A Review of Circuits and Systems for Advanced Sub-THz Transceivers in Wireless Communication

Abstract

Sub-terahertz (sub-THz) frequencies (100–300 GHz) are gaining prominence in the development of next-generation wireless communication systems, promising ultra-high data rates and wide bandwidths essential for applications like 6G networks and beyond. Despite the immense potential of these frequencies, several design and implementation challenges remain, especially in transceiver architectures, high-order modulation, and beam-forming capabilities. In this paper, we survey recent advances in sub-THz transceiver design, with a particular focus on D-band frequencies. We explore the latest developments in circuit performance and architectures, including innovative transmitter and receiver designs that utilize direct-digital modulation (and demodulation) and phased-array systems. To ensure a comprehensive and up-to-date analysis, this work selects over 100 data points from top-tier conferences and journals, with most publications dating within the past five years, reflecting the state of the art in the field. Meanwhile, we discuss practical challenges, future directions, and opportunities to optimize sub-THz systems for high-speed, high-capacity wireless communication.

1. Introduction

The rapid growth of wireless data demand and the push for ubiquitous connectivity have driven the exploration of higher frequency spectra, specifically sub-terahertz (sub-THz) frequencies, for next-generation wireless communication systems. Frequencies in the sub-THz range (100–300 GHz) lie between millimeter wave and terahertz, combining the W-band (75–110 GHz), D-band (110–170 GHz), G-band (140–210 GHz), and J-band (220–320 GHz), as illustrated in Figure 1 [1,2]. Sub-THz frequencies hold immense potential, offering wide bandwidths that can support data rates greater than 100 Gbps, critical for emerging applications such as 6G networks, immersive augmented reality (AR), wireless interconnects in data centers, imaging, and high-capacity wireless backhaul [3,4,5]. These frequencies offer advantages that include improved spatial resolution, high-speed data transmission, and lower latency, all essential for addressing the needs of dense urban environments, industrial automation, and long-range satellite and inter-satellite links [1,6,7,8]. Furthermore, the integration of sub-THz frequencies into wireless systems aligns with the global standardization efforts for 6G and beyond, providing a path to bridge the gap between terahertz research and practical implementation. According to Edholm’s law, the bandwidth and data rate of telecommunications exhibit an exponential rate of growth, which is similar to Moore’s law [9]. The development of data rates in wireline, nomadic, and wireless systems is plotted in Figure 2, clearly showing the trend of Edholm’s law.

Despite the promising attributes of sub-THz communication, several challenges remain. High free-space path loss, limited power efficiency, and signal propagation constraints due to atmospheric absorption are inherent at these frequencies, making the design of effective sub-THz transceivers particularly challenging [10]. On the other hand, the sub-THz spectrum offers a largely untapped resource, with significant portions of it allocated as unlicensed spectrum by the Federal Communications Commission (FCC), particularly in the 120 to 170 GHz range [1]. This unlicensed sub-THz spectrum is critical for future wireless standards that demand higher data rates, lower latency, and enhanced reliability to support applications like 6G, industrial automation, and ultra-fast wireless backhaul.

The link capacity of sub-THz wireless communication is intricately tied to the modulation scheme, the achievable signal-to-noise ratio (SNR), and the error vector magnitude (EVM) requirements for different modulation indexes. Shannon’s capacity theorem provides the theoretical limit of the data rate:

$$C=B·{log}_2(1+\mathrm{SNR}),$$

(1)

where B is the channel bandwidth. Achieving high data rates in sub-THz systems necessitates high-order modulation schemes, which demand stringent SNR and EVM performance. The theoretical limits of link capacity provided by Shannon’s theorem guide the practical selection of modulation schemes to maximize efficiency under realistic constraints. The dependency of the bit error rate (BER) on EVM for various modulation formats, such as QPSK, 16 QAM, and 64 QAM, is illustrated in Figure 3. For instance, transitioning from QPSK to 16 QAM increases the spectral efficiency but requires significantly lower EVM to maintain a tolerable BER. This trade-off underscores the critical role of linearity and noise performance in transceiver design. At sub-THz frequencies, maintaining low EVM is particularly challenging due to high phase noise, non-linearity in power amplifiers, and mixer-induced distortions, which collectively degrade signal quality.

Sub-THz wireless communication requires innovative transceiver architectures that can overcome these physical and technical challenges. This work surveys the latest research status in sub-THz communication systems, with key enabling features including high-order modulation schemes, high data rates, direct-digital modulation and demodulation techniques, and advanced phased-array systems. On the transmitter (TX) side, architectures employing direct-conversion [11,12,13] or direct-digital modulation [14,15,16,17] are shown to simplify design complexity while supporting high data rates. Transmitters employ advanced power amplifiers (PAs) to address trade-offs between efficiency, linearity, and power output using techniques such as multi-way power-combining [18,19] and outphasing [20,21,22]. Advances in receiver (RX) include heterodyne [23] and direct-conversion architectures [24], along with sophisticated noise mitigation and EVM management techniques that enable robust performance even at high frequencies. Direct-digital demodulation methods are also proposed. Integrated transceiver (TRX) systems now incorporate full-duplex designs with self-interference cancellation (SIC) and LO feedthrough suppression to improve efficiency and reduce signal degradation. Phased-array systems enhance spatial multiplexing and beam steering, utilizing both analog and hybrid beam-forming techniques for precise directional communication. These innovations collectively highlight the potential of sub-THz communication to meet the demands of next-generation wireless networks.

The advancement of semiconductor technology plays a crucial role in enabling mm-Wave and sub-THz communication systems. SiGe BiCMOS and advanced CMOS nodes have become dominant contenders in this space, benefiting from improved transit frequency ($f_t$) and maximum oscillation frequency ($f_{max}$), allowing the realization of high-performance RF front-ends with increased integration levels [8,25]. Though III-V technologies, such as InP HBTs and GaAs pHEMTs, still provide superior power efficiency at high frequencies, their cost and fabrication complexity make them less attractive for large-scale commercial applications [26]. It should also be noted that FinFETs have gained popularity in recent years due to competitive power efficiency and noise performance ($N{F}_{min}$) [27,28,29].

