1. INTRODUCTION

Sixth-generation (6G) wireless technology has promised to deliver a peak data rate of 1 terabits/s with less than 100 µs latency [1,2]. Terahertz spectrum has been identified as one of the enablers for the realization of 6G broadband connectivity [3,4]. Attenuation of terahertz radiation by atmospheric gases such as water vapor and oxygen molecules limits terahertz transmission to specific frequency windows [e.g., D-band (110–170 GHz), H-band (220–325 GHz), and Y-band (325–500 GHz)] [1,5,6], which have relatively low atmospheric absorption, leading to the potential for high-speed communication with hundred GHz of bandwidth and pico-second-level symbol duration [1,7,8].

Recently, integrated localization and communication have gained momentum. Localization is the process of estimating the position of the target and, if required, the orientation, which is essential for location-aware communications [9,10] and tactile internet [11,12]. Using the current wireless network, localization and tracking have been demonstrated in outdoor scenarios such as vehicles on the road with meter-level accuracy using a carrier frequency of 28 GHz [13], and indoor scenarios using WiFi at the 2.4 and 5 GHz bands [14]. In addition, positioning with relative lateral and longitudinal accuracy of 0.1 m and less than 0.5 m, respectively, has been demonstrated using the 5G network for self-driving vehicles [15]. However, this is still not accurate enough to support applications such as future assistant robots in smart factories, where high precision at the centimeter level (${\lt}{2}\;{\rm cm}$) is needed for opening doors and picking up items [16]. A radar function has also been integrated with existing communication systems, IEEE 802.11 (including but not limited to the 2.4, 5, 6, and 60 GHz frequency bands) [17–20] and the 3rd Generation Partnership Project (3GPP, also known as 5G New Radio) frequency bands [21–23], which can only provide centimeter-level resolution due to limitations on the bandwidth and data rate.

In future 6G networks, integrated localization and communication is vital to achieve ubiquitous connectivity with high data rates and low latency. The terahertz band can offer improved localization performance due to large bandwidth availability and short wavelengths. Heretofore, there is some research work on the integration of imaging within terahertz communication systems [24,25], while, to the best of our knowledge, integrated localization and communication are yet to be widely investigated for deployment in the terahertz band. Moreover, most of limited existing works use conventional frequency modulated signal for target detection, which has challenges that include the requirement of an additional waveform apart from the communication signals, relatively low range resolution, and two functions not working simultaneously. For example, a range resolution of 1.58 cm has been demonstrated using linear frequency modulation (LFM) for radar, and orthogonal frequency-division multiplexing (OFDM) signals in a 340 GHz (Y-band) photonics-based communication system [26]. In terms of the D-band, a joint radar-communication complementary metal–oxide–semiconductor (CMOS) transceiver using an electronics-based method has been demonstrated, achieving a 1.25 cm range resolution with a frequency-modulated continuous-wave (FMCW) signal at 150 GHz by switching between the communication and radar modes [27]. To the best of our knowledge, a simultaneous localization and communication system with millimeter-order resolution at the D-band has not yet been reported.

Here, we demonstrate integrated high-resolution localization using a D-band photonics-based communication system. Millimeter-order accuracy (${\lt} 4 \;{\rm mm}$) can be simultaneously achieved using the same signal as the 5 Gbps communication link. The content is organized as follows: First, we describe the D-band communication system with a 5 Gbps data rate and a 1.5 m link distance and investigate the system’s capabilities and the hardware limitation for the communication link. We then utilize the same communication signal for measuring the distance of the target, followed by a discussion of the detection accuracy and resolution.

2. TERAHERTZ COMMUNICATION SYSTEM

Currently, two approaches are utilized for realizing terahertz communication systems: photonics-based and electronics-based [8]. Electronics-based systems use frequency multiplying chains to up-convert a lower-frequency signal from the microwave and millimeter-wave bands [6,28,29]. These systems have a simple setup and high terahertz power (order of milliwatt [30,31] and watt [32,33]). However, electronics-based systems suffer from waveform distortion and high phase noise due to the nonlinear effects of frequency multipliers and mixers. The high gain usually leads to a reduction in the bandwidth [34].

