**3. Results**

#### *3.1. Measurement Accuracy*

First, we do the measurement accuracy experiment. The experience was carried on the author's campus from 22 September to 21 October 2021. The campus is located at 120◦55 E and 28◦51 N. The weather was sunny during the test. The campus is spacious. The parameters of cc1101 are a carrier frequency of 433 MHz, a baud rate of 100 kbps, and a modulation mode of 2 FSK. The transmission power of the NB-IoT circuit is 13 dBm, the antenna gain is 3 dB, and the transmission rate is 3.9 Kbps. Each test point continuously sends and receives 1000 data packets. The packet loss rate of the whole network is less than 1% within 8 km, and the packet loss rate is 0% within 400 m. We read the humidity data and temperature data 10 times of 6 sensor nodes every day for a week and compare the data of thermometers and hygrometers which are put near the sensor node simultaneously. Then, the error was calculated and drawn in Figure 11a,b. The temperature error is less than 1 ◦C, as shown by the red line on Figure 11a. The average of temperature error is about 0.5 ◦C, as shown by the black line on the Figure 11a. The humidity error is less than 5% RH, as shown by the red line on Figure 11b. The average of humidity error is about 2% RH, as shown by the black line on the Figure 11b. The error of temperature and humidity are mainly decided by the DHT11 chip precision.

**Figure 11.** The measurement error. (**a**) temperature error (**b**) humidity error.

Then, one sink node and its three sensor nodes were placed in a mobile car. We put a Leica GNSS (teaching edition) in the car, which is a professional position for the measuring instrument. The car was moved in different speed, and we read the latitude and longitude information of the monitoring center and Leica Receiver simultaneously. Let the longitude and latitude test by Leica GNSS are *LonA*, *LatA*. Let the longitude and latitude test by our system are *LonB*, *LatB*. Then the position error can be calculated by the Expression (14).

$$\begin{array}{c} \Delta Ion = (LonA - LanB) \times 1000 \times \left(111.413 \times \cos\left(LatB \times \frac{\pi}{180}\right)\right) \\ \quad - 0.094 \times \cos\left(3 \times LatB \times \frac{\pi}{180}\right)) \\ \Delta Lat = (LatA - LatB) \times 1000 \times \left(111.133 - 0.59 \times \cos\left(2 \times LatB \times \frac{\pi}{180}\right)\right) \\ \quad \text{Distance} = \sqrt{\Delta Ion^2 + \Delta lat^2} \end{array} \tag{14}$$

The relationship between positioning error and vehicle speed is shown in Figure 12.

**Figure 12.** The relationship between positioning error and vehicle speed.

The positioning error is decided by several factors, such as the number of satellites used for positioning, vehicle speed, NB-IoT life cycle, etc. The average number of observable satellites under the GPS/BDS dual mode is 9, and the positioning performance is better than that of GPS single system and BDS single system. The main reason is that the number of available satellites increases, and the geometry configuration is enhanced. Under BDS/GPS dual positioning system, more positioning satellites can be obtained, so the accuracy is higher than that of GPS or BeiDou single positioning system. When the vehicle is static, the positioning error is about 2 m. When the car moves, the positioning error increases. The faster the car speed, the greater the positioning error. When the life period of NB-IoT is set as 3600 s, and the speed is less than 40 km/h, the positioning error is less than 10 m. When the speed is about 60 km/h, the positioning error is about 20 m. The larger the life cycle of NB-IoT, the greater the positioning error, because the larger the life cycle of NB-IoT, the greater the transmission latency.

#### *3.2. Network Performance Test*

Network performance test includes data compression rate and transmission packet loss rate. Limited by the experimental conditions, it is impossible to obtain a large number of test data of sensor nodes. Therefore, the experimental data on data compression rate is taken from the temperature data of Intel-Berkeley University Joint Research Laboratory in reference [40]. Compare the compression ratio between the AOZS algorithm proposed in this paper and the commonly used DCCM (Differential Code Compression Method) algorithm, as shown in Figure 13. Under the condition of the same amount of node data collection, the compression ratio of AOZS algorithm is lower than DCCM algorithm, and the compression performance is better, because AOZS algorithm makes full use of the correlation between data, and the coding based on the optimal bit factor removes the redundant information to the greatest extent. The more data the node collects, the higher the time correlation of the data. The coding factor of AOZS algorithm can describe more original data and fully mine the time correlation of data. Therefore, the compression ratio becomes smaller and smaller and tends to be stable gradually. With the increase of the number of sensor data, the compression rate of DCCM algorithm is maintained at about 50%, and that of AOZS algorithm is maintained at about 10%.

