The Wireless Internet of Things (WIoT) is experiencing rapid growth, driven by applications in smart homes, industrial automation, and medical monitoring. These applications often require low-frequency (tens of kHz) long-range communication with central clouds for energy-efficient operation and decision-making. A typical WIoT data transmission link consists of data acquisition sensors, pre-processing hardware, a frequency converter, and radio frequency modules. The frequency converter is a critical component, synthesizing signals at different frequencies based on band requirements and mixing them with sensor datastreams to improve transmission efficiency. Current state-of-the-art frequency converters rely on CMOS-based digital and analog circuits with separate frequency synthesizer and mixer circuitry. This approach, while functional at high frequencies (GHz), introduces excessive latency and power consumption, particularly problematic for low-frequency WIoT applications. The need for a frequency converter capable of in-situ synthesis and mixing within a single module to improve latency and energy efficiency is paramount for the advancement of WIoT technology. Memristors, with their tunable oscillation frequencies stemming from rich ion dynamics and electrical behaviors, offer a compelling alternative to CMOS-based solutions. While oscillatory bionic neurons based on memristors have been reported, their application in frequency converters remains unexplored. A calibratable memristor oscillator array could potentially achieve in-situ synthesis and mixing in a single module, greatly improving latency and power efficiency. This study aims to address this gap by presenting a novel VO₂ memristor-based frequency converter designed for energy-efficient in-situ frequency synthesis and mixing in low-frequency WIoT applications. The research focuses on the development of a highly uniform and calibratable VO₂ memristor oscillator array, its integration into a frequency converter, and its performance evaluation in a real-world WIoT experimental setup.

Existing literature extensively covers the advancements and challenges in WIoT networks, emphasizing the need for energy-efficient data transmission. Numerous studies detail the use of CMOS-based frequency converters in various applications, showcasing their capabilities at higher frequencies. However, the limitations of CMOS-based solutions in low-frequency, energy-constrained WIoT environments have been identified, highlighting the need for alternative approaches. Research on memristors and their unique properties, particularly their potential for creating tunable oscillators, has also received considerable attention. While the use of memristors in neuromorphic computing, mimicking biological neurons, has been demonstrated, the specific application of memristor oscillators in frequency converters for WIoT is relatively unexplored. This paper aims to bridge this gap by presenting a novel approach to frequency conversion based on VO₂ memristors, addressing the efficiency challenges inherent in current CMOS-based solutions for low-frequency WIoT communication.

This research involved the design, fabrication, and characterization of a VO₂ memristor-based frequency converter. The study began with the fabrication of high-quality epitaxial VO₂ memristors using pulsed laser deposition (PLD) on c-Al₂O₃ substrates. The quality of the VO₂ films was meticulously characterized using techniques such as scanning electron microscopy (SEM), transmission electron microscopy (TEM), X-ray diffraction (XRD), Raman spectroscopy, and X-ray photoelectron spectroscopy (XPS). These characterizations confirmed the high crystalline quality and the desired electrical properties of the VO₂ film, crucial for achieving high uniformity and reliability. A detailed electro-thermal compact model was developed using SPICE simulations, enabling accurate prediction of the memristor's I-V characteristics. A VO₂ memristor-based oscillator was then designed, consisting of a VO₂ memristor connected in parallel with a capacitor, utilizing the memristor's negative differential resistance (NDR) for self-oscillation. The programmability of the oscillator was investigated by varying the applied current and parallel capacitance. To improve device-to-device uniformity, a parallel calibration resistor was incorporated into the oscillator circuit. The self-oscillation behavior was carefully analyzed, and the relationship between oscillation frequency and applied current, capacitance, and calibration resistance was experimentally determined. This analysis was then used to design an 8x8 VO₂ memristor array-based frequency converter capable of in-situ frequency synthesis and mixing. The converter utilizes a compact circuit design enabling multiple current sources to drive the individual memristors in the array, with the output signals from the individual memristors combined to achieve both frequency summation and subtraction. An end-to-end WIoT experimental setup was designed, incorporating a software-based pre-processor for encoding and modulation of sensor data (acoustic, vision, and spatial), the VO₂-based frequency converter, and a receiver for demodulation and decoding. The performance of the VO₂-based frequency converter was compared to a conventional CMOS-based design in terms of power consumption and bit error rate (BER) across a range of signal-to-noise ratios (SNRs).

The research successfully demonstrated a highly uniform and calibratable VO₂ memristor oscillator. The oscillator exhibited excellent cycle-to-cycle and device-to-device uniformity due to the high crystalline quality of the epitaxially grown VO₂ and the incorporation of parallel calibration resistors. The oscillation frequency could be effectively programmed by adjusting the driving current and calibration resistance. The developed 8x8 VO₂ memristor array-based frequency converter demonstrated the capability for in-situ frequency synthesis and mixing for 2-8 channels, with frequencies up to 48 kHz. The experimental results showed that the output signal from the converter accurately reflected the summation or subtraction of the individual input frequencies. In the end-to-end WIoT experimental setup, the VO₂-based frequency converter outperformed the conventional CMOS-based design, exhibiting a power consumption improvement of 1.45x-1.94x with minimal (0.02-0.21 dB) BER degradation across different sensor data types (acoustic, vision, and spatial). The high-crystalline-quality epitaxial VO₂ memristors played a crucial role in achieving this performance. The addition of the parallel calibration resistor significantly reduced device-to-device variations, enhancing the overall uniformity and reliability of the frequency converter. The combination of in-situ synthesis and mixing capabilities, significant power savings, and minimal performance degradation compared to CMOS-based solutions showcases the potential of the proposed VO₂ memristor-based frequency converter for practical application in WIoT systems.

The findings of this research directly address the challenges associated with current CMOS-based frequency converters in low-frequency WIoT applications. The demonstrated power savings of 1.45x-1.94x and minimal BER degradation underscore the significant potential of VO₂ memristor-based frequency converters for enhancing energy efficiency in WIoT networks without compromising performance. The ability to perform in-situ frequency synthesis and mixing within a single module reduces latency, which is crucial for real-time data transmission in many WIoT applications. The successful integration of the VO₂-based frequency converter into an end-to-end WIoT experimental setup further validates its practical applicability. The high crystalline quality of the epitaxially grown VO₂ and the innovative use of calibration resistors address the issues of device uniformity, ensuring the reliability and scalability of the design. This work opens a new avenue for designing more efficient and reliable frequency converters for low-frequency WIoT applications. Future work could focus on further optimizing the design, exploring different memristor materials, and investigating potential applications beyond frequency conversion in other areas of WIoT.

This research successfully demonstrated a novel VO₂ memristor-based frequency converter capable of in-situ synthesis and mixing, offering significant power savings and minimal performance degradation compared to CMOS-based counterparts in a real-world WIoT setting. The high uniformity and programmability of the VO₂ memristor oscillator array are key contributors to its superior performance. This work highlights the potential of VO₂ memristors for developing energy-efficient and high-performance components for next-generation WIoT systems. Future research could explore larger-scale memristor arrays, explore integration with other low-power communication technologies, and investigate applications in diverse areas within WIoT.

While this research presents compelling results, some limitations should be noted. The current design and implementation are limited to an 8x8 memristor array, and scaling to larger arrays might introduce challenges related to device uniformity and control circuitry complexity. The experimental setup used a simplified AWGN channel model for noise simulation. More comprehensive channel models including multipath fading and interference could provide a more realistic performance evaluation. The temperature dependence of the VO₂ memristor's oscillation frequency, although considered, might require further investigation for optimal operation across diverse environmental conditions.