Using Compressive Sampling to Fill Interbatch Data Gap From Low-Cost IoT Vibration Sensor

A low-cost wireless vibration sensor can be built using a 3-axis accelerometer, such as ADXL345, attached to a low-cost Wi-Fi microchip, such as ESP8266. In an Internet of Things (IoT) setting, a large number of such inexpensive sensor nodes can be setup with the widely used direct-read-and-send method which samples and sends individually acquired vibration data points from the sensor through the Internet to a server. In this work, we show that such a method is not effective. As the microcontroller alternates between sampling and sending the data, the micro delays of transmission will affect the sensor sampling rate and cause the data points to space unevenly, making the acquired data inaccurate. We propose that vibration should be sampled and transmitted in batches, as such data are acquired continuously without interruption and data points are more evenly spaced. However, the proposed batch-read-and-send will have interbatch gaps that need to be filled. Thus, the key contribution of this work is the novel use of compressive sampling (CS) technique to bridge those gaps. Experimental results show that the direct-read-and-send method loses more information and can only achieve a maximum sampling rate of 350 Hz with a standard uncertainty of 12.4, whereas the proposed solution can measure the vibration wirelessly and continuously up to 633 Hz. Gaps with up to 160 missing points can be filled using CS and achieve better accuracy, with a mean absolute error (MAE) of up to 0.048 and a standard uncertainty of 0.001, making the low-cost wireless vibration sensor a cost-effective solution in an IoT setting.

Automated summary

This paper proposes the construction of a low-cost wireless vibration sensor using a 3-axis accelerometer and Wi-Fi microchip, and solves the inaccuracy issue of the direct-read-and-send method through compressive sampling technique, enabling the sensor to continuously and efficiently measure vibration data in an IoT environment.