By providing a comprehensive overview of recent research, this review identifies the current challenges and future directions for sub-THz transceivers. This article is organized as follows. Section 2 focuses on transmitter design considerations, covering key components such as heterodyne and direct-conversion architectures, signal generation methods using frequency synthesizers and multipliers, phase-shifting techniques, and advanced power amplifier designs. Section 3 delves into receiver design challenges, comparing direct and heterodyne down-conversion, LNA-first and mixer-first architectures, and recent advances in demodulation methods (RF-DACs, six-port, and Kramers–Kronig coherent receiver). Section 4 explores transceiver system design, addressing full-duplex operation with single and double antennas, phased-array beam steering for high-data-rate applications, and multiple-input multiple-output (MIMO) systems. The article concludes with a summary of key findings and future directions for sub-THz communication technologies.

2. Transmitter System and Component Design Considerations

Figure 4a,b provide a comprehensive overview of sub-THz wireless communication link performance, highlighting key metrics such as data rate, modulation scheme, and transmitter’s saturated output power ($P_{sat}$) or its effective isotropic radiated power (EIRP). More than 70 data points were collected, predominantly sourced from studies published in the last five years, to capture the forefront of research developments. Data rates peak at 200 Gbps for the D-band [30,31] and 160 Gbps for the J-band [32]. Among the frequencies of interest, the highest modulation scheme recorded by mm-Wave direct-digital transmitters is 512 QAM on the E-band [33], 256 QAM on the D-band [34,35], and 128 QAM on the J-band [36]. While the majority of studies employ QAM modulation, alternative schemes such as OOK [37,38], QPSK [39,40], APSK [41], and FSK [42] have also been documented. There are also works that showcase the multi-channel operation at sub-THz [43,44,45,46]. Additionally, there is a perceptible decrease in the sub-THz transmitter’s $P_{sat}$ and power efficiency as frequencies increase, as marked in Figure 4b. This is due to the degradation of the performance of the devices and the loss of metal at higher frequencies.

2.1. Heterodyne, Direct-Conversion, and Direct-Digital Modulation Transmitter

TX architectures in sub-THz communication systems are typically based on either traditional heterodyne (or multi-stage upconversion) designs or homodyne (direct-conversion) designs. In recent years, new architectures such as the direct-digital modulation transmitter have also been reported. Each method presents its own set of benefits and hurdles, especially as the operating frequencies approach sub-THz ranges. Figure 5 provides an overview of three primary architectures, which will be discussed in this section.

Heterodyne TX: In a heterodyne architecture (shown in Figure 5a), the signal generation process involves an intermediate frequency (IF) stage before final upconversion to the desired RF output. This approach allows the modulation to be performed at a lower frequency, where components such as mixers and modulators can achieve better linearity, accuracy, and stability [47,48,49]. One of the primary advantages of a heterodyne architecture is that it alleviates the need for the modulator to operate directly at sub-THz frequencies, which can be challenging and costly. For instance, achieving high linearity at sub-THz frequencies is more difficult than at a much lower IF, such as 100 MHz.

Nevertheless, heterodyne transmitters also come with trade-offs. In these systems, the complex-modulated IF signal requires higher resolution for accurate digitization or generation, as the IF signal contains precise amplitude and phase information. This level of detail is difficult to reproduce accurately and demands higher resolution components than those used in baseband (BB) generation. Furthermore, the multi-stage upconversion introduces potential challenges with image rejection and unwanted spectral components, which require filtering or complex frequency planning.

Homodyne TX: In contrast, homodyne, or direct-conversion transmitters, perform modulation directly from the baseband to the final RF output frequency, bypassing the need for an IF stage (shown in Figure 5b). Direct-conversion has gained traction in sub-THz applications because it offers a simpler architecture with fewer components, reducing power consumption and minimizing design complexity [11,12,13,50,51,52,53,54]. Direct-conversion simplifies signal generation by operating at baseband, where multi-level signals can be digitally generated with relative ease and transmitted directly to the output frequency. This design reduces power requirements and component count, making it highly suitable for compact, integrated sub-THz transmitters.

However, direct-conversion also has limitations. The lack of an IF stage requires high-precision mixers and modulators that can maintain linearity and spectral purity at sub-THz frequencies. Additionally, direct-conversion systems are susceptible to DC offset, I/Q imbalance, and LO feedthrough, which can degrade signal quality and reduce effective SNR. While digital pre-distortion and other calibration techniques can mitigate these issues, they add complexity to the system. In the context of complex-modulated signals such as 16 QAM, a homodyne system benefits from directly generated baseband signals that are easier to process, given that they contain only fixed amplitude levels without the added phase precision required in IF signals.

Direct-Digital Modulation TX (DDMT): DDMTs represent an innovative evolution in transmitter architectures for sub-THz communication, focusing on transforming bits directly into high-order modulation constellations (shown in Figure 5c). These transmitters bypass traditional components such as high-speed digital-to-analog converters by employing advanced modulation schemes to directly generate signals at the desired frequency using binary-weighted current-steering RF-DACs (shown in Figure 6, or other similar components), enabling high data rates with greater energy efficiency [14,55,56,57,58].

Recent advancements demonstrate the capability of DDMTs to achieve outstanding performance metrics. In addition to using RF-DACs, other methods of achieving direct-digital modulation have been reported. For example, a 135 GHz direct-modulation transmitter using separate bits for amplitude modulation and phase modulation on I/Q paths (shown in Figure 7a) achieved 16 QAM with 32 Gbps data rate, while consuming only 130 mW of DC power [59]. The design of the PAM2 amplitude and phase modulator is straightforward, which reduces the system overhead. Another design achieves 64 QAM with a data rate of 120 Gbps by summing the vectors of three QPSK outputs with different amplitudes, as shown in Figure 7b [17,60]. This method ensures flexibility in modulation schemes, allowing seamless transitions between QPSK, 16 QAM, or even higher-order modulation schemes based on the system SNR constraints. No power back-off is required for this design, which eases the output power limitation of the normal transmitter. The measured frequency spectrum, eye diagram, and constellation are shown in Figure 8. Furthermore, DDMTs inherently simplify the transmitter chain by eliminating IF stages and associated calibration challenges, thus reducing power consumption and system complexity. These results underscore the potential of DDMTs in meeting the demands of next-generation sub-THz wireless communication systems. However, direct modulation also introduces challenges related to limited signal-to-noise-and-distortion ratio (SNDR), I/Q mismatches, and higher sensitivity to EVM, which can impact the achievable modulation indexes. To address these limitations, techniques such as digital predistortion (DPD), adaptive biasing, and optimized circuit topologies are employed to improve signal integrity. Despite these challenges, recent implementations have demonstrated that direct modulation can support modulation schemes as high as 512 QAM with competitive power efficiency, making DDMTs a keystone technology for sub-THz wireless communication systems.

