The design of low-power sensor nodes has become a pivotal technology driving advancements in smart cities, environmental monitoring, and health management. Monitoring systems centred on temperature and humidity, PM2.5, and human body infrared sensors must extend node operational lifespans from days to years while meeting real-time and accuracy requirements. This paper systematically outlines core low-power design strategies and engineering practices across four dimensions: sensor selection, power management, data processing, and communication protocols.

I. Power Optimisation Pathways in Sensor Selection
The energy consumption characteristics of sensors directly influence overall node power usage. Within temperature and humidity sensing, significant power differences exist between digital and analogue products. Taking the SHT31 digital sensor as an example, its operating current is 80μA (at 1.8V supply), whereas the analogue-output HM1500 consumes only 15μA under identical conditions, though it requires an external ADC conversion circuit. Practical testing demonstrates that a solution employing the HM1500 paired with a 16-bit ADC reduces average daily power consumption by 62% compared to the SHT31 solution, operating at a 10-second sampling interval.
PM2.5 sensor selection requires balancing accuracy and power consumption. Laser scattering sensors (e.g., Plantower PMS7003) provide tri-channel PM1.0/PM2.5/PM10 data but consume 80mA, rapidly depleting batteries during continuous operation. In contrast, infrared scattering sensors (e.g., Sharp GP2Y1010AU0F) operate at just 20mA but offer limited accuracy. Practical application at an environmental monitoring node demonstrates that employing a dynamic switching strategy—activating the laser sensor every 30 minutes under normal conditions and shortening this to 5 minutes during anomalies, while maintaining continuous monitoring with the infrared sensor—reduces average daily power consumption from 12mAh to 3.8mAh.
Power consumption optimisation for passive infrared (PIR) sensors focuses on enhancing detection algorithms. Traditional PIR modules operate at approximately 50μA but suffer from false triggering. Introducing a threshold adaptive algorithm reduces detection sensitivity by 30% during stable ambient temperatures, cutting invalid triggers by 75%. Field data from a smart security node shows this strategy reduced the PIR module's average daily power consumption from 0.6mAh to 0.15mAh while maintaining 98% detection accuracy.

II. Refined Design of Power Management Systems
The selection of power management integrated circuits (PMICs) directly impacts energy conversion efficiency. Taking the TPS62740 as an example, this chip achieves 90% efficiency in light-load mode (output current < 1mA), representing a 40% improvement over linear regulators. In an agricultural monitoring node application, the TPS62740 powers sensors alongside a supercapacitor for transient energy buffering. This configuration maintains voltage fluctuations within ±2% across temperatures from -20°C to 60°C while reducing static power consumption from 15μA to 3μA.
Integrated energy harvesting technologies further extend node lifespan. The combination of solar cells and thermoelectric generators (TEGs) is particularly exemplary. In indoor environments, TEGs harness the temperature differential between the human body and surroundings (typically 2–5°C) to generate 50–200 μW of electricity. Testing on a wearable device demonstrated that, paired with a 200 mAh lithium battery, the TEG module extended device runtime from 7 to 21 days. Key design elements include employing MPPT algorithms for real-time maximum power point tracking and utilising low-leakage capacitors for energy storage.
Dynamic Voltage Scaling (DVS) technology significantly reduces processor power consumption by adjusting supply voltage in real-time according to load demands. Taking the STM32L0 series MCU as an example, when running temperature and humidity data processing algorithms, reducing the voltage from 3.3V to 1.8V can decrease power consumption by 58% while only reducing processing speed by 12%. In practical node design, voltage should be elevated during data acquisition phases and lowered during idle periods, synchronised with sensor sampling cycles, to form a dynamic power management closed-loop.

