Terahertz communication: detection and signal processing

The development of 6 G networks has promoted related research based on terahertz communication. As submillimeter radiation, signal transportation via terahertz waves has several superior properties, including non-ionizing and easy penetration of non-metallic materials. This paper provides an overview of different terahertz detectors based on various mechanisms. Additionally, the detailed fabrication process, structural design, and the improvement strategies are summarized. Following that, it is essential and necessary to prevent the practical signal from noise, and methods such as wavelet transform, UM-MIMO and decoding have been introduced. This paper highlights the detection process of the terahertz wave system and signal processing after the collection of signal data.


Introduction
Terahertz waves, which are a possible solution for developing fifth-generation networks (5 G) further, have been widely utilized for simulating, designing, and fabricating related THz detection systems.As a submillimeter -wave, THz wave ranges from 0.1−10 THz with a wavelength between 3 mm and 30 µm [1].Considering the establishment of the system, it can be divided into three parts: THz radiation source, THz detector, and THz signal processing [2].More specifically, THz sources can be divided into lasers such as quantum cascade lasers [3] and high-emissivity thermometer calibrators, including blackbody sources [4].Simultaneously, THz Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence.Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.detectors, as one of the critical components in terahertz detection systems, play a central role in capturing and converting terahertz radiation into measurable electric signals.To better absorb radiation and decrease the reflectivity on the surface of the THz detector, theories and structural designs based on different mechanisms have been proposed.Broadband THz detectors are of great importance and have huge potential for non-destructive real-time detection and imaging applications in the biomedical field [5].Finally, high-speed remote wireless communication is the primary goal of current research and commercialization.The actual electronic signal collected after the detection process is quite low, which is mixed with the environmental noise as well.To better distinguish and amplify the detected THz signal, researchers have applied several strategies such as wavelet transform [6].
In terms of the structural design and material selection for THz detectors, numerous novel materials have been integrated into the THz devices, which perform tunable and broadband THz detection characteristics.Compared with the traditional metal materials, Semiconductor materials such as GaAs [7]  and InAs [8] have some unique properties including on/off switches and changeable voltage control.With the development of nanotechnology, nanoscale materials have shown great potential for applications in various fields.Several lowdimensional carbon-based materials, including CNTs [9][10][11][12][13], which exhibit as one-dimensional nanomaterials, and GRs [14][15][16][17] which represents as two-dimensional nanomaterials, have been widely applied in THz detection.In addition, other nanomaterials such as TMDC [18], BP [19], and MXenes [20,21] have attracted considerable attention.GR offers unique optoelectronic characteristics, including exceptional electrical conductivity and excellent RF absorption with a near-zero band gap in few-layer dimensions.BP is a stable allotrope of phosphorus that can be used for the fabrication of semiconductor devices.MXene nanomaterials are a family of 2D transition metal carbides, nitrides, and carbonitrides, which can generally be expressed as M n+1 X n T x [22].This paper introduces the THz detectors based on their mechanisms and theories, followed by their structures and related applications.In every subtopic, the detectors, characteristics, and challenges faced by researchers were included based on the experimental results.To better solve the problems and improve the performance of devices, research originating from the past ten years has been introduced and discussed.In the next section, detailed strategies based on the signal processing are introduced.Table 1 lists all the abbreviations applied in this review.

Responsivity (R)
Responsivity denotes the capacity of a material or device to generate an electrical response in response to the incident terahertz wave.Due to the variation of the electrical signal, responsivity can be calculated based on the equation: R = Generated electrical signal Incident power . (1-1) In most cases, the electrical signal is output and collected via the form of current or voltage, resulting in the final unit of amperes per watt (A/W) or volts per watt (V/W).The higher responsivity the detector achieves, the more voltage generates, the more efficient conversion the detector performs eventually.

Noise-equivalent power (NEP)
Most THz detectors operate under non-biased conditions, where Nyquist-Johnson noise predominates as the primary source of noise.To assess the impact of noise on the final performance of the detector, another crucial parameter, known as NEP, has been introduced and compared.Following with the noise-equivalent voltage (V N ), NEP can be expressed as where k B refers to the Boltzmann constant, T represents the absolute temperature of material, and R refers to the resistance of the device.The smaller the Noise-equivalent voltage and the Noise-equivalent power are, the more signal the detector performs, the better signal-to-noise ratio the detector achieves.

Detectivity (D * )
The detectivity parameter considers both the sensitivity to the signal changes and the ability to discriminate against noise sources, offering a comprehensive assessment of a detector's performance.Defined as the reciprocal of the noise-equivalent voltage, it can be calculated as: where A stands for the effective area of the photothermoelectric detector.A high detectivity denotes a detector's capacity to detect feeble signals against a noisy background.

Response time (τ )
The response time indicates how quickly detectors respond to shifts in incident terahertz (THz) radiation.This measurement requires determining the duration for the detector's signal to reach a specific percentage (generally 90% or 95%) of its final steady-state value following exposure to a sudden change in the input THz signal.A faster response time allows the detector to capture rapid changes in THz signals, making it ideal for applications such as real-time monitoring or highspeed data communication.

THz detector: theory, structure and application
Among the field of detection, several types of detectors have been applied based on different mechanisms.They can generally be divided into three types: thermal detectors, rectification detectors, and resonant detectors [14].In the following chapter, different detectors, including the detailed structure design, figure of merit, drawbacks and so on are introduced and assessed.In addition, recent advanced developments based on the applied materials and fabrication techniques are summarized.By investigating possible applications ranging from real-time screening to medical imaging, this chapter highlights the significant role of terahertz detectors in unlocking the great potential of terahertz radiation for a wide range of practical applications.

Thermal detector
THz thermal detectors (thermometer) belong to one of the sensors used for detecting and measuring electromagnetic wave radiation, which converts the temperature change to other detectable signal based on the principle of thermal effects caused by the absorption of THz radiation [23].When THz radiation interacts with the absorbing material in a detector, it transfers energy to the lattice structure of the material.As a result, the increasing localized temperature of the material causes charge carrier transport inside the structure based on the Seebeck effect.As a result, electrical carrier transfer generates a photoinduced voltage, followed by output and collection via the readout circuit.Based on this, several types of THz detectors including Golay detector [24], bolometer [25], thermoelectric detector [26], and pyroelectric detector [4], have been proposed and fabricated.