TX Above ${\mathbf{F}}_{\mathbf{max}}$: For transmitter designs operating beyond the transistor’s $f_{max}$ (>200 GHz), three key topologies are employed: mixer-last, multiplier-last, and cubic-mixer-last architectures [61]. The block diagrams are shown in Figure 9. The mixer-last topology upconverts an IF signal to the target frequency using a mixer driven by a LO signal at high frequency, offering high linearity and supporting complex modulation schemes such as QAM, but at the cost of increased power consumption and circuit complexity [62,63]. The multiplier-last approach amplifies the modulated IF signal before multiplying it to the desired frequency, allowing PAs to operate in saturation for higher efficiency, but limiting modulation to constant-envelope schemes such as QPSK [39]. The cubic-mixer-last topology combines frequency multiplication and mixing by leveraging non-linearities (sub-harmonic mixer), reducing circuit complexity, and achieving efficient signal generation with moderate linearity [46]. Each topology represents a trade-off between efficiency, linearity, and modulation complexity, tailored to the challenges of sub-THz transmitter design.

2.2. Signal Generation: VCO vs. Frequency Multiplier

In high-frequency TX designs for sub-THz communication systems, LO signal generation poses a challenge in balancing phase noise, power efficiency, and spectral purity. Two primary approaches for generating high-frequency signals include using voltage-controlled oscillators (VCOs) directly at the target frequency (Figure 10a) or employing frequency multipliers to upconvert a lower-frequency signal (Figure 10b). Due to their superior phase noise characteristics and better energy efficiency, frequency multipliers are often preferred over direct high-frequency VCOs in sub-THz systems.

One of the main reasons for using frequency multipliers over high-frequency VCOs is the inherent phase noise characteristics. As frequency increases, achieving low phase noise directly through high-frequency VCOs becomes challenging due to the degradation of the quality factor (Q) of on-chip passive components, which results in higher phase noise levels [64]. Frequency multipliers, on the other hand, allow for the generation of a high-frequency signal from a stable, low-frequency VCO with minimal phase noise degradation. Phase noise is typically increased by 20 × log(N) dB, where N is the multiplication factor, but this still results in a lower noise profile than that of a high-frequency VCO operating directly at sub-THz frequencies [65]. Additionally, the use of frequency multipliers can lead to more power-efficient designs. High-frequency VCOs generally require substantial power to maintain oscillations at sub-THz frequencies, whereas frequency multipliers consume less power by amplifying and selectively filtering other harmonics generated from lower-frequency signals.

Phase noise directly affects the EVM and, consequently, the achievable modulation index. For advanced modulation schemes such as 16 QAM or 64 QAM, low phase noise is critical for maintaining signal fidelity. The low phase noise of the frequency multiplier facilitates the use of higher modulation indices by providing a cleaner LO signal and supporting robust high-order modulation without degrading signal quality [64].

In summary, the preference for frequency multipliers over direct high-frequency VCOs in sub-THz TX designs stems from a strategic trade-off between phase noise reduction, power efficiency, and support for high-order modulation. Sub-THz systems are capable of achieving the requisite spectral purity and phase noise characteristics crucial for high-capacity wireless communication by producing high-frequency signals via multipliers.

2.3. Phase Shifter

I/Q generation in sub-THz transmitter designs, accurate phase rotation between in-phase (I) and quadrature (Q) signal paths is crucial for effective modulation and demodulation, particularly in systems using QAM. This phase difference, typically set at 90 degrees, can be achieved using three primary methods: couplers, poly-phase filters, and delay lines. Each approach offers unique trade-offs in terms of frequency range, precision, and integration complexity.

Couplers: Couplers are widely used in high-frequency systems to split an input signal into two signals with a 90-degree phase shift, producing the required I/Q components (Figure 11a). A commonly used coupler in I/Q generation is the quadrature hybrid coupler, which can achieve a stable 90-degree phase shift over a specified bandwidth [15,39,51,66,67]. Couplers are particularly effective for wideband systems, as they offer consistent phase accuracy across a broad frequency range. However, implementing couplers in sub-THz systems can be challenging due to size constraints and the need for low insertion loss. High-frequency couplers also require precise manufacturing, as even minor mismatches in the design can lead to phase errors that degrade the signal quality.

Delay lines: Delay lines provide a third approach to achieve the required phase difference between the I and Q paths by introducing a time delay to one of the signal paths [12,30]. This delay corresponds to a phase shift, which can be tuned to a certain degree by adjusting the line length or delay element configuration (Figure 11b). Delay lines are advantageous in that they can be adjusted dynamically, allowing phase tuning at different frequencies. However, implementing delay lines at sub-THz frequencies is challenging due to the precise length and low-loss requirements needed to maintain signal integrity. Furthermore, delay lines consume a large silicon area, which is a limiting factor in highly integrated systems.

Poly-phase filters: Other phase-shifting approaches include poly-phase filters (PPFs) [68] and switchable phase shifters (SPSs) [40]. PPFs are passive networks that use resistors and capacitors arranged in a configuration that splits the signal into two paths with a phase difference of 90 degrees. The advantage of poly-phase filters lies in their compact design and low power consumption, making them ideal for integrated circuit applications where space and power efficiency are priorities. SPSs comprise L-C networks and single-pole dual-throw (SPDT) switches that control the phase shift by bits (Figure 11c). However, PPFs typically have a limited bandwidth, as their phase accuracy depends on the matching of passive components, which can vary with frequency. This limitation makes these methods more suitable for applications with a fixed or narrowband frequency requirement.

Digital-controlled transmission line: Digital-controlled transmission lines provide precise phase manipulation essential for high-frequency communication systems. These phase shifters adjust phase shifts digitally, allowing for fine control with minimal analog tuning complexity. The fundamental principle relies on switched transmission lines or lumped-element networks, where digital control signals toggle the phase shift stages to achieve the desired phase states [69]. An example model of the digital-controlled phase shifters is shown in Figure 12 where the switches control the dielectric constant of the transmission line and thus the electrical length. Recent advances have focused on improving resolution, minimizing insertion loss, and achieving low power consumption. For instance, a compact D-band phase shifter achieves 0.1-degree phase resolution with an RMS phase error of 0.8 degrees [70]. Additionally, a low-loss digital-controlled transmission line in 45 nm RF-SOI technology implements a trombone-like structure, enabling calibration-free operation with a phase tuning range of 360 degrees and an insertion loss of 11.5 dB [71]. Though increasing digital control complexity, these innovations highlight state-of-the-art developments in digital phase control, enabling high-performance beam-forming and precise calibration for next-generation wireless communication systems.