**Figure 13.** The Comparison of data compression ratio.

Another indicator of network communication reliability is packet loss rate. Sx1268 is a new generation 433 MHz LoRa half duplex transceiver chip produced by Semtech in 2018. It is also one of the commonly used Lora chips at present. So we compare the communication reliability between Sx1268 LoRa module and our wh-nb75-ba NB-IoT module. Figure 14 is the comparison of packet loss rate of our NB-IoT module and LoRa Sx1268 module under the same transmitting and receiving condition. When the distance is less than 250 m, the packet loss rate of both circuits is nearly 0. With the increase of distance, the packet loss rate of LoRa module increases significantly, while that of NB-IoT module increases little. When the distance is 400 m, the packet loss rate of LoRa module is about 1.5%, that of NB-IoT module is still nearly 0. when the distance is 600 m, the packet loss rate of LoRa module is about 2%, that of NB-IoT module is about 0.5%. when the distance is 800 m, the packet loss rate of LoRa module is about 5%, that of NB-IoT module is about 0.7%. when the distance is 800 m, the packet loss rate of LoRa module is about 5%, that of NB-IoT module is about 0.7%. When the distance is 1000 m, the packet loss rate of LoRa module is about 10%, that of NB-IoT module is about 1%. When the distance is 1200 m, the packet loss rate of LoRa module is about 15%, that of NB-IoT module is about 1.2%.

**Figure 14.** The packet loss rate of our NB-IoT module and LoRa sx1268.

#### **4. Discussion**

Through these tests aboved, the monitoring system realizes the higher position precision. It is shown that the BDS/GPS dual mode position have higher position precision than that of single BDS or GPS. When the monitored target is stationary, the positioning accuracy is only determined by the positioning module. The position calculation formula under the

dual-mode positioning module is deduced as above. When the monitored target moves, the positioning accuracy is jointly determined by the positioning module, vehicle speed and life cycle. However, under the same vehicle speed and the same life cycle of NB-IoT, the monitoring system accuracy of dual-mode positioning is still higher than that of single BDS or GPS positioning mode. Considering the characteristics of sensor data in monitoring system, an adaptive optimal zero suppression (AOZS) algorithm based on time correlation is proposed in this paper. After testing and comparing with the commonly used differential code compression method (DCCM) algorithm, the data compression rate of the new algorithm can be as high as 90%, which greatly reduces the amount of data transmission in the communication network and improves the network performance and transmission efficiency. With the increase of the number of sensor data, the compression rate of DCCM algorithm is maintained at about 50%, and that of AOZS algorithm is maintained at about 10%. Packet loss rate is the main indicator of communication network performance. We tested and compared the packet loss rate of the monitoring system based on wh-nb75-ba NB-IoT module and the monitoring system based on sx1268 LoRa module which is mainly used. When the distance is less than 250 m, the packet loss rate of both circuits is nearly 0. With the increase of distance, the packet loss rate of LoRa module increases significantly, while that of NB-IoT module increases little. The greater the distance, the greater the difference between the packet loss rate data of the two circuits.

#### **5. Conclusions**

A new monitoring system is proposed in the paper, based on NB-IoT and BDS/GPS dual-mode positioning. The whole monitoring system includes three parts: sensor node, sink node and monitoring center. The sensor node which is based on cc1101 RFID circuit realizes the detection of surrounding temperature and humidity and data transmission. The sink node receives and compresses the temperature and humidity data from the sensor node, obtains the positioning information through at6558 BDS/GPS positioning module, and uploads these data to the cloud through wh-nb75-ba NB-IoT module. The monitoring center can download data from the cloud and save it to the local machine, and can analyze the historical data through an operation interface.

Experiments and analysis show that the proposed scheme has better positioning accuracy, better data compression ratio and transmission performance. The temperature and humidity error are less than 1 ◦C and 5% RH especially with the selected chip. The position error is decided by several factors, including the number of satellites used for positioning, the monitored target moving speed and NB-IoT module lifetime period. When the monitored target is stationary, the positioning error is about 2 m, which is less than that of the single GPS or BDS mode. The AOZS compression algorithm is used to improve compression ratio (CR). The CR is about 10% when the data amount increasing.

The scheme of this paper had encouraged experiments and was efficient and practicable in monitoring system. However, many aspects, still need to be further studied, such as transmission delay, multi-sensor nodes and low-power circuits. Furthermore, optimizing the network structure to reduce its consumption and accomplishing end-to-end network will be the main direction of our work.

**Author Contributions:** Conceptualization, Z.X. and R.Z.; methodology, Z.X.; software, Z.X. and R.Z.; validation, J.F. and L.Z.; formal analysis, J.F.; investigation, L.Z.; writing—original draft preparation, Z.X.; writing—review and editing, R.Z.; visualization, J.F.; funding acquisition, Z.X. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was supported by the Commonweal Projects of Zhejiang Province (Grant No. LGN20F010001) and General Project of Zhejiang Education Department (Grant No. Y201940951).

**Conflicts of Interest:** The authors declare no conflict of interest.