Thermocouple.
As a basic and widely used thermal detector, the thermocouple acts as a transducer that directly converts thermal energy into electrical energy [27].Inside the structural design, two different metal wires combine into a terminal block to form a hot junction.The end was connected to the thermal reservoir.The other end of each wire, referred as the cold junction or reference junction the, are connected to a voltmeter to read the voltage difference.The produced thermoelectric voltage (V) can be calculated based on the equation: where the ∆T represents the temperature difference between different junctions, and α s represents the Seebeck coefficient of the materials.To better shield the wires from external influences and increase the mechanical strength of the thermocouple, a protective sheath made of stainless steel or ceramic materials has been used.As mentioned above, a thermocouple generates a voltage signal that is directly related to the temperature difference, making it a reliable and widely used temperature measurement device for various applications.This type of device has several advantages including a wide temperature detection range, fast response time, and simple fabrication process [27].However, research has met some challenges as well.Firstly, since the structural design of a thermocouple device is primitive, it is difficult to make precise measurements.Secondly, the generated emf curve is nonlinear with respect to temperature increase.The nonlinearity arises from the fact that the Seebeck coefficient, which quantifies the amount of voltage generated per unit temperature difference, is not constant across all temperatures for most materials.The Seebeck coefficient can vary with temperature, and different materials have different temperature dependencies.Calibration curves or electronic linearization were required to increase the accuracy of the final temperature readings.Finally, in accordance with the equation (2-1), the final voltage strongly depends on the temperature difference [28].In the actual testing process, the temperature of the reference junction changes owing to the limited distance between the wires.A temperature premeasurement or temperature compensation process is required, which increases the complexity of using thermocouples [29].
With the introduction of nanostructures and nanomaterials, the performance of thermocouples has improved synchronously [30].Nanoscale-based thermocouples, called nano-thermocouples, can achieve fast response time up to several picoseconds (ps) and can be combined with other MEMs.To exploit silicon materials with high Seebeck coefficients, Assumpcao et al [31] proposed a flexible metal-on-silicon nano-thermocouple, (shown in figure 1(a)) which enhances thermal sensitivity by a factor of 17-30 compared to traditional alloy-based thermocouples.

Pyroelectric detector.
A pyroelectric detector is another type of thermally sensitive detector that employs the pyroelectric effect to measure electromagnetic wave radiation [25].Here, the pyroelectric effect refers to a unique phenomenon that occurs in pyroelectric materials where a spontaneous polarization can be altered by temperature changes.When exposed to terahertz radiation, the temperature of the pyroelectric material increases, causing the internal polarization to change.This change in polarization results in the generation of an electric charge or a voltage across the material.The detailed pyroelectric current i can be calculated as: where A denotes the sensing area of the pyroelectric detector, ρ denotes the pyroelectric coefficient of the pyroelectric material, and dT/dt is the rate of change in temperature with time.
Under the same rate of temperature difference, a higher pyroelectric current requires a larger sensing area with a high pyroelectric coefficient.Compared with other thermal detectors, the detectors can also operate at room temperature [32], which simplifies the minimum requirement for the detection.Except that, they could be used for detecting a wide wavelength range of terahertz waves, making them suitable for various applications.Pyroelectric detectors are commonly used in applications such as thermal imaging, motion sensors, and spectroscopy.
The sensitivity and performance of a pyroelectric detector can be enhanced by optimizing the choice of pyroelectric material, electrode design, and readout circuitry.Different pyroelectric materials, such as TGS [33,34], LT [35][36][37], and PVDF [35,38], offer varying temperature ranges, pyroelectric coefficients, and Curie temperature (T C ) characteristics.Here, the Curie temperature refers to the crucial minimum temperature at which the internal polarization of the materials is reduced to zero.As one of the most widely studied materials, a single TGS crystal has a high pyroelectric coefficient and a high figure of merit.However, with further research, the weakness of this material becomes obvious: the low chemical stability of the TGS shortens the lifetime of the final detector.In addition, the Curie temperature of the TGS is relatively low, which limits the range of working temperatures.Several modification methods related to TGS materials include doping with other materials [30].Compared with the TGS material, LT materials have higher stability with slightly lower pyroelectric coefficients and lower T C .Sun et al have successfully designed and fabricated a pyroelectric detector based on LT material (Shown in figure 2) [37].By coupling with the quartz substrate via the UV adhesive method, the pyroelectric detector performs expectable thermal and mechanical characteristics.In a latest study, Arose et al proposed a wavelength-selective pyroelectric THz detector based on a thin LT wafer [35].After comparing the results from the simulation with the real measurement, they confirmed that the final responsivity is on the order of 1 V W −1 with 10 5 Jones signal-to-noise ratio within the frequency range of 0.3-1.5 THz.Polymer-based pyroelectric materials, particularly PVDF materials, have become an alternative solution for designing pyroelectric detectors.Although they have slightly poor pyroelectric characteristics, their better compatibility and excellent mechanical properties have broadened their structural design, which makes it possible to design large-area pyroelectric detectors or pyroelectric detector arrays.Several thin-film pyroelectric sensors based on poly (vinylidene fluoride-co-trifluoroethylene) (PVDF-TrFE) copolymers have been proposed and fabricated [38].In summary, owing to their fast response time, wide spectral range, and ease of operation, pyroelectric detectors have been widely used in various applications that require the detection and measurement of terahertz radiation.