In conclusion, each method for phase shifting and I/Q generation presents specific advantages and trade-offs, and the optimal choice depends on the application’s requirements for bandwidth, power efficiency, and integration level. By selecting the appropriate phase shifting technique, sub-THz TX designs can achieve the necessary phase accuracy for effective modulation and signal integrity across various frequency ranges.

2.4. Power Amplifier

Sub-THz power amplifiers (PAs) are crucial components in sub-THz transmitters, responsible for boosting signal power to levels necessary to overcome high free-space path loss and ensuring reliable communication links. The design of sub-THz PAs faces challenges related to efficiency, linearity, and integration, particularly at high frequencies. The trend of saturated power ($P_{sat}$) of sub-THz PA with various technologies is plotted in Figure 13, where a boundary of around 25 dBm is illustrated.

PA Topologies: Three common PA topologies used in mm-Wave designs are power-combining, Doherty, and output-phasing architectures.

Due to the limited output power capability of individual transistors, power-combining techniques are often employed to increase the total output power. Power combiners aggregate the output from multiple amplifier stages, using methods like current combining, Wilkinson combiners, or transformer-based combining. For instance, two-way to eight-way current-combining architecture demonstrated in sub-THz PA efficiently increases output power by leveraging symmetrical transformer-based or coupled-line coupling, which reduces sensitivity to parasitics and enhances power delivery [18,19,73,74,75,76]. The block diagrams of two commonly used power combination topologies based on current and voltage are shown in Figure 14 [77,78]. Additionally, multi-drive stacked-FET approaches utilize series-stacked transistors to support higher voltage swings, achieving higher output power while maintaining device reliability in advanced CMOS processes [79,80].

The Doherty power amplifier (DPA) is widely used for its ability to enhance power efficiency, particularly at back-off power levels. In a DPA, the main and auxiliary amplifiers operate in tandem to optimize efficiency across a range of output power levels. mm-Wave DPAs have been demonstrated in the E-band [81], but the transformer-based power-combining network becomes lossy at sub-THz frequencies. To mitigate this issue, DPAs that operate at higher frequencies apply a distributed network to enhance efficiency [74].

Outphasing PAs, or LINC (linear amplification using non-linear components), split the input signal into two phase-modulated components. These components are amplified separately and then recombined at the output, achieving high linearity and efficiency. The output-phasing technique is beneficial for high-order modulation schemes as it reduces the effects of amplitude distortion by relying on phase modulation, making it well suited for sub-THz transmitters that use complex modulations such as 16 QAM and 8PSK [20,21].

PA’s Non-linearity: Non-linearity in PAs is a critical concern at mm-Wave frequencies, as it directly affects signal quality and modulation performance. The primary non-linear effects include the 1 dB compression point $O{P}_{1\mathrm{dB}}$, amplitude-to-amplitude (AM/AM) distortion, and amplitude-to-phase (AM/PM) distortion [82].

The $O{P}_{1\mathrm{dB}}$ point indicates the output power level at which the amplifier gain drops by 1 dB due to saturation effects. It serves as a key metric for PA linearity, especially at high frequencies, where maintaining a high $O{P}_{1\mathrm{dB}}$ is challenging due to reduced transistor breakdown voltages. In recent D-band PA designs, gain expansion techniques have been employed to push $O{P}_{1\mathrm{dB}}$ to 20.4 dBm while maintaining high efficiency [83].

AM/AM distortion refers to the non-linear relationship between input amplitude and output amplitude, while AM/PM distortion involves undesired phase shifts as a function of input amplitude. Both types of distortion degrade the signal quality, affecting the EVM and reducing the achievable modulation indexes [84]. An illustration of these two non-linearities’ effect (together with AWGN) on the constellation is shown in Figure 15. In sub-THz PAs, stacked-FET architectures and adaptive biasing are often used to mitigate these non-linear effects by providing a more linear amplification profile and stabilizing the phase response across varying input levels [79].

Digital Calibration Techniques: To address the non-linearity and efficiency challenges in mm-Wave PAs, digital calibration techniques have become an integral part of modern PA design. Digital pre-distortion (DPD) is a widely used technique to linearize PAs by applying an inverse distortion function to the input signal, pre-compensating for the PA’s inherent non-linearity. DPD algorithms adaptively adjust the input signal to minimize AM/AM and AM/PM distortions, significantly improving linearity and enabling the use of high-order modulation schemes. Recent designs implement DPD using low-latency digital signal processing blocks integrated directly into the baseband circuitry of the transceiver, allowing real-time calibration [85].

Thermal effects can cause performance degradation in PAs, particularly in high-power and high-frequency applications. Thermal management techniques such as dynamic biasing and active cooling help manage the heat generated by the PA, stabilizing its performance. Dynamic biasing adjusts the bias current based on temperature feedback, optimizing efficiency while preventing device breakdown under high currents [85,86].

Dynamic biasing involves adjusting the operating point of the amplifier in real time to optimize performance based on signal power and environmental conditions [87]. This approach improves efficiency, particularly at lower output power levels, by reducing the bias current when full power is not needed. It is especially effective in stacked-FET PAs, where it helps manage voltage stress and ensures consistent performance across varying supply voltages and temperatures.

3. Receiver Design Considerations and Challenges

A summary of the performance of the sub-THz wireless communication link receiver, including data rate, modulation schemes, and receiver noise figure (NF) and conversion gain (CG), is presented in Figure 16. More than 60 data points have been analyzed, most of which are collected from works published over the past five years to highlight the latest advances in receiver technology. The highest data rate achieved is 128 Gbps in the D-band [88]. For modulation schemes, the most frequently reported is QAM, the highest level being 2048 QAM in the W-band with a passive mixer and an IF amplifier [58], 512 QAM achieved by an ultra-wideband dual receiver system in the D-band [47], and 32 QAM in the J-band [89]. NF is reported to range from 4.7 dB [90] to 15 dB in most receiver designs, with lower values achieved in systems optimized for specific narrowband applications. There are also receivers on multi-channel communication in sub-THz [45]. CG is observed in the range of 20 to 40 dB in most receiver architectures, with a noticeable reduction at higher frequencies. An observable trend is the increase in NF accompanied by a reduction in CG as the operating frequency increases above 200 GHz.

This section will introduce the system design considerations for sub-THz receivers, including the classical heterodyne and homodyne architectures, direct-digital demodulation, and other advanced receiver concepts.

