Measure the node spacing of wireless sensors using audio signals

Measure the node spacing of wireless sensors using audio signals

[Date:]

Source: China Power Grid Author: Meng lingjun, Xia shanhong

[Font:Large Medium Small]

　

1 Introduction

A wireless sensor network node system that uses audio signals to achieve autonomous measurement of node spacing. The system includes dsPIC6014A microcontroller, 512 kb sram, 2.4G band RF transceiver module, audio transceiver module and power management module. The time difference between the RF synchronous signal and the audio signal is measured to measure the distance between nodes. The Node increases the signal-to-noise ratio of the Ranging Signal by accumulating the average of the measured data multiple times and using the IIR digital filter technology, range detection is used to determine the arrival time of ranging signals. Test data shows that the distance to the node's remote location is up to 30 m, and the error is less than 3.5%.

The main parameters used for measuring the distance between nodes are:

Received Signal strength (RSS), Signal Time Difference (TPOA), angle (AOA)/signal arrival direction (DOA ). Among them, there are many studies on RSS and RF plus ultrasonic ranging. The propagation attenuation of RF signals is related to many parameters, such as the initial transmit power, the height of the antenna from the ground, the reflection, and the carrier frequency. If no verification is performed, the error may exceed 50%. The ultrasonic frequency of RF and ultrasonic positioning is 40 kHz. The attenuation in the stored air determines that the distance between the two locations is generally less than 10 m, and the direction is strong. This method is suitable for indoor use. The attenuation of acoustic waves in air towel decreases with the decrease of frequency. In the case of several kHz, the distance measurement within dozens of meters can be realized by using the commercial audio Transceiver Technology as low as small, it is an effective technology for achieving long-distance and high-precision positioning. The article [] describes how to use pseudo-random code + DSP-related processing to achieve centimeter-level sound source localization accuracy, and the system structure is complex. The work in the document [5] is similar to the work in this paper. Using the General Mica2 node platform, the high-power sound transmitter and analog phase-locked loop are used to detect the audio Ranging Signal.

Mature and commercialized nodes are Mica series and Telos nodes developed by UC Berkeley. These nodes only provide one basic hardware platform and must use a Dedicated Interface Board to expand other functions. The purpose of this thesis is to explore a node hardware system that can be used in the field and has long-distance and high-precision self-positioning. Design a new node structure.

2-node hardware system design

Based on the above considerations, node Structure 1 is implemented. The node uses the dsPIC6014A microcontroller of Microchip. It has 12-bit ADC and 8 kb ram, 16-bit command operation and I/O control. It supports C Programming and some DSP functions, flexible clock and power consumption control, available in 3 ~ The 5 V voltage range works at a maximum operating speed of 20 MIPS at 3.3V. The node is configured with a kb SRAM. A 16-bit port of dsPIC6014A is used as the low position of the SRAM address bus. The 3-bit high is controlled by the other 3 I/O bits, the eight sub-buckets are used to save the collected audio signal data and temporary data for digital signal processing.

The RF transceiver module uses nRF24L01 to exchange data with the CPU through the SPI interface. The audio signal generator uses commercially available Standard piezoelectric beeps. After testing and analyzing the noise spectrum in nature, it is found that most of the audio signal frequencies are concentrated in 20 ~ Therefore, the central frequency of the buzzer is 3000Hz and the sound pressure is greater than 90 dB. The audio receiving sensor is a polar microphone with a gain of about 60 dB for two-level amplifiers. To improve the anti-interference capability, a second-order butworth band-pass filter with a center frequency of 3000Hz is added to the node, the circuit structure 2 is shown. Circuit test results: the center frequency is 3000Hz, and the-3 dB bandwidth is about 987Hz.

