wearable air quality sensors, present the data acquisition process as well as the user view of the system, which, in contrast to similar applications, provides a personalized real-time notification mechanism to mobile application users. The solution was used in an air quality measurement campaign “Sense the Zagreb Air” performed in the City of Zagreb, Croatia, in early July 2014 with 20 participants.

Design of sensor node for air quality crowdsensing

Information on air-quality in urban environments is typically measured only at limited number of sites, due to cost of measurement of atmospheric concentrations of toxic gases (CO, NO2, SO2) within accuracy boundaries defined by regulative bodies. Low spatial resolution of the mentioned environmental parameters hinders their applications in localization of the air-pollution sources, traffic regulation or studies of chronic respiratory diseases related to personal pollution exposure. Thus, we propose complementing the existing air quality monitoring infrastructure by a network of mobile sensors enabling the citizens to participate in measurement (e.g. “crowdsensing”). In this paper, we present the design of such battery-powered, wearable sensor node, housing two electrochemical gas sensors, temperature, relative humidity and atmospheric pressure sensors, with Bluetooth connectivity. Electrical, mechanical and software design are shown. Next, sensor node was characterized by evaluating the sensing accuracy and the autonomy in laboratory conditions. Accuracy within ±1 °C, ±2% RH, ±2 hPa, and ±0.6 ppm CO is shown. Autonomy is estimated at 65 h. Preliminary results of the outdoor functional test are demonstrated.

Empowering smartphone users with sensor node for air quality measurement

We present an architecture of a sensor node developed for use with smartphones for participatory sensing of air quality in urban environments. Our solution features inexpensive metal-oxide semiconductor gas sensors (MOX) for measurement of CO, O3, NO2 and VOC, along with sensors for ambient temperature and humidity. We focus on our design of sensor interface consisting of power-regulated heater temperature control, and the design of resistance sensing circuit. Accuracy of the sensor interface is characterized. Power consumption of the sensor node is analysed. Preliminary data obtained from the CO gas sensors in laboratory conditions and during the outdoor field-test is shown.

Towards Continuous Wheeze Detection Body Sensor Node as a Core of Asthma Monitoring System

This article presents a wheeze detection method for wearable body sensor nodes used in management of asthma. Firstly, a short review of current state of telemonitoring in management of chronic asthma is given. A concept of the asthma monitoring system built around a body sensor node analysing respiratory sounds is proposed, with a smart phone as a self-management center and additional sensor nodes for environment monitoring. In search for a wheeze detection algorithm suitable for low power continuous operation on wireless sensor node, a simple algorithm based on the 4-th order linear prediction coefficients (LPC) method is presented. Predictor error energy ratio of Durbin’s algorithm is used as the only feature. Algorithm is implemented on low power digital signal processor (DSP) to evaluate its performance. Sensitivity (SE) of 70.9%, specificity (SP) 98.6% and accuracy (ACC) of 90.29% are achieved using pre-recorded test signals. Program complexity is analysed in order to identify possibilities of lowering power consumption.

Prototype of Respiratory Sounds Monitoring System Based on Compressive Sampling

In this study we demonstrate end-to-end prototype of the respiratory sounds monitoring m-health system based on compressive sampling (CS). We show a low power implementation of the involved pseudorandom sampling on the wearable wireless acoustic sensor featuring a 16-bit microcontroller and Bluetooth radio chip. CS reconstruction using the orthogonal matching pursuit (OMP) reconstruction algorithm was implemented on the Android smartphone.We measure network datarate reduction in comparison with associated reconstruction accuracy of compressible DCT spectrum on synthetically generated harmonic signals and prerecorded respiratory sounds containing asthmatic wheezing.

Energy-efficient atmospheric CO concentration sensing with on-demand operating MOX gas sensor

Collaborative mobile air-pollution monitoring, employing context aware sampling scheduling, requires on-demand sensing of gas concentration. If MOX gas sensors are used, their energy consumption can be reduced by on-demand heating, with a risk to compromise the sensing repeatability. We experimentally investigate response of MiCS-5525 CO MOX sensor to intermittent heating sequences consisting of a train of short rectangular pulses. We analyse stability of response, energy consumption and sensitivity at low concentrations, for various combinations of pulse waveform period and duty-cycle. We obtained stable readings for the heating sequences consuming 200-300 mJ, saving over 30% of energy per reading.

Extending lifetime of battery operated wireless sensor node with DC-DC switching converter

In this paper we study lifetime extension of a wireless sensor node operated from primary battery through a DC-DC switching converter. The node was developed in course of the environmental monitoring project MasliNET. A step-up switching converter is needed to drain the battery when its voltage drops below the operating voltage of the node. In order to minimize the losses caused by the DC-DC converter it is important to shut it down whenever possible and keep the difference between its input and output voltage low. We have developed a controllable power-supply unit and a dynamic control algorithm that manages the use and the output voltage of the DC-DC converter based on the current nodes task requirements and the battery state. We have experimentally verified the solution by comparing it to the node permanently operated from a fixed output voltage DC-DC converter. It has been shown that the proposed power-supply design and the control algorithm prolong battery lifetime as much as 30 %.

Monitoring Respiratory Sounds: Compressed Sensing Reconstruction via OMP on Android Smartphone

We present a novel respiratory sounds monitoring concept based on compressive sensing (CS). Respiratory sounds are streamed from a body-worn sensor node to a smartphone where processing is conducted. CS is used to simultaneously lower sampling frequency on the sensor node and over-the-air data rate. In this study we emphasize compressed sensing reconstruction via orthogonal matching pursuit (OMP) on Android smartphone. Accuracy of the reconstruction and execution speed are investigated using synthetic signals. We demonstrate applicability of the technique in real-time reconstruction of at least 10 components of compressible DCT spectrum of respiratory sounds containing asthmatic wheezing, acquired at 4x lower sampling rate.

Energy-Efficient Respiratory Sounds Sensing for Personal Mobile Asthma Monitoring

Current medical practice of long-term chronic respiratory diseases treatment lacks a convenient method of empowering the patients and caregivers to continuously quantitatively track the intensity of respiratory symptoms. Such is asthmatic wheezing, occurring in respiratory sounds. We envision a mobile, personalized asthma monitoring system comprising of a wearable, energy-constrained acoustic sensor and smartphone. In this paper, we address the energy-burden of acquisition and streaming of acoustic respiratory signal, and lessen it by applying the concept of compressed sensing (CS). First, we analyse the adherence of normal and pathologic respiratory sounds frequency representations (discrete Fourier transform and discrete cosine transform) to the sparse signal model. Given the pseudo-random non-uniform subsampling encoder implemented on MSP430 microcontroller, we review tradeoffs of accuracy and execution time of different CS algorithms, suitable for real-time respiratory spectrum recovery on smartphone. Working CS respiratory spectrum acquisition prototype is demonstrated, and evaluated. Prototype enables for real-time reconstruction of spectra dominated by approximately eight frequency components with more than 80% accuracy, on Android smartphone using Orthogonal Matching Pursuit algorithm, from only 25% signal samples (with respect to Nyquist rate) acquired and streamed by sensor at 8 kb/s.