Bolometer.
A bolometer is a specialized instrument designed to measure the power of the incident electromagnetic radiation.It operates by converting incident electromagnetic radiation into heat, causing a change in the temperature of the bolometer.This temperature change is then translated into a corresponding electrical resistance change, which can be readily measured [39].The bolometer's history dates to the late 19th century when it was invented by American astronomer Samuel Pierpont Langley [25].Since then, bolometers have been a fundamental tool for exploring various scientific frontiers, from measuring minute changes in cosmic microwave background radiation in astrophysics to probing the quantum nature of light in quantum optics.
The functioning of a bolometer is underpinned by the principal theory of black body radiation, coupled with the distinct attributes inherent materials-specifically, their temperature coefficient of resistance (TCR).The TCR quantifies the variation in a material's electrical resistance contingent upon a unit change in temperature.In the context of an appropriate bolometric material, a high TCR is imperative to secure a substantial alteration in the electrical signal in response to a specified incident power [40].
Figure 3 shows a schematic representation of the fundamental structure of a typical bolometer, which primarily consists of two integral components: a radiation absorber and a thermometer.Both elements are interconnected via a thermal link that extends to a thermal reservoir.In certain instances, a single component can simultaneously fulfil the roles of both the radiation absorber and thermometer.
When a radiation absorber is subjected to incident electromagnetic waves, a thermogenic response is elicited, causing an increase in the temperature of the absorber material.Subsequently, the thermometer near the absorber quantifies the modification in resistance within the absorber material, which is instigated by the thermal elevation.This resistance alteration is subsequently transmuted into a more quantifiable parameter, such as the voltage, which is utilized to assess the power of the incident radiation.
During this thermodynamic process, two key factors come into play: a discernible delay in the manifestation of the temperature variation and a finite conductance that facilitates the dissipation of the generated heat towards the thermal reservoir.Consequently, the dual roles of the absorber and thermal link can be encapsulated by two key thermal elements: a thermal capacitor (C th ), which represents the absorber, and a thermal conductance (G th ), which symbolizes the thermal link.The intricate interplay between these elements governs the overall performance and efficiency of the bolometer in detecting and quantifying incident radiation.According to the simplified model, a bolometer can be effectively represented by a thermal RC network, as illustrated in figure 1(b), which is analogous to an electrical RC network.The absorbed incident power, denoted as P, can be mathematically described by the heat transfer equation: Here, the temperature increase induced by illumination is denoted as ∆T (t), where the thermal response time can be determined based on the relationship between the heat capacity (C th ) and thermal conductance (G th ).This relationship implies that the thermal response time is equivalent to the ratio of C th to G th .It is important to note that if the power absorption from the ambient temperature is considered, the term P in the equation will also encompass the ambient term as well.Hence, equation (2-3) provides valuable insights into the thermal dynamics and response characteristics of the system, aiding the analysis and optimization of its performance [41].
In contemporary research, there is a substantial impetus to augment the performance of bolometers to satisfy the requirements of diverse high-sensitivity applications.A promising trajectory in this regard is the development of superconducting bolometers.Superconducting materials, subjected to cooling near absolute zero, demonstrate zero electrical resistance.Marginal fluctuations in temperature, instigated by incident radiation, can subsequently propel the superconductor beyond its critical temperature, eliciting significant resistance alterations.This characteristic culminates in a bolometer that exhibits superior sensitivity compared to those fabricated from conventional materials [42,43].As the original edition of hot electron bolometer (HEB) device, one attractive sandwich-like structural design-superconductor-insulator-superconductor (SIS) has been widely used in superconductor-based detectors.The insulator acts as a barrier which prevents the mutual flow of hole-electron pairs, resulting in the quantum tunneling phenomenon that happened between the two superconducting layers.Although this structure has shown good performance in terms of terahertz detection, it is hard to achieve more than 1.4 THz operating frequency [44].In contrast, HEB can break the frequency restriction and perform superior sensitivity with low noise and wide bandwidth.It operates based on the principle that the resistance of a superconducting material increases under high-frequency THz radiation.Chen et al [45] have successfully designed and fabricated HEB mixers including a planar antenna and a niobium nitride (NbN) connecting bridge between the antenna terminals and the silicon substrate.The results show that the device can work normally with low noise under 1.6 THz and 2.5 THz radiation.In addition, Su et al [46] have symmetrically examined a NbN-based HEB device via a THz biasing method.Although a relatively low operating frequency-0.65THz has been applied, the results show that the device can achieve up to 10 5 A W −1 current responsivity with enhanced NEP value of 2.7 pW Hz −1/2 .Nonetheless, the use of superconducting materials introduces new challenges, such as the need for cryogenic cooling.To address this, research is being conducted on high-temperature superconductors, which can operate at more accessible temperatures [47,48].
Microbolometers are another important class of bolometers.In these devices, the sensing element is miniaturized to the micrometer scale, which reduces the thermal mass and hence increases the speed and sensitivity [47][48][49][50].Microbolometers have been widely used in thermal imaging applications, such as night-vision and firefighting cameras, where they convert invisible infrared light into visible images [51].Despite their extensive application, bolometers have several constraints.Firstly, the response time of a bolometer, which is determined by its thermal relaxation time, can exhibit significant sluggishness.This restricts the capacity of the bolometer to assess the swiftly oscillating signals.Secondly, the prerequisite for continuous temperature calibration during bolometer operation can present challenges in specific field applications [52].Tu et al [53] have designed and fabrication a microbolometer based on Nb 5 N 6 material, which achieved up to 61.5 V W −1 maximum voltage responsivity with 8.5×10 −11 W Hz −1/2 NEP value.Aji et al [54] combined the microbolometer with another dipole antenna, which performs up to 530 V W −1 responsivity with 42 pW Hz −1/2 NEP value within the regime of 0-1 THz.
In conclusion, bolometers are crucial tools for diverse scientific and industrial applications.The continued advancement of bolometer technology is important for enhancing our capabilities in areas such as astrophysics, quantum optics, thermal imaging, and more.Their improvement and innovation are the subjects of ongoing research that present promising exciting developments in the future.

Photothermoelectric detector.
As an advanced type of evolutionary sensing device, PTE detectors combine the photothermalelectric effect with the bolometric sensing mechanism, which acts a significant role in the detection and imaging of THz radiation.Under the THz radiation, the surface of the PTE detector absorbs the electromagnetic radiation, which induces a localized heating effect.Owing to the application of thermoelectric elements, the localized temperature difference causes the charge carrier transport inside the structure based on the Seebeck effect.As a result, the electrical carrier transfer generates a photoinduced voltage, followed by output and collection via the readout circuit.The photoinduced voltage, denoted as U, can be determined based on the equation: (2-4) Here, S represents the Seebeck coefficient of the material and ∆T represents the temperature difference.
To achieve high sensitivity and high efficiency characteristics, three essential components with different working functions have been considered: the absorption layer, the thermoelectric element, and the readout circuit.Firstly, the absorption layer (also called the absorber), is applied to absorb the THz wave radiation.Typically, a patterned metamaterial [26,55] or a semiconductor layer [56,57] is used as the absorber to enhance the absorption of THz waves and maximize the interaction between the THz radiation and the detector.Secondly, as mentioned above, the thermoelectric element transforms the absorbed THz radiation into a detectable electrical signal with the assistance of an absorption layer.Typically, it comprises a thermally sensitive material, especially a nanostructured material, which experiences temperature variations upon exposure to the THz radiation.According to recent research findings, 2D nanomaterials including GR [15-17, 54, 55], BP [58], MXene [21,59,60] and TMDC [59] exhibit a good photoelectric conversion performance.The atoms within the 2D materials establish layers characterized by strong covalent or ionic bonding, whereas the individual atomic layers are interconnected through relatively weak vdWs interactions in the perpendicular direction.By adjusting the stacked layers with suitable crystal structures, the electrical band can be effectively tuned, which helps to realize broadband THz detection [61].Finally, the readout circuit helps realize signal amplification, noise suppression and signal optimization, which ensures accurate and reliable measurement of the THz radiation and facilitates further analysis or visualization study.
Several recent studies have contributed to the development and improvement of THz detectors.For instance, since thermocouples are widely used temperature sensitive devices, a novel PTE detector design that couples the metamaterial absorber with a thin-film thermocouple has been proposed [26].Owing to the high Hall coefficient and acceptable photoelectric response within the THz region, bismuth material (black) was used with gold material (yellow) as the thermocouple layer (shown in figure 4(a)).The final simulation results demonstrate that the PTE detector exhibits a 35 mV W −1 with 6.6 × 10 −6 W Hz −1/2 at a resonance frequency of 1 THz.
As another improvement approach, the introduction of heterojunctions can also improve the photothermoelectric performance of detectors.In contrast to the traditional junction of a single material, a heterojunction is established between different materials or between the same materials with different crystal structures.According to equation (2)(3)(4), the concept of heterojunction can further improve the Seebeck coefficient difference and causes higher photovoltage.shows a heterojunction photothermoelectric detector comprising graphene and Ta 2 NiSe 5 [62].However, the mismatch between the rigid substrate and PTE components limits and increases the difficulty during the fabrication process.Several flexible polymer-based thin-film substrates have become available for application in flexible photothermoelectric devices.
3.1.5.Golay cell.Golay cells are pneumatic detectors used at ambient temperature within a broad range of wavelengths from approximately 20-1000 µm [63].It consists of a small metallic cylinder filled with inert gas.As shown in figure 5 [64], there is a thin film window on one side of the cylinder, and a flexible diaphragm on the other side.The terahertz radiation is allowed to fall onto the outer surface of the thin film window and absorbed.When the radiation is absorbed, it heats the internal gas, causing an increase regarding the internal gas pressure.As a result, displacement occurs on the surface of the flexible membrane, which depends on the strength of the THz waves incident on the front window.One light produced from an LED device has been focused by the focusing optics and illuminated onto the outer surface of the diaphragm.After displacement happens, the light reflected from the flexible mirror back to the focusing optics varies as well, which eventually changes the intensity of the light absorbed by the photoelectric cell.Since the electrical signal converted via the photoelectric cell is quite low, a preamplifier circuit will be applied to output the final signal.
Considering the adverse impact of platform vibration on the performance of the Golay cell, the Golay unit has mostly been installed on a vibration-isolated base.This base provides facilities for adjusting the position of Golay cells to receive incident radiation.The basic requirement of a vibration-free platform, coupled with the huge and complex structure of the Golay unit and its high cost, sometimes makes practical onsite measurement inconvenient.Other drawbacks include its fragility and slow response with a typical rise time of 25 ms [24].Meanwhile, it is obvious that this detector can operate at room temperature, and no cooling condition is required when repeatedly detecting THz radiation.These advantages provide relatively reliable measurements and promote their application in the astronomical field [65].