III. Power Control via Data Processing Strategies
Data compression algorithms reduce transmission energy consumption. For temperature and humidity data's periodic characteristics, a combined differential encoding and Huffman compression approach is employed. Test data indicates that the original data packet size of 12 bytes is reduced to 3 bytes after compression, shortening transmission time by 75%. In NB-IoT communication scenarios, the energy consumption per data transmission decreases from 4.2mJ to 1.3mJ, reducing the node's average daily power consumption by 0.8mAh.
The introduction of edge computing enables localised data processing. Taking PM2.5 monitoring as an example, nodes incorporate threshold-based algorithms that activate communication modules only when concentrations exceed safety thresholds. Implementation in an industrial park demonstrated a 92% reduction in communication frequency. Combined with LoRaWAN's low-power characteristics, this extended node endurance from three months to two years. Key technologies include: designing lightweight neural network models for data anomaly detection, and optimising Flash memory read/write strategies to reduce energy consumption.
Dynamic adjustment of sampling frequency is central to balancing accuracy and power consumption. In human infrared sensor applications, historical trigger data analysis establishes temporal models, shortening sampling intervals to 1 second during peak occupancy periods and extending them to 30 seconds during off-peak periods. Testing on a smart office node demonstrated this strategy reduced PIR module power consumption from 0.6mAh daily to 0.2mAh, while maintaining over 95% event capture rate.

IV. Low-Power Implementation of Communication Protocols
Within Low-Power Wide-Area Network (LPWAN) technologies, LoRa and NB-IoT exhibit markedly different power consumption characteristics. LoRa consumes merely 12mA reception current at spread factor SF=12 yet achieves transmission distances up to 15km; NB-IoT, whilst offering direct cellular network access, incurs standby power consumption as high as 200μA due to its persistent connection mode. Comparative testing on an agricultural monitoring system demonstrated that LoRa nodes, uploading data four times daily, exhibited an annual power consumption of 1.2Ah – a 65% reduction compared to NB-IoT implementations.
Bluetooth Low Energy (BLE) 5.0's periodic advertising mode offers a novel solution for short-range communication. By extending the advertising interval from 100ms to 1s while enabling 2Mbps high-speed mode for compressed data transmission, a health monitoring wristband reduced its daily power consumption from 1.8mAh to 0.7 mAh. Key optimisations included: utilising BLE's channel selection algorithm to avoid interference bands, and designing an adaptive retransmission mechanism to ensure data reliability.
Wake-on-Receive (WOR) technology enables on-demand activation of communication modules. Traditional RF receivers continuously monitor channels, consuming up to 15mA; WOR circuits intermittently activate front-end amplifiers, reducing average power consumption to 50μA. In a wildlife tracking node application, combined with an FSK modem, this reduced energy consumption per communication from 3mJ to 0.8mJ, extending node endurance from 45 days to 180 days.

V. Power Consumption Analysis of Typical Application Cases
The design of a smart building environmental monitoring node serves as a highly representative example. This node integrates an SHT40 temperature and humidity sensor (operating current 1.2μA), a GP2Y1010AU0F PM2.5 sensor (20mA@5V), and an EKMB1101111 PIR sensor (50μA), utilising an STM32L476RG ultra-low-power MCU. Through implementing the following strategies: temperature and humidity sampling at 10-minute intervals, PM2.5 sampling at 30-minute intervals (shortened to 5 minutes during anomalies), and enabling a dynamic threshold algorithm for the PIR sensor at night; employing LoRa for the communication module with three daily data uploads; and integrating the TPS62740 with a 200mAh lithium battery for power management. Field testing demonstrates a 18-month operational lifespan in typical office environments, representing a fourfold improvement over conventional designs.
Designing low-power sensor nodes constitutes a system-level engineering endeavour, requiring end-to-end optimisation spanning sensor characteristics, power architecture, data processing, and communication protocols. With breakthroughs in energy harvesting technology and the convergence of AIoT, future nodes will achieve ‘zero-power’ operation—self-sustaining through environmental energy and intelligent task scheduling, thereby eliminating the maintenance burden of battery replacement. This technological evolution will not only drive large-scale deployment of IoT devices but also lay the foundation for building sustainable smart ecosystems.