3.1. Heterodyne and Homodyne Receiver

The block diagrams for the heterodyne, homodyne and direct-digital demodulation receivers are shown in Figure 17. The heterodyne architecture is popular, particularly for D-band applications [50,89,91,92,93,94]. Heterodyne receivers typically use a single mixer and a lower LO frequency, which helps minimize signal loss and noise gain. Furthermore, the use of a single mixer improves image frequency rejection, which is particularly important in the D-band, where image frequency interference is a significant concern. In the design of heterodyne receivers, the quality of the LO signal is a critical consideration, as phase noise and linearity significantly affect the overall performance of the receiver. These factors are important in both heterodyne and direct-conversion (zero-IF) architectures, but heterodyne systems require particular attention to the selection of IF and LO frequencies. In heterodyne systems, the selection of the IF is crucial. The choice of IF should be made with precision to ensure compatibility with both RF and LO frequencies, aiming to minimize any LO leakage into the RF and IF bands and to attain strong image rejection. For instance, in a D-band receiver, an IF range of 26–34 GHz is selected. This choice is based on its near compatibility with existing 5G FR2 chips, facilitating further down-conversion of the IF signal to baseband [23].

Due to the simplicity of the system, the homodyne architecture (direct-conversion) is attractive for achieving compact and low power in receiver design [95,96,97]. One of the key advantages of direct-conversion is the ability to convert the RF signal directly to baseband, avoiding the complexities associated with IF stages. This reduces the number of components required, simplifies the overall design, and improves the system’s power efficiency. Furthermore, direct-conversion minimizes the risk of image frequency interference, as it directly converts the RF signal to the baseband without additional mixing stages that could generate image frequencies. This is especially useful in systems where eliminating image frequency interference is crucial. Despite the advantages of direct-conversion, its application in high-frequency bands such as the D-band presents several challenges.

For example, the generation and distribution of I and Q components in the LO waveform are particularly demanding at high frequencies. Quadrature systems suffer from significantly higher phase noise—up to 12.6 dB higher compared to differential topologies—and consume more power. Additionally, I/Q mismatches at frequencies like 300 GHz require extremely precise timing alignment. For example, a phase mismatch of 10° translates to approximately 45 fs, making direct-conversion architectures especially difficult to implement effectively [91].

In D-band applications, dual down-conversion architectures are used to optimize receiver performance [24]. Compared to traditional single-stage down-conversion, dual down-conversion offers several advantages such as improved image ejection, bandwidth expansion, and design flexibility as it allows adjustments to be made based on application requirements and frequency characteristics.

3.2. LNA-First Versus Mixer-First Architectures

LNA-First RX: In receiver designs, the LNA-first architecture is a majority choice due to its excellent noise performance at the front-end of the receiver chain. The LNA, as the first component in the signal chain, plays a critical role in amplifying weak signals while maintaining a low NF, which is crucial to improving receiver sensitivity. With low NF and high gain of LNA, the following blocks in the RX path have less stringent NF requirements, as indicated by the Friis equation. LNA-first designs have demonstrated strong performance across various frequency bands.

However, while the LNA-first architecture excels in terms of noise performance, it also introduces inherent challenges, particularly in linearity and blocker suppression. The LNA amplifies all signals, including strong interfering signals (blockers), which diminishes the receiver’s dynamic range. Specifically, when subjected to strong signals, the LNA can enter a non-linear region, resulting in signal distortion. This leads to a decrease in the 1 dB compression point ($I{P}_{1dB}$), limiting the receiver’s ability to handle high-power signals (equivalently reducing the dynamic range) [98].

Mixer-First RX: The mixer-first architecture, in which the mixer is placed at the front-end of the receiver chain to enable direct down-conversion of the received signal, was initially proposed to address wideband impedance matching challenges in broadband receivers [99,100,101,102] (Figure 18). In the sub-THz range, it effectively mitigates the limitations of LNA-first designs, particularly in terms of linearity and blocker suppression. By shifting the signal to a lower IF early in the signal chain, this architecture allows the system to tolerate higher levels of blocker interference, thereby reducing the stringent linearity requirements typically imposed on high-frequency stages.

Furthermore, mixer-first architectures are particularly advantageous in high-frequency ranges, where maintaining low NF in the LNA becomes increasingly challenging. For 28 nm CMOS technology, the upper-frequency limit is typically around 280 GHz, and as the frequency approaches half of this value, the performance of active parts such as the LNA deteriorates significantly. In sub-THz, further compromising LNA performance. As a result, many designers opt for mixer-first architectures at these higher frequencies to avoid the inherent limitations of LNA performance [103].

3.3. Direct-Demodulation Receiver

Demodulation plays a critical role in communication systems by extracting transmitted data from received signals. Traditional demodulation techniques typically rely on high-speed ADCs to digitize received analog signals, followed by complex digital signal processing (DSP) to decode transmitted information. However, as the demand for higher data rates increases, particularly for advanced modulation schemes, these conventional ADC-based demodulation methods encounter significant challenges such as high power consumption and system complexity.

In response to these challenges, direct demodulation receivers have emerged as a viable solution, offering significant reductions in power consumption and complexity. Direct analog demodulation techniques bypass the need for high-speed ADCs by extracting amplitude and phase information directly from the received analog signal. For example, methods such as NR-STAR-MQAM use analog circuitry, such as root-square computation to determine the amplitude ($d_r$) and a phase detector to extract the phase (${\varphi }_r$), enabling a compact and low-complexity circuit design. These techniques significantly reduce power consumption—up to 31 times lower than digital methods—while maintaining comparable error performance [104]. By simplifying the receiver architecture, direct demodulation enables ultra-high-speed communication in sub-THz systems while addressing key limitations of traditional ADC-based designs.

Recent innovations have introduced RF-correlation-based direct-demodulation methods, which offer a more efficient and power-effective alternative for high-order modulations such as 8PSK. This method leverages multi-phase RF correlation to directly demodulate 8PSK symbols, eliminating the need for high-speed ADCs and complex digital processing. As described in the work of [105] (Figure 19), the proposed method involves correlating the received signal with multiple phase-shifted versions of the local oscillator (LO) signal. This process allows the system to detect the phase of the received signal without relying on amplitude-based decisions. Using simple sign-check comparators (e.g., BPSK decision logic), the system can detect symbols with significantly lower power consumption and reduced hardware complexity. The approach is particularly beneficial in high-frequency applications, such as mm-Wave communication, where traditional high-speed ADCs are not only power hungry but also require significant computational resources.