Rectification detector
Terahertz rectification detectors operate based on rectification, which is a nonlinear process that transforms alternating THz radiation into a DC electrical output.During this process, rapidly oscillating THz waves can be rectified to a relatively more measurable form.Commonly used detector devices involve the Schottky diodes, which exploit the nonlinear behavior of metal-semiconductor junctions, as well as the field effect transistor, which applies an electric field to control the current flow between the internal structures of semiconductors.Other types of rectification detectors including quantum cascade detectors and nonlinear optical crystals have been applied for terahertz detection.In the following paragraphs, different terahertz rectification detectors are introduced and reviewed in detail.

Schottky diode.
A Schottky diode, also referred to as a hot-carrier diode or Schottky barrier diode, is a type of semiconductor diode that has a metal-semiconductor junction rather than a semiconductor-semiconductor junction, as in conventional diodes.It was named after the German physicist Walter H. Schottky.This metal-semiconductor junction creates a barrier (the Schottky barrier) for electrons at the interface, which has unique properties compared to the standard p-n junction of a typical diode [39].
Schottky diodes are characterized by their fast-switching times, low forward voltage drops (usually 0.25-0.4volts for a metal-silicon junction), and low junction capacitance.These properties make Schottky diodes particularly useful in highfrequency applications, such as RF circuits and power rectification.The fast-switching characteristics of SBDs can be represented by their high cut-off frequency, which is given by: In equation (2)(3)(4)(5), a simplified SBD model is represented, which predominantly consists of a junction capacitance (C 0 ), measured at zero bias, and a total series resistance (R s ).Although this model provides a basic understanding of the behavior of SBDs, more comprehensive device models require additional components to accurately portray the contributions originating from various sections of the device.The cutoff frequency ( f T ), which is essentially the reciprocal of the RC time constant, typically exhibits high values in SBDs, primarily attributed to the diminutive RS, consequently leading to reduced charging/discharging durations.Noteworthy advancements in SBD technology have enabled the attainment of cutoff frequencies exceeding 10 THz, as demonstrated with a GaAs-based SBD, while Si-based SBDs have achieved record f T values of approximately 4 THz.One detailed example for the Si-based SBD structure has been shown in figure 6.These frequencies are an order of magnitude higher than the cutoff frequencies observed in transistor technologies based on analogous material systems, underscoring the superior highfrequency performance of the SBDs.
In the context of THz radiation detection, Schottky diodes play a critical role due to their high-frequency characteristics.The low junction capacitance and fast switching time allow these diodes to function effectively, even at extremely high frequencies in the THz range.When THz radiation impinges upon the Schottky diode, it induces an AC current in the diode owing to the interaction between the incident radiation and charge carriers in the diode's depletion region of the diode.This AC current can then be rectified into DC current, which effectively converts the THz radiation into an electrical signal that can be measured.This is the fundamental principle by which Schottky diodes are used for THz detection.
Moreover, the sensitivity of Schottky diodes to THz radiation can be enhanced by applying a bias voltage, which adjusts the Schottky barrier height and thus controls the number of charge carriers available for conduction.This allows for the optimization of the diode response to incident THz radiation, facilitating more accurate and sensitive detection.
Recent developments, such as zero-bias Schottky diodebased THz detectors, have been designed to operate effectively even without the application of a bias voltage, making them more practical for certain applications.Ongoing research in the design and fabrication of Schottky diodes continues to improve their performance and expand their applications in the field of THz detection.
The field of THz detection has seen significant advancements in recent years, particularly in the context of Schottky diode technologies.One notable development is the fabrication of solution processed ZnO Schottky diodes.These devices can function within microwave and millimeter-wave frequency ranges, featuring an intrinsic cut-off frequency exceeding 100 GHz.Their design includes two asymmetric metal electrodes separated by a gap of approximately 15 nm, with a layer of ZnO or aluminum doped ZnO deposited from solution.When these diodes are integrated with additional passive elements, they can deliver output voltages of 600 mV and 260 mV at frequencies of 2.45 GHz and 10 GHz, respectively [67].
Recent advancements have been realized in the field of THz detection with the advent of zero-bias, room-temperature operable Schottky diode-based THz detectors.These state-of-theart detectors have demonstrated operational capabilities of up to an upper limit of 5.56 THz.Their performance characterization has been conducted utilizing a dual-method approach: a table-top system for frequencies up to 1.2 THz and a FEL facility for isolated frequencies ranging from 1.9 to 5.56 THz.The employment of two distinct measurement techniques enabled differentiation between the sub-nanosecond and millisecondscale responsivities, with a recorded NEP of 10 pW Hz −1/2 .Schottky diodes, in conjunction with HEMTs, represent a class of rapid detectors capable of achieving sub-nanosecond temporal resolutions, with their speed largely constrained by their RC time constants, which can exceed 1 THz.The compact size, room-temperature operability, and cost-effectiveness of these detectors enhance their practicality and ease of handling, rendering them suitable for a wide range of applications, notably diagnostics at THz-generating particle accelerator facilities [68].
In conclusion, the development, and recent advances in Schottky diodes for THz detection underscores the potential of these devices for high-frequency applications.Ongoing research and technological innovations in this domain continue to push the boundaries, suggesting a promising future for THz detection and its myriad applications.

