IMPROVING THE ACCURACY AND ENERGY EFFICIENCY OF FREQUENCY SENSOR SIGNAL MEASUREMENTS IN WIRELESS NETWORK NODES

Abstract

The article addresses the actual scientific and applied problem of improving measurement accuracy and reducing power consumption in wireless sensor network (WSN) nodes that use primary transducers with a frequency output signal. Traditional approaches to integrating WSNs with cloud platforms under the Sensor-Cloud concept face the challenge of a strictly limited energy budget of autonomous edge nodes, which restricts their operational lifetime. The use of frequency sensors makes it possible to eliminate power-consuming analog-to-digital converters from the signal processing pipeline; however, it imposes specific and non-trivial requirements on the frequency measurement algorithms implemented on the microcontroller side. It has been proven that the classical direct frequency counting method is highly inefficient for low-frequency signals, since it requires extended processor active time proportional to the measurement gate interval, which leads to a linear increase in energy consumption and introduces unacceptable latency in data transmission to the gateway. To overcome this fundamental limitation, the expediency of applying the reciprocal counting method (period measurement) combined with Edge Computing algorithms is rigorously justified. The use of simple linear regression — the ordinary least squares method — for processing an array of signal edge timestamps directly on the microcontroller is proposed as the core algorithmic contribution. This approach significantly reduces the impact of phase jitter and comparator hardware noise on the overall quantization error without the need to increase the measurement time window. Mathematical analysis and simulation modelling of the proposed algorithms were carried out in the Python environment with AWGN jitter model (σ = 5 μs). It is demonstrated that the regression-based frequency estimator reduces the error variance inversely proportionally to the square root of the number of processed samples, thereby providing stable and predictable resolution across a wide operating frequency range of the sensor. Additionally, the impact of the selected measurement method on the overall energy balance of the node is analysed when using lightweight IoT application-layer data transfer protocols such as MQTT-SN and CoAP. It has been established that shifting the computational load to the sensor node through local data aggregation and adopting an event-driven message delivery model makes it possible to reduce the volume of transmitted radio traffic by a factor of N, where N is the number of processed edges. The results of the study can be used in the design of energy-efficient and highly autonomous information-measuring systems intended for long-term unattended operation in IoT and Sensor-Cloud deployments.