One key advantage of the RF-correlation-based method is its tolerance to phase errors. Even if the LO phase deviates from the ideal angle by up to 22.5°, the system can still accurately detect the symbols. This robustness to phase errors is a critical improvement over conventional methods, where small phase deviations can lead to substantial errors in symbol detection. Moreover, by reducing the dependence on high-speed ADCs, this method enhances the sensitivity and linearity of the receiver, making it suitable for ultra-high-frequency applications.

Recent research has shown that ADC-less demodulation techniques can also be applied to QAM modulation schemes, offering a promising solution for low-power and efficient receiver designs. Conventional QAM receivers typically require ADCs to sample the I and Q channels and subsequent digital processing. In the architecture shown in Figure 20, it reduces hardware complexity by amplitude and phase detector on the I and Q paths. However, this method is restricted to lower modulation schemes; with higher modulation indices, the amplitude and phase resolution exceed the capabilities of the detectors, and the BER will increase as a result.

Six-Port Direct-Demodulation Receiver: Another type of direct-demodulation receiver employs the six-port theory, or the so-called linear interferometric technique, offering distinct advantages over traditional non-linear mixer-based architectures like homodyne and superheterodyne systems [106,107,108]. These advantages include low power consumption, simple and compact configuration, wideband performance, and robustness to power-level variations. The six-port receiver fundamentally operates on the principle of linear interference, utilizing passive networks of hybrid couplers, power dividers, and power detectors to superimpose the received RF signal and LO under controlled phase conditions. The relative amplitude and phase shifts of the input signals at the various ports allow the direct extraction of components I and Q without requiring high-speed ADCs or complex digital signal processing (Figure 21).

The six-port receiver operates by linearly combining the RF input signal $S_{\mathrm{RF}}$ and the LO signal $S_{\mathrm{LO}}$ through a passive network. The equations for $S_{\mathrm{RF}}$ and $S_{\mathrm{LO}}$ are given as follows:

$$S_{\mathrm{RF}}\left(t\right)=A_{\mathrm{RF}}cos({\omega }_{\mathrm{RF}}t+{\varphi }_{\mathrm{RF}}),$$

(2)

$$S_{\mathrm{LO}}\left(t\right)=A_{\mathrm{LO}}cos\left({\omega }_{\mathrm{LO}}t\right).$$

(3)

The output voltages at the four power detectors are functions of the relative amplitudes and phases of these signals. The six-port junction linearly combines the input signals with specific phase shifts ${\theta }_k$ for each port k and the output power at the k-th port is proportional to the square of the voltage. The voltage and power at the k-th port can be derived as follows:

$$V_k\left(t\right)=S_{\mathrm{RF}}\left(t\right)+S_{\mathrm{LO}}(t+{\theta }_k),$$

(4)

$$P_k={\left|V_k\right|}^2={\displaystyle \frac{1}{2}}\left(A_{\mathrm{RF}}^2+A_{\mathrm{LO}}^2+2A_{\mathrm{RF}}{A}_{\mathrm{LO}}cos({\varphi }_{\mathrm{RF}}+{\theta }_k)\right).$$

(5)

Then, the I and Q components can be derived from differences in power at the ports:

$$I=P_3-P_5=2A_{\mathrm{RF}}{A}_{\mathrm{LO}}cos\left({\varphi }_{\mathrm{RF}}\right),$$

(6)

$$Q=P_4-P_6=2A_{\mathrm{RF}}{A}_{\mathrm{LO}}sin\left({\varphi }_{\mathrm{RF}}\right).$$

(7)

Recent innovations, such as joint multiband receivers, extend the six-port architecture’s utility by enabling concurrent operations at sub-THz frequencies [109], combining signals from both domains into a unified platform. These systems leverage the six-port receiver’s ability to simultaneously handle modulated signals across diverse frequency bands with minimal power consumption and straightforward hardware.

Kramers–Kronig Receiver: The Kramers–Kronig (KK) receiver is a direct-demodulation approach that simplifies coherent signal reception by eliminating the need for a high-frequency LO and mixers [110,111]. Instead, it takes advantage of the mathematical relationship between the amplitude and phase of an analytic signal. This technique, adapted from optical communication, is particularly effective for wireless systems operating at terahertz frequencies, offering significant advantages in simplicity and power efficiency.

An analytic signal $U\left(t\right)$ can be expressed as follows:

$$U\left(t\right)=U_0+U_s\left(t\right),$$

(8)

where $U_0$ is a strong carrier and $U_s\left(t\right)$ represents the data signal. The KK receiver ensures that $U\left(t\right)$ satisfies the minimum-phase condition:

$$|U_s\left(t\right)|<U_0,$$

(9)

such that the amplitude $\left|U\right(t\left)\right|$ contains sufficient information to reconstruct the phase $\varphi \left(t\right)$. The phase can be recovered using the Kramers–Kronig relation:

$$\varphi \left(t\right)={\displaystyle \frac{1}{\pi }}\mathcal{P}{\int }_{-\infty }^{\infty }{\displaystyle \frac{ln\left|U\right(\tau \left)\right|}{t-\tau }}d\tau ,$$

(10)

where $\mathcal{P}$ represents the Cauchy principal value.

The KK receiver employs an envelope detector, such as a Schottky barrier diode (SBD), to measure the amplitude $\left|U\right(t\left)\right|$. Advanced digital signal processing (DSP) techniques are then used to reconstruct the phase $\varphi \left(t\right)$ and demodulate the signal. An example is shown in Figure 22. This approach is compatible with advanced modulation schemes, such as 16 QAM, and enables high data rates, achieving up to 115 Gbps over significant distances (e.g., 110 m [111]). The simplified architecture of the KK receiver reduces both power consumption and hardware complexity, making it an ideal candidate for sub-THz wireless networks.

4. Transceiver System Design

4.1. System Considerations: Full Duplex with Single/Double Antenna, SIC, and LO Feedthrough

In sub-THz and mm-Wave transceiver (TRX) systems, one of the critical design choices is the implementation of full-duplex communication, which enables simultaneous transmission and reception of signals. Achieving efficient full-duplex operation in these high-frequency systems requires careful consideration of antenna configurations, self-interference mitigation, and LO feedthrough suppression. Two common configurations in full-duplex systems are single-antenna and double-antenna configurations, each presenting unique trade-offs in terms of isolation, integration, and system complexity.