Field effect transistor.
THz FETs play an essential role in the field of terahertz detection, offering a promising direction for high-frequency signal amplification and modulation.Similar with its counterparts in basic electronics, the THz FET is designed with three terminals, the source, the drain, and the gate [69].The rectification process that happens during the FET works involves the transistor's ability to selectively allow or blow the current flow based on the voltage applied on the gate terminal.When a positive voltage applied onto the gate terminal, an electric field has been induced which modulates the charge carriers within the channel, either enhancing or depleting their concentration.Dyakonov and Shur [70] have firstly demonstrated the possibility of applying the FETs for terahertz radiation detection, providing another pioneering direction for terahertz detectors.Based on that, further research based on THz FET devices have emerged from more innovative structural design including HEMTs (MPDFETs) based on III-V semiconductor materials.HEMTs sometimes called MODFETs, represent an improved generation of semiconductor devices.T. Mimura with his team from Fujitsu Lab first proposed and designed the first generation of HEMT devices in 1980 [71], following with the theory proposed by Ray Dingle et al based on the doping process within a III-V heterostructure in 1978 [72].Compared with traditional MOSFETs, HEMTs have obvious advantages in rectification owing to their relatively high electron mobility, making them ideal for applications requiring fast switching and high-frequency operation.The central structure that distinguishes HEMTs is their unique heterostructure, which usually consists of a thin layer of semiconductor materials, such as GaAs [73] and InP [74], between two different semiconductor layers.This sandwichlike structural arrangement formed a 2DEG at the interface between the layers.In this case, electrons in the 2DEGs have higher mobility than electrons in bulk materials, which enables faster internal charge transport.To better investigate the impact of different parameters including thickness, length, and spacing, plenty of research has been conducted in the past.Khan et al [75] examined the performance of an AlGaN/GaN HEMT device by varying the thickness of the AlGaN barrier layer.According to their results, increasing the thickness can effectively increase the drain current.Nevertheless, the short-channel effect significantly decreases the internal electrostatic control of the gate terminal when the thickness reaches a specific value, which confirms the trade-off between electrostatic control and drain current.In addition, one of the possible strategies to improve the performance of the final HEMT device is to increase the spacing distance between the gate and the drain channels.Zhi et al [76] proposed a series of experiments based on T-gate InAlAs/ InGaAs InP-based HEMT devices, which confirmed that it is possible to effectively diminish the electron avalanche multiplication, thereby alleviating the intensity of the electric field by increasing the distance of the spacing.Kim et al [77] systematically investigated and compared the influence of channel layer thickness on InP-based HEMT devices.Based on their research, several analytical relationships between detailed parameters were derived and proposed.Owing to the space limitation of the review article, specific formulas and detailed conclusions are not introduced.Overall, two distinct types of HEMTs, depletion-mode (D-mode) and enhancement-mode (E-mode), have been developed and applied for THz communications.A detailed comparison of these two HEMT structures is presented in the following table 2. Several research [77][78][79][80] have confirmed that D-mode HEMTs are more suitable to utilize for THz related applications than E-mode HEMTs due to their actual experimental RF performance results.In addition, several nanomaterials, especially 2D materials like GR, MXene and BP, exhibit superior electron mobility and high carrier velocities for the efficient detection performance at terahertz frequencies.The application of these materials can improve the performance of the FET devices further.Vicarelli et al [81] fabricated an efficient top-gated FET device based on the graphene material, which exhibited a maximum 150 mV•W −1 responsivity under 0.3 THz radiation.Simultaneously, physical imaging was successfully performed without breaking the closed package.Furthermore, Bandurin et al [82] successfully fabricated a FET device based on dual-gated graphene encapsulated in hBN structural design, which can perform up to 30 V W −1 responsivity and minimum 0.6 nW Hz −1/2 NEP value under 0.13 THz sub-THz radiation.Additionally, Delgado-Notario et al [83] designed and fabricated a few-layers graphene-based THz FET devices with two asymmetrically top metallic terminal structural design.According to the examination under illumination of 0.3 THz radiation, the device achieves the maximum value of 0.216 A W −1 current responsivity with the minimum value of 0.81 pW Hz −1/2 NEP under 4.5 K temperature.Significantly, a substantial enhancement has been realized when appropriate biasing between the top and the back gate terminals.Owing to the various combinations of the metallic and non-metallic elements, the MXene family continues to expand and explore.However, related research has confirmed that MXene exhibits better absorption within the mid-infrared band.Few experiments have been designed and fabricated using MXene FETs.According to the published research, the bandgap energy value of BP crystals will increase when decreasing the number of layers inside structure, which have been examined as 0.3 eV for bulk BP crystal [84] and 2.0 eV for monolayer BP flake [85].Viti et al [86] fabricated a top-gated FET device based on thin flake black phosphorus material, which performed up to 8 V W −1 responsivity with maximum 500 signal-to-noise ratio within 0.265-0.375THz range.Most of the FET devices have exhibit a broad-band or non-resonant response towards THz radiation, but notably, they could also achieve resonant detection within the relatively high THz frequency region.To be precise, in terms of resonant photodetection in THz FET detectors, when the quality factor (Q) is substantially larger than 1, the resonant THz detection can occur in plasmonic FETs regarding to the high THz region.The quality factor can be calculated based on the following equation: where ω represents the expected resonant frequency and τ represents the momentum relaxation time of charge carriers in the system.Knap et al [87] have proved the possible resonant photoresponse happened in the GaAs/AlGaAs FET device under 0.6 THz illumination, which coheres with the plasma wave detection theory.The increase of the carrier density in the transistor channel transfers the resonance position to a higher gate voltage, which coheres with the plasma wave detection theory.Following with that, Bandurin et al [88] designed and fabricated an antenna-mediated coupling FET device based on graphene encapsulated the hexagonal boron nitride (hBN) crystals material, which performs up to 240 V W −1 and 0.2 pW Hz −1/2 NEP under 0.13 THz radiation.Meanwhile, the fabricated FET device could also act as a resonant THz photodetector due the plasmon resonance in the graphene channel, providing a potential plasmonic application in strong magnetic field.In the latest research, Caridad et al [89] have successfully fabricated single-layer graphene FETs and observed the plasmonic resonant THz detection for the first time.The fabricated short-channel FET detectors could achieve larger than 1 quality factor (Q ≫ 1), even under room temperature.In summary, FET devices have become one of the most advantageous and widely applied detectors in the field of THz detection over the past two decades.Their potential applications in THz detection continue to be a subject of ongoing exploration.