The node uses a 3.7 V rechargeable lithium-ion battery as the power supply. During the hibernation period, the power supply of all non-working units is turned off for energy saving. A power management unit controlled by RTC (Real-time clock) allocates and manages all levels of power supply. After the system is powered on, the Power Supply Management Unit is set to valid, and the CPU sets the RTC wake-up time. After the work is completed, the CPU closes the power supply management unit output, at this time, only the RTC and power management units are working, and the power consumption is 12 μW. When the scheduled wake-up time is approaching, RTC outputs an interrupt signal, enables the power, and the node enters the working state, in this way, node operation and sleep cycle are controlled. The unique feature of a node is that it uses the timer interrupt setting function provided by RTC from several minutes to several days to realize flexible control of node operation and sleep cycle, and achieves a low power consumption design.

When the outdoors uses audio signal ranging, atmospheric temperature, wind speed, and wind direction have a certain impact on the Sound speed. There are still large technical barriers to achieve Wind Speed Measurement on nodes, the temperature is mainly affected when the wind speed is low. Here, the formula c = 331.4 + 0.6T is used to compensate the Sound speed. In the formula, T is the atmospheric temperature (℃ ). The temperature sensor is the DS1624 of Maxim and has a standard I2C interface.

3 ranging signal arrival time Algorithm

In this paper, a TOA estimation method for ranging signals based on digital rectification is proposed. Its basic principle is to process ranging signals by digital signal processing, obtains the amplitude variation information of the Ranging Signal with a higher signal-to-noise ratio, and uses the amplitude variation trend analysis to estimate Toa. It includes the following processes:

(1) signal-to-noise ratio of the Ranging Signal. Ranging signals can be expressed as follows: F (t) = Av + assin (ω st + φs) + N (t). AV is the DC component generated after signal collection, N (t) is a random distribution of noise. According to the signal analysis theory, to improve the signal-to-noise ratio, you can use digital filtering or multi-sample accumulation to calculate the average value. Considering the node's computing capability and hardware structure, the method of calculating the average value after 4 sampling times is adopted.

(2) Remove the DC component. Calculate the average value of f (t) Av, then perform subtraction, filter out the DC component in the signal to make it an AC signal J (T), and the audio Ranging Signal is an AC signal, filtering out DC components helps to separate large Ranging Signal ranges during subsequent processing.

(3) digital full-wave rectification. After (2) processing, the signal is a positive and negative converted bipolar signal, and then Z (t) = | j (t) | processing, that is, digital full-wave rectification, returns a single polarity signal.

(4) low-pass filtering. The second-order IIR low-pass filter is used to filter the Z (t) and obtain a signal B (t) similar to the Z (t) envelope ).

(5) analyze the amplitude change trend of B (t. In the data segment where the Ranging Signal starts to appear, the amplitude difference between adjacent data points is large and is continuously increasing (the experiment can determine the minimum number of continuously increasing data ), find the starting point N (I) for the continuous increase of the amplitude, that is, the signal arrival time point, as shown in 3.

4 Test Results

The ranging signal used in the test is a single-frequency sine signal with a frequency of 3000Hz, a sampling frequency of 23.8 kHz, and a sampling length of 4096 points (12-bit ADC ). Figure 3 shows the source signal waveform and the data waveform during data processing. For the original signal, there is a large error or failure to identify the starting point of the Ranging Signal by using the amplitude or frequency of the signal. The method described in this article can obtain the signal starting point with higher accuracy. When the system clock is 10 MHz, the entire computing process takes about 1.5 s, which can meet the positioning requirements of static or slow moving nodes. The maximum ranging error at 30 m is about 3.5%.

Conclusion 5

A wireless sensor network node with the audio positioning function is implemented. It has an independent RTC + power management unit design, implements Low Power sleep, and can measure the node spacing 30 m away. The TOA estimation method for ranging signals implemented by single-chip microcomputer is proposed to obtain high accuracy of arrival time estimation, which provides an effective method for realizing high precision node location. This node can be used to build a wireless network that is used for long distance node spacing such as forests and farmland.