Single-antenna full-duplex systems use a shared antenna for both TX and RX paths, typically employing an SPDT switch to alternate between transmit and receive operations [30,47]. This configuration offers the advantage of reduced size and integration complexity, making it particularly suitable for compact, highly integrated TRX systems at mm-Wave frequencies. However, single-antenna configurations face significant challenges related to self-interference. The high power of the transmitted signal can easily saturate the receiver front-end if adequate isolation is not maintained. To address this problem, self-interference cancellation (SIC) methods are utilized in the antenna, RF, and digital domains.

Recent works on sub-THz full-duplex phased-array transceivers have demonstrated SIC by employing a combination of analog and digital methods. Analog SIC is primarily achieved through directional couplers or cancellation loops that generate a replica of the transmitted signal and subtract it from the received signal. This approach effectively suppresses self-interference at the RF front-end, preventing receiver saturation. Digital SIC, on the other hand, further refines cancellation by leveraging adaptive filtering and signal processing algorithms to mitigate residual interference in the baseband domain.

For instance, a sub-THz full-duplex phased-array transceiver demonstrated effective self-interference cancellation using a combination of analog domain suppression and digital signal processing, achieving over 60 dB of cancellation without degrading the received signal quality. It typically involves the use of directional couplers or cancellation loops that subtract a copy of the transmitted signal from the received signal, while digital SIC leverages adaptive filters and signal processing algorithms to further suppress residual interference [112]. An example TRX work using SIC is shown in Figure 23, which achieves more than 20 dBc suppression of interference.

Another key issue in single-antenna systems is LO feedthrough, where the local oscillator signal leaks into the transmit or receive path, creating unwanted spurs and thus degrading the SNR. LO feedthrough can be exacerbated in direct-conversion transceivers, where the LO signal is mixed directly with the baseband signal. Techniques such as differential signal routing, improved LO buffering, and feedthrough compensation circuits have been implemented to mitigate these effects [112,113].

Double-antenna configurations utilize separate antennas for the TX and RX paths, inherently providing greater isolation between transmitted and received signals [114,115,116]. This approach reduces the impact of self-interference and allows for a higher dynamic range and better receiver sensitivity. The physical separation of TX and RX antennas can be particularly beneficial at sub-THz frequencies, where high-gain directional antennas are employed to counteract significant path loss.

Although double-antenna configuration is common, it also introduces challenges related to integration and size, particularly in compact, wafer-scale designs. Recent advances in TRX integration for full duplex TRx involve flip-chip bonding [56] and on-chip antenna arrays, allowing for high-density packaging without compromising performance. This approach was successfully demonstrated in a D-band radio-on-glass module, which integrated separate TX and RX with advanced packaging techniques, achieving stable performance across a wide frequency range [35].

In the most recent sub-THz transceiver works, TX and RX use a correlated LO signal to cancel phase noise and achieve a higher data rate or modulation scheme [15,50,117]. However, synchronized LO signal increased system complexity and does not apply to all wireless communication scenarios [118]. Instead, a recent work has reported 256 QAM with 8 Gbps in the D-band with uncorrelated TX and RX, which is shown in Figure 24 [35].

In summary, as the mm-Wave and sub-THz TRX systems aim for higher data rates and efficient full-duplex communication, significant progress has been made to address issues like self-interference and LO feedthrough, as analyzed in this section. Both single- and double-antenna configurations have their advantages, and the optimal choice depends on a combination of factors, including chip size, power budget, data rate constraints, isolation requirements, integration complexity, and overall system performance trade-offs to best suit the intended application.

4.2. Phased-Array System and Beam-Forming

Phased-array systems have become a key component in high-frequency transceiver designs, particularly in mm-Wave and sub-THz applications. The ability to electronically steer the beam without physically moving the antenna enables high-gain, directional transmission and reception, which is crucial for overcoming significant free-space path loss at these frequencies. Phased-arrays offer enhanced link performance through adaptive beam-forming, spatial multiplexing, and interference management, making them integral to modern wireless communication systems targeting 5G, 6G, and beyond.

The deployment of phased-array systems plays a crucial role in overcoming the inherent challenges associated with high path loss at sub-THz frequencies. The Friis transmission equation is widely used as follows:

$$P_r=P_t+G_t+G_r+20{log}_{10}\left({\displaystyle \frac{\lambda }{4\pi D}}\right),$$

(11)

where the last term represents the free-space path loss in logarithmic scale. As frequencies increase, free-space path loss follows a quadratic relationship with frequency, significantly impacting the signal strength over long distances. To mitigate this effect, phased-array antennas leverage beam-forming techniques to focus radiated power in specific directions, effectively increasing the link gain and compensating for path loss.

Beam-forming in phased-array systems is achieved by controlling the relative phase of the signals fed to each antenna element. By introducing a phase shift to the signals of individual elements, the array can direct the beam in a specific direction, effectively focusing the signal energy and enhancing the received signal strength. It not only increases EIRP but also mitigates the impact of multi-path fading in dense environments [119,120].

Phased-arrays typically employ three primary beam-forming approaches: analog, digital, and hybrid beam-forming. In analog beam-forming, phase shifters or delay elements adjust the phase of the RF signal before transmission or reception, allowing for a compact and power-efficient design. However, analog beam-forming has limited flexibility, as the beam can only be steered in one direction at a time. Digital beam-forming, on the other hand, processes signals in the digital domain, allowing for simultaneous multi-beam steering and improved interference mitigation [8,121,122]. Hybrid beam-forming applies both analog and digital approaches with more complex TRX architectures.

Phased-array TRX systems can be implemented using several architectural approaches, including RF beam-forming, IF beam-forming, and LO beam-forming. Each architecture offers unique trade-offs in terms of complexity, power efficiency, and beam-forming capabilities. The block diagrams of those architectures are shown in Figure 25.

RF Beam-forming: In RF beam-forming, the phase shift is applied directly to the RF signal using phase shifters or delay lines [30,34,47,92,123]. This approach minimizes the number of required data converters, reducing power consumption and hardware complexity. However, RF beam-forming typically lacks the flexibility needed for advanced multi-beam steering. A recent example of an RF beam-forming system is the D-Band 8×8-Element transmit–receive arrays that achieved an angular range of 60 degrees in the H plane and 38 degrees in the E plane [47]. The measured 2D and 3D patterns are shown in Figure 26 with a sidelobe level (SLL) of 12.5 dB.

IF Beam-forming: IF beam-forming applies phase shifts at the IF stage, allowing better control of the signal phase before it is upconverted to the RF frequency [49,93,124,125,126]. This method offers a compromise between the simplicity of RF beam-forming and the flexibility of digital beam-forming. IF beam-forming systems can achieve efficient multi-beam operation while reducing the phase shifter’s frequency dependency. A recent D-band TRX system used IF beam-forming to provide efficient phase control across a wide frequency range, supporting high-data-rate communication with robust beam steering capabilities.