Quantum cascade detectors.
As derived from QCLs, QCDs have similar design structures and functions.QCDs are a type of intersubband photovoltaic detector that has been widely applied for photodetection within the mid-IR and near-IR region [90].One core process, called LO-phonon scattering, occurs during the operation.Nevertheless, since the photon energy is lower than the LO-phonon energy under THz radiation, it is difficult for the QCLs devices to work with.In this case, strategies including optimizing the interface roughness and material selection have been proposed and examined.Scalari et al [91] established a QCL-like structure based on the Al 0.15 Ga 0.85 As/GaAs material.Based on their design concept, the detector operated under 3.6 THz, and a semi-insulating substrate was utilized for low loss guiding related to surface plasmons.When a strong magnetic field was applied perpendicularly to the plane of the detector, significant modulations of the photocurrents were observed.However, the body effects on the final energy shift remain unknown and require further investigation.Giorgetta et al [92] successfully designed and fabricated a QCD device based on InGaAs-AlAsSb material that exhibits up to 10 11 Jones detectivity under around 3.5 THz radiation at 108 K.

Resonant detector
IR waves are a type of EMR whose wavelength is longer than visible light and shorter than that of radio waves.As a type of uncooled IR detectors, resonant IR detectors have been widely used in security monitoring [93], mess sensing [94], and human signal detection [95] due to their operation at room temperature, lack of cooling requirements, small size, low power consumption, long lifespan, and relatively low cost.
With the advancement of silicon micromechanical systems and surface micromachining technology, research on materials, structures, and circuit systems for resonant IR detectors has led to breakthroughs in terms of device volume, integration, sensitivity, and characterization.Longer IR wavelengths between 30 and 100 µm are sometimes included as part of the terahertz radiation range.Therefore, it is possible for researchers to apply IR detectors for detecting THz waves.Generally, resonant IR detectors are categorized into two types: electrothermal and electrostatic excitations.The former produces mechanical displacement through thermal expansion and contraction of materials, and thermal energy is induced by an alternating current.The latter utilizes the changes in the Coulomb force between two charged electrodes or charges in piezoelectric materials to produce motion.In addition to the driving electrodes providing the excitation, detectors also include vibrating elements and sensitive elements that should be sensitive to IR or heat.
Micro-cantilevers and thin-film resonators are vibrating elements in resonant IR detectors.After the detector absorbs the IR radiation and converts it into thermal energy, the vibrating element changes stress or produces volume expansion owing to the temperature change, resulting in a certain shift in the resonant frequency.Additionally, changes in the amplitude or phase of the output-sensitive element can be used to reflect the measured changes.In practical applications, the frequency signal is less sensitive to external electromagnetic interference and has a strong anti-interference ability.This type of signal is easier to convert into a digital signal, which can ensure the stable operation of the detector [96,97].
Several efforts have been made to enhance the performance of sensitive and vibrating elements.Initially, these elements are quite simple and can be easily manufactured, which mainly include silicon and silicon dioxide materials.The challenge is that their low extinction coefficients, low absorption rates, low-TCR, and low-quality factors impact the conversion of super thermal energy from IR radiation and signal collection.This can affect the sensitivity and resolution of the detector [98,99].By separating the sensitive and vibrating elements into two parts, signal conversion can be effectively enhanced.In recent years, AlN [100], LiNbO3 [101], metal [99], quartz [102], and SiN [103] have received widespread attention as IR absorption and vibrating elements because of their high IR  absorption rates and high sensitivities in terms of the temperature change.In addition to common IR-sensitive materials, Qian et al [104] added a layer of ultrathin graphene on AlN, as shown in figure 7, which not only shows a resonant frequency of 307 MHz and quality factor of 450% of the original but also significantly improves the IR absorption rate based on the subwavelength thickness of the transparent graphene film, increasing it by ten times at 5 µm and possibly a hundred times at 3.4 µm.
The operational mechanics of sensory films provide an innovative route to augment the efficiency of detectors.The pivotal action of this film is delineated as a 'twisting motion,' characterized by the bending and contortion of the thin film torsion bar consequent to the absorption of IR radiation, as per [105].Upon IR radiation, the incident IR wave is absorbed, leading to a design-induced upward curvature of the resonator.This is a result of the central layered structure undergoing a more significant thermal expansion, as illustrated in figure 8.This bend sets the torsion bar at an angle to the straight line connecting the static ends, thereby reinforcing the rigidity of the torsional spring.The resultant effect of this chain of events increases the resonant frequency of the detector.Thus, the core principle of IR sensing lies in the precise measurement of this frequency surge, further solidifying the crucial role of the working mode of the sensory film in improving the detector performance.As illustrated in figure 8, the selectivity of the detector remains a paramount criterion.To address this issue, spectrally selective IR detectors have been engineered by exploiting subwavelength plasmonic gratings.This technology capitalizes on the occurrence of SPR within sub-wavelength metallic gratings.This strategic design element enables the absorbers to trap IR radiation within a particular spectral range.Intriguingly, the central wavelength of this spectrum can be predetermined during the design phase, thereby enabling a high degree of spectral control.As shown in figure 9, results from Gokhale et al support this idea [106].Their research substantiates the efficiency of this approach, demonstrating that resonant detectors based on this principle display an impressive absorbance rate of 46%, accompanied by a FWHM of 1.7 µm.This study underscores the potential of subwavelength plasmonic gratings for delivering precise spectral selectivity to detectors.
Improving the system is a method for customizing detector features.The first strategy is establishing the vibration system.The dual-mode resonant IR detector, utilizing an AlN-based Film Bulk Acoustic Resonator (FBAR) with resonant modes at around 2.5 GHz and 3.5 GHz, optimizes the IR detection [107].Sensitivity to IR illumination allows the frequency and reflection coefficient, S11, and valley value change to be tracked, facilitating IR signal detection.Dual modes enable frequency-hopping sensing, bolstering resilience against electromagnetic interference.The AlN FBAR, responsive to IR owing to its temperature-dependent resonance and charge carrier generation, enables four concurrent IR sensing signals within a single device.The final fabricated device exhibits ultra-low noise with a relatively compact design, which suggests immense potential for high-resolution and uncooled IR detection applications.The proposed sensor system, benefiting from advances in semiconductor technology, incorporates threshold-triggered micromechanical photoswitches (MPs) to selectively harness energy from specific IR radiation bands [108].The MPs turn on the sensor when the intensity of IR is strong enough.When intensity of IR is not strong enough, the MPs remain inactive, which means there is no power used when on standby.This design facilitates the use of near-zero power use during inactive periods.
The exploration of various device designs for the concurrent detection of multiple IR wavelengths on a single Focal Plane Array (FPA) necessitates a delicate balance between performance trade-offs [109].An array of pixels sensitized to disparate wavelengths offers several advantages but simultaneously restricts the FPA's per-wavelength availability.Layering detector materials can release several constraints yet complicates both the growth and fabrication processes.Similarly, the integration of an actively tunable filter into the detector introduces additional complexity and delay layers.A promising solution to these challenges is a Nanoantenna-Enabled IR Detector (NED).This inventive design permits the simultaneous readout of two-color bands within each pixel, without compromising the quantum efficiency or the pixel fill factor.By incorporating multiple nanoantenna coupled with a modified backside readout geometry, the NED design demonstrates promising potential through a combination of simulations and empirical measurements.This innovative approach provides an avenue toward a more efficient and versatile multispectral IR detection.

