*6.2. Results and Analysis*

The experimental data of all nodes is organized as shown in Table 4. Since nodes 6 to 10 were without algorithms, the transmission power was set to a maximum of 10 dBm and the data rate was set to the lowest at 50 kbps. The PERs of nodes 6 to 10 were lower than the PERs of nodes 1 to 5 under the data rate and the transmission power control algorithm. However, the nodes with algorithms maintained a PER of less than 1% except for node 1, and the average current consumption was much lower than that the nodes without algorithms.


**Table 4.** Node experimental results.

Comparing node 1 with node 6, the PER of node 1 was higher than node 6. However, node 1 saved 78.74% more power consumption than node 6 in response packet. In the overall average current consumption, node 1 saved 51.64% more of the energy than node 6.

Node 2 and node 7 were the nearest nodes for the bridge node. In Table 4, the PER of node 2 was 0.6322%, and node 7 was 0.1518%. Although the PER of node 2 was larger than that of node 7, its overall PER was still less than 1%. In the response packet, node 2 saved 73.67% more power consumption than node 7, and in the overall average current consumption, it saved 48.43% more power consumption.

The PER of node 3 was 0.9532%, and node 8 was 0.0912%. In the response packet, node 3 saved 70.49% more power consumption than node 8, and in the overall average current consumption, it saved 46.36% more power consumption.

The experimental result is showed in Figure 14. The PER of node 4 was 0.8429%, and node 9 was 0.3331%. In the overall average current consumption, node 4 was 44.991 uA, and node 9 was 74.281 uA. In the response packet, node 4 saved 59.59% more power consumption than node 9, and in the overall average current consumption, it saved 39.43% more power consumption.

**Figure 14.** Experimental results of (**a**) node 4 and (**b**) node 9.

The PER of node 5 was 0.4011%, and node 10 was 0.3376%. In the overall average current consumption, node 5 was 32.516 uA, and node 10 was 74.281 uA. In the response packet, node 5 saved 85.83% more power consumption than node 10, and in the overall average current consumption, it saved 56.23% more power consumption.

Obviously, the RSSI values of nodes with algorithms are close to the sensitivity values corresponding to the current data rate from the experimental results. If the RSSI value is lower than the sensitivity value, the probability of packet error will increase. Moreover, since the position of the nodes is affected by the people in the office and the class, the RSSI value of each node floats dramatically during the daytime and the probability of packet error is high. However, at night, the RSSI value is so stable that the probability of packet error is low. If the node is supplied by AAA battery with 1200 mAh, the execution duration could approach about 4.21 years with the proposed algorithm. The nodes without the proposed algorithm could run only for 1.8 years.

The data of Table 4 is drawn with a bar graph of the nodes' average current consumption as Figure 15. It is clear the response packet of the nodes with the algorithm saved the most significant energy in the TX Mode, and the ranking of the power saving was sequentially ranked as nodes 5, 1, 2, 3, and 4. The reason for the difference in the amount of power consumed by each node was assumed to be the positional relationship.

**Figure 15.** Nodes overall average current consumption.

For justification of the proposed algorithm, the experimental results for long-term testing are presented in Figure 16 and Table 5. Figure 16 shows the battery voltage variation in node 4 and node 9, which ran with and without control algorithm for 69 days of execution. The battery voltage of the node 9 in Table 5 was obviously lower than that of node 4, and the results also verify the proposed algorithm workable.

**Figure 16.** The battery voltage of (**a**) node 4 and (**b**) node 9 within 69 days.


**Table 5.** The battery voltage of each node at the 69th day.

#### **7. Conclusions**

A wireless sensor network based on Sub1G-Hz and star topology was constructed in this paper, and the TDMA wireless communication protocol and the transmission power and data rate control algorithm were proposed to reduce power consumption on sensing nodes usefully. According to the dynamic environment, sensing nodes with hybrid control algorithms automatically adapt the transmission power and data rate to achieve good communication quality and low power consumption simultaneously. The above algorithms will increase the performance and reduce power consumption on wireless communication. Then, communication devices could be operated at very low power consumption when using wireless communication.

The experimental results show that PER states of nodes can effectively be controlled near the target value, 1%, which can prove the good reliability of communication. In addition, because all nodes run in the TDMA architecture's wireless protocol, TDMA can enable a wireless transmission in a low duty cycle. The average current consumption of the node without the hybrid control algorithm was calculated as 74.281 uA, and the power consumptions of algorithm nodes were different and depended on the positions of the nodes. According to the experimental results, when the power consumption of the response packet in the transmission mode was compared, the power consumption saved up to 85.83%. The overall consumption saved up to 56.23% of the power consumption, which indicates that the algorithms proposed in this paper actually have an energy-saving effect for wireless communication. If the node is powered by AAA battery with 1200 mAh, the node could run approximately 4.21 years with proposed algorithm. The other TDMA is discussed in [17] for the LoRa system but not included about ADR and TP, and the authors suggest that the battery life is about 3 years.

In summary, the proposed hybrid control algorithm is complex, and the payload and node number in the WSN are also limited. However, the system architecture and control algorithm proposed in this paper could lower several important things such as the power consumption, system complexity, maintenance fee, etc.

**Author Contributions:** All authors contributed ideas, discussed the results. C.-W.H. and H.-J.Z. designed the experiment, and performed most of the analysis. H.-J.Z. compiled the data. C.-W.H. and W.-T.H. wrote most of the main text. Y.-D.Z. supported the important revision of the article. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Ministry of Science and Technology, ROC, grant number under contract No. MOST 109-2221-E-224-023-, 108-2221-E-224-045- and 108-2218-E-150-004-.

**Conflicts of Interest:** To the best of our knowledge, the authors have no conflict of interest, financial or otherwise.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Sensors* Editorial Office E-mail: sensors@mdpi.com www.mdpi.com/journal/sensors

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18