LO Beam-forming: Similarly, if a phase change occurs on the LO path, then this method is called LO beam-forming [127,128]. LO beam-forming avoids additional loss on the signal paths and causes minimal gain variations. However, steering the phase at sub-THz frequencies requires a large amount of power, and phase error is another area of concern.

Other Challenges: Designing efficient phased-array TRX systems for mm-Wave and sub-THz applications involves several challenges, including phase noise, mutual coupling between antenna elements, and calibration of phase and amplitude mismatches.

At sub-THz frequencies, phase noise is a significant issue that can degrade the signal quality and limit the achievable data rate. High phase noise affects the beam’s coherence, reducing the effective gain and increasing the EVM. To mitigate phase noise, recent phased-array designs employ low-noise VCOs coupled with frequency multipliers, alongside digital calibration techniques that dynamically correct phase errors. For instance, a 60 GHz phased-array TRX system integrated adaptive phase noise compensation to enhance beam coherence and improve modulation fidelity in high-order QAM schemes.

Mutual coupling between closely spaced antenna elements can distort the beam pattern and reduce overall system performance. This coupling leads to unwanted changes in the impedance of each antenna element, which impacts the phase and amplitude response. Techniques such as decoupling networks, spatial filtering, and optimized array layouts have been employed to reduce mutual coupling.

The precise calibration of phase and amplitude mismatches is essential for maintaining beam-forming accuracy. Small errors in phase and amplitude can significantly distort the beam pattern, reducing the array’s effective gain and limiting the achievable data rate. Digital calibration methods, such as least mean squares (LMS) algorithms and adaptive equalization, are commonly used to dynamically adjust phase and amplitude errors.

In conclusion, phased-array systems are integral to modern mm-Wave and sub-THz TRX designs, offering enhanced beam-forming and spatial multiplexing capabilities. Using advanced beam-steering techniques and addressing associated design challenges, phased-arrays enable robust, high-performance communication links in next-generation networks.

4.3. MIMO Systems

Multiple-input multiple-output (MIMO) technology offers a robust solution to achieve high data rates in the sub-THz frequencies [31,121,129]. By utilizing multiple antennas at both the transmitter and receiver, MIMO enables the simultaneous transmission of multiple independent data streams over the same frequency band, effectively increasing spectral efficiency and enhancing link reliability through spatial diversity. An example photo of 4 × 4 MIMO measurement achieving a total data rate of 640 Gbps is shown in Figure 27. By employing spatial multiplexing, MIMO allows the simultaneous transmission of multiple independent data streams over the same frequency resources, significantly enhancing spectral efficiency. Furthermore, MIMO’s diversity gain plays a crucial role in ensuring link reliability and robustness in multipath environments. By receiving diverse copies of the transmitted signal, diversity gain mitigates the effects of deep fading and multipath interference, enhancing the system’s resilience to channel impairments.

In particular, Choi et al. explored the use of concurrent 20-beam transmission to significantly enhance spatial multiplexing efficiency while maintaining low power consumption [130]. A key contribution of this work is the implementation of a joint static/dynamic beam-forming system that combines static phased-array techniques with dynamic space–time modulation. Through the integration of these technologies, our work not only advances the state of the art in high-frequency MIMO systems but also lays the foundation for scalable, low-cost sub-THz communication infrastructures suitable for future wireless standards.

5. Conclusions

In this paper, we reviewed the latest advancements in sub-THz transceiver architectures, covering high-frequency signal generation, power amplifier design, direct modulation schemes, phased-array beam-forming, and MIMO technologies. While significant progress has been made, several challenges remain in realizing practical sub-THz communication systems.

One of the primary obstacles is the commercialization and standardization of sub-THz technology. Regulatory bodies such as the FCC have initiated efforts to allocate spectrum and define standardization frameworks. However, industry-wide collaboration is still required to ensure interoperability, efficient spectrum utilization, and cost-effective semiconductor solutions. Despite ongoing developments, the adoption of sub-THz communication is currently constrained by fabrication costs, energy efficiency limitations, and the complexity of high-frequency circuit integration. Future research must address these barriers by developing scalable semiconductor processes, advanced packaging techniques, and energy-efficient system architectures.

From a technical perspective, data rates and modulation schemes in sub-THz communication have been pushed close to the limits of thermal noise and transistor performance. As further improvements yield diminishing returns, research should shift toward enhancing power efficiency rather than solely increasing data rates. The development of direct-digital modulation transmitters and direct-digital receivers introduces new opportunities to optimize energy efficiency and system complexity. In terms of the building blocks, digital-controlled transmission lines are becoming key elements that offer improved phase tuning precision compared to conventional passive phase shifters. Additionally, PA efficiency at sub-THz frequencies remains a major bottleneck. Innovations in PA architectures, outphasing, and load modulation are required to enhance $P_{sat}$ and power-added efficiency (PAE). Transceiver design techniques, such as the self-interference cancellation, can further enhance the overall performance of the system. Lastly, the widespread adoption of phased-array and MIMO architectures at sub-THz frequencies will be essential to overcome propagation challenges and enable high-capacity wireless links. Techniques such as hybrid beam-forming could provide a practical trade-off between performance and power consumption. Together, these methods pave the path for the next-generation wireless communication system.

Addressing these commercial, regulatory, and technical challenges will be crucial for bridging the gap between academic research and real-world deployment. Although this work provides a comprehensive review of advanced sub-THz transceivers, it has certain limitations. This study primarily focuses on circuit- and system-level advancements in sub-THz TRX design. However, other crucial aspects, such as advanced semiconductor technologies that enable these circuits, have not been extensively covered. Furthermore, measurement techniques, including vector network analyzers (VNAs), oscilloscopes, and spectrum analyzers, play a fundamental role in the characterization and validation of sub-THz transceivers but are beyond the scope of this work. Furthermore, on-chip antennas and packaging technologies, which are integral to the practical implementation of sub-THz transceivers, warrant further exploration. Future work could provide a more detailed discussion of these topics to offer a more holistic perspective on the development and deployment of sub-THz wireless communication systems. As semiconductor technologies and circuit design techniques continue to advance, sub-THz wireless communication is expected to play a critical role in future sub-THz wireless networks, unlocking new possibilities for high-speed connectivity and ultra-low-latency applications.