Photoconductive antennas
With the development of mobile devices, antennas play a crucial role in transmitting and receiving electromagnetic signals, enabling communication between electronic devices, such as radios, televisions, wireless routers, and mobile phones.Photoconductive antennas (PCAs) operate based on the photoconductive effect, wherein the electrical conductivity of a material changes when exposed to the electromagnetic wave radiation.Auston et al [110] firstly introduced the successful THz generation based on a picosecond photoconductive dipole structure in 1984, which offers a pioneer direction for coupling the photoconductive material with the THz wave emission and detection.There are two essential elements existed in the PCA detectors, a semiconductor-based substrate with an antenna structural design.When a laser with femtosecond pulse emits the surface of the PCA, the semiconductor substrate absorbs photons, leading to the generation of photoinduced electron-hole pairs.Rather than rely on an external DC bias to establish the electric field in PCA emitters, the incident THz pulse induces an electric field in the active region between the antenna arms, resulting in the directional movement of the carriers.As a result, a photocurrent, whose magnitude is proportional to the intensity of the THz emitter, is generated and detected.
From the perspective of the working mechanism, the PCA device has enormous potential in terahertz detection applications.However, due to the mismatch and the complex fabrication process regarding to the substrate and the antenna layers, only few strategies have been employed to increase the amount of photoinduced carriers and thereby enhance the quantum operating efficiency of the fabricated detector.Burford et al [111] made the first effort to integrate the plasmonic nanodisk with thin-film photoconductive layers, aiming to design an innovative PCA device.In their structural design, low temperature-GaAs layer has been used as the thin-film substrate which attempts to overcome the relatively low conversion efficiency challenge for traditional PCAs.Comparing the performance of the plasmonic thin-film photoconductive antenna with the modified conventional antenna, the conclusion of the performance improvement can be obtained based on the obtained THz electric filed under different optical pump laser illumination within the regime of 0-9 THz. Lee et al [112] designed and fabricated a PCA device based on GaAs substrate.By comparing the results with and without plasmonic grating structures between the electrodes, they have demonstrated that a highly localized electric field forms near the surface between the GaAs layer and the grating structure, which enhances the photo-absorption of the final PCA detector.Overall, photoconductive detectors are crucial in various applications, including THz communication, imaging, and spectroscopy system.

Summary of THz detectors
Table 3 summarizes the advantages and disadvantages introduced in the above subtitles.It can be seen that each of these detectors has unique properties and requires few improvements for future research.Researchers generally expect and hope to achieve several parameters, including fast response time, high signal-to-noise ratio, and great sensitivity.

Signal processing for terahertz communication systems
As terahertz technology continues to expand and deepen in the fields of ICT, non-destructive and security testing, biomedicine, environmental monitoring, and astronomy, it has shown broad prospects and potential [116].Taking THz band wireless communication and sensing systems as an example, the mainstream system architectures are solid-state electronics and photoelectric combinations.Researchers have achieved 10 Gbit/s high-speed remote wireless communication, and directly modulated terahertz communication systems have been derived and developed at an extremely fast pace in recent years.
The ongoing demand for broader bandwidth and faster data transmission rates [104,105] requires advanced THz devices and innovative signal processing techniques.This trend promotes the development of terabytes per second (Tbps) links [106].Meanwhile, it is also necessary to consider schemes such as Distance Adaptive Absorption Peak Modulation in the physical layer to achieve stealth and combat security and eavesdropping problems in terahertz communication links.This section provides an overview of state-of-the-art techniques for processing, analyzing, and extracting signals of the 'terahertz gap' in systems using terahertz sensors.High frequency, wide bandwidth, and narrow beam are the characteristics of the terahertz band, which lies between MMW and IR.However, these factors affect the quality and accuracy of terahertz signals because of different types of noise, such as      2. Simply the structural design [61,111,112] thermal noise, ambient noise, instrument noise, phase noise and mixed noise considerations.To improve the final quality of the obtained signal and reduce the noise as much as possible, filtering and amplification are required to attenuate or even eliminate the noise and interference among the terahertz signals.

Denoising
The wavelet transforms, as an extension of the Fourier transform, is widely used for denoising terahertz signals.It has a time-frequency localization property shared with the terahertz time-domain spectrum (THz-TDS), which is conducive to the sparse representation of the THz-TDS [117], and is therefore widely used in the denoising of terahertz signals.Wavelet transform-based denoising methods typically employ wavelet shrinkage.To satisfy the ideal state, the selected wavelet basis must have properties such as orthogonality, high vanishing moments, tight branching, symmetry, and anti-symmetry.The mainstream thresholding functions include BayesShrink and SureShrink methods, and thresholding functions such as hard thresholding, soft thresholding function proposed by Donoho [118], and the semi-soft thresholding function proposed by Gao and Bruce [119] can be selected.In terahertz signal processing, the above methods play an important role in improving the quality of terahertz signals.

Modulation
In terahertz communication systems, orthogonal frequency division multiplexing (OFDM) as a multi-carrier modulation technique suitable for high-speed data transmission can be compatible with existing systems with the following expressions, which can alleviate the problem of large Doppler frequency shift expansion; however, it has been welcomed by many researchers for its ability to improve detection accuracy and throughput, often working together with multiple input multiple output (MIMO) to solve, for a longer period of time in dominated high-speed wireless standards [120], but it is hampered by high PAPR.A single carrier (QAM) has a weaker transmission performance than OFDM with a lower computational complexity.Researchers have proposed CE-OFDM [121], CP-OFDM [122], MIMO-OFDMA, discrete Fourier transform-OFDM (DFT-s-OFDM), and SI-DFTs-OFDM based on OFDM, the latter of which has a 10-fold improvement in accuracy over conventional OFDM modulation schemes.QAM-based OQAM/FBMC [123] and DFT-s-OTFS have also been actively researched as schemes that can provide a lower PAPR and reduce the interference caused by Doppler expansion.In the future, all these multi-carrier modulation schemes will be investigated further.

Precoding
MIMO is a technique for transmitting data that uses multiple transmitting antennas and receiving antennas simultaneously and is now widely used in wireless communication.Precoding, as a beamforming (BF) for multiple streams, is a core functional module in the physical layer processing of LTE, can be considered as a category of MIMO, and it can be broadly classified into three categories: digital precoding, analogy precoding, and hybrid precoding, where hybrid precoding can combine the advantages of excellent performance and low cost of the former two systems [124].For the PCBS, AS, and BDS algorithms compared to DAoSA, DAoSA has lower power consumption, and it is noted that two low complexity solutions (shown in figure 10), EBE and VEC, have been developed for DAoSA in combination with ultra-massive MIMO (UM-MIMO) systems [125].The signals in the THz band are less susceptible to free-space diffraction or interantenna interference [126] but may be affected by fading and scattering during propagation, leading to high path loss, diffusion loss and distortion of the signal [127].Channel estimation must be accomplished in a timely manner to compensate, correct, and reconfigure the signals to overcome the power limitations [128].The UM-MIMO system, as an extension of MIMO technology that uses graphene and metamaterials to produce large antenna arrays and transceivers can provide considerable spatial multiplexing, introduce beamforming gains to extend the transmission distance, achieve higher spectral efficiency, and overcome the atmospheric terahertz fading effect [129].The precoding used in terahertz communication is slightly different from conventional techniques in that digital precoding results in more power consumption.Therefore, hybrid precoding, broadband precoding, and RISassisted THz precoding have been focused, the first of which is particularly promising.Park et al compared DAoSA and its two special cases, FC and AoSA, in terms of spectral efficiency and power consumption, and FC has higher power consumption while AoSA has higher performance loss [130,131].To maximize the spectral efficiency, more studies have attempted to achieve this through semidefinite relaxation, alternate optimization, and so on.In addition, THz broadband hybrid schemes against beam-splitting effects [131], as well as intelligent modulation of RIS electromagnetic waves, also offer new possibilities for terahertz communication and coding.In the future, researchers will continue to focus on unavoidable power consumption issues in terahertz UM-MIMO systems, security, and hardware shortages that also pose some limitations for precoding.

Decoding
Decoding the signal is also required by data detection, channel coding and other techniques to carry out the decoding of the signal, and both require extremely high baseband processing modules, innovative but less complex schemes should be considered, using appropriate joint algorithms, parameters, and models to establish co-optimization to achieve more accurate high-performance signal processing, capable of countering conflicts between terahertz channels and digital baseband systems.The three main decoding schemes are Turbo, which is performed on the data flow graph, LDPC, and Polar, which are performed on the tree structure [132].Finally, to further improve the efficiency of terahertz baseband signal processing, schemes that combine signal modulation, channel modulation, and data monitoring, similar to terahertz wireless bridges [133], which do not need to perform the three previously mentioned steps, may have a better prospect of design.

Machine learning
Owing to the channel characteristics of terahertz, conventional signal processing methods cannot fully achieve the desired performance in the terahertz band, and scientists have gradually introduced machine learning methods [134].Compared to conventional signal processing methods, deep learning can provide better robustness for ISAC in terms of data detection at the receiver [135], thus advancing 6 G and beyond [120][121][122].Machine learning plays an important role in signal preprocessing in terahertz systems, and more models applying deep learning such as SISR and improved GAN models have been proposed [133], providing the possibility of high-quality images for non-destructive detection.Meanwhile, in the preprocessing part of the terahertz signals, SNV, min-max normalization, and SG filtering are capable.The raw data with high latitude and high redundancy need data compression and dimensionality reduction.Following the extraction of right features for comprehensive signal acquisition and automated classification, a benchmark model is established for anomaly detection.As figure 11 shown, effective data identification and classification are then conducted using feature extraction methods such as PCA, PLS, t-SNE, NMF, and deep learning techniques like SVM, kNN, DA, and NB [127].These methods can also be applied for beam formation and efficient data detection in generalized index modulation schemes.The signal processing capability is compensated by the emerging IRS by simulating a deep learning-based configuration for UM-MIMO channel estimation [129].Signal processing plays a crucial role in the system and is a key link to realize the efficient realization of terahertz technology in various applications of its capabilities.There have been many recent advances in terahertz signal processing techniques for channel modelling, noise reduction and modulation, which open new perspectives for the system-level development of terahertz technology.

Conclusion
In conclusion, this paper provides a comprehensive overview of different types of THz detectors and signal processing techniques.It can be seen that these detectors have underscored critical roles and made extensive progress in sensitivity, bandwidth, and response time, revealing high potential in diverse applications, ranging from imaging and spectroscopy to communication.Meanwhile, with the assistance of signal processing methods, complex THz signals can be extracted into meaningful information, which enhances the utility of THz technology.
Nevertheless, forward research progress is not without the challenges.Researchers pursue even higher sensitivity, improve signal-to-noise ratios, and expand the THz spectral range to maintain the existing performance.Additionally, issues related to size reduction, lifetime, and reliability need to be addressed for the practical utilization of THz systems.Furthermore, ensuring experimental safety standards for both THz radiation and sensitive information is of great importance as well.As we look any further, the future of THz detectors and signal processing holds significant promise.Addressing and solving these challenges require an interdisciplinary combination of materials science and electrical engineering.Future research should focus on novel materials, innovative structural designs, and progressive signal processing algorithms.
In summary, the collocation between advanced terahertz detectors and sophisticated signal processing technologies will promote and revolutionize many fields.By overcoming the existing challenges and pursuing new avenues of research, related research is at the stage of unlocking the full potential of terahertz technology, which will further improve communication.

Figure 3 .
Figure 3.The typical structure of a bolometer.

Figure 5 .
Figure 5. Schematic diagram of an internal structure design of a Golay cell.Reproduced from [64].CC BY 4.0.

Figure 7 .
Figure 7.The structure of the detector with GR, AlN, and platinum.Reproduced from [104].CC BY 4.0.

Table 1 .
List of full name terminologies with acronyms.

Table 2 .
Comparison between the D-mode HEMTs and E-mode HEMTs.

Table 3 .
Summary of different THz detectors.