The photonics-based systems use photomixing to achieve optical-to-terahertz down-conversion. Two laser beams are used to generate the terahertz beat note [35–37]. Photonics-based systems come with much lower terahertz power than the electronics-based systems (${\lt}{10}\%$). Nevertheless, terahertz signals generated from photonics-based systems have higher spectral purity and lower amplitude and phase noise compared to electronics-based systems. The existing optical modulators can provide a high bandwidth and modulation index. Furthermore, the photonics-based systems have the potential to be integrated into existing fiber optic networks and harness the photonics advances for the miniaturization of the hardware [34,38]. This work is based on the photonics-based system.

The first communication band beyond 100 GHz is the D-band, where communication links using electronics-based approaches have been demonstrated for the D-band [39–44]. However, photonics-based communication systems have been demonstrated mainly in the H-band [28,29,37,45–48] and Y-band [5,49–52]. Some D-band photonics-based communication systems have been demonstrated to include 4K video streaming [35], equalizers [53–55], bidirectional transmission [56], optical-carrier-suppressing signal generation [57], and multiple-input multiple-output (MIMO) [58].

A. Photonics-Based Terahertz Communication System

Figure 1 shows the schematic of the photonics-based terahertz system including wireless communication and target localization parts [35,59]. Since there is only one terahertz receiver available, the detector is moved for communication and target localization measurements (dash boxes in Fig. 1). The system consists of two tunable 1550 nm lasers with 1 MHz linewidth and minimum frequency step size of ${\lt}{10}\;{\rm MHz}$ (Toptica DFB pro). The lasers operate at the terahertz difference frequency, in this case, the D-band. The baseband signal is modulated on one laser before coupling with the second laser (Fig. 1), providing a single sub-carrier modulation. The modulation can also happen after the coupling of two lasers, providing a double sub-carrier modulation, which is claimed to have a higher terahertz signal-to-noise ratio (SNR) [60].

 

Fig. 1. Schematic of the photonics-based terahertz system for wireless communication and target localization. (AWG: arbitrary waveform generator; EDFA: erbium-doped fiber amplifier; EVOA: electronic variable optical attenuator; and LNA: low-noise amplifier).

Download Full Size | PDF

An arbitrary waveform generator (AWG, Keysight M8190A with a 5 GHz bandwidth) is used to generate the baseband signal with 400 m ${V_{{\rm pp}}}$ amplitude. The generated RF signal is first amplified to 5.5 V by a modulator driver (Thorlabs MX40A with 40 Gbps maximum data rate) and then modulated (amplitude shift keying modulation, ASK) to one laser using an electro-optic Mach–Zehnder modulator (Thorlabs LN05S-FC with a 40 GHz bandwidth). The modulator driver has a lower cutoff frequency of 100 KHz that, as demonstrated later, leads to errors in the modulation of signals with low frequencies, e.g., acoustic signals. The bias voltage of the modulator is controlled by the modulator driver, which continuously monitors the quadrature bias point with a positive slope using a dither tone (1 kHz, 600 m ${V_{{\rm pp}}}$) to correct drift in the bias voltage. Since the output optical power of the modulator is low (${\lt}{2}\;{\rm mW}$), an erbium-doped fiber amplifier (EDFA, Thorlabs EDFA100P) is used after the modulator to amplify the laser power to ${\sim}35\;{\rm mW}$, a similar power level to the unmodulated laser 2. A 3 dB coupler is used to combine the lasers. An electronic variable optical attenuator (EVOA, Thorlabs EVOA1550A) is connected before the terahertz transmitter to limit the optical power to avoid overloading the photomixer. Figure 2 shows the optical spectrum of the two lasers with a frequency separation of 160 GHz measured using an optical spectrum analyzer, where the baseband signal is modulated on laser 1. A slight increase in the bandwidth is observed after modulation (red curve), compared to the optical signal without modulation (blue curve). The signal-to-noise ratio (SNR) is higher than 50 dB for both modulated and unmodulated optical signals.

 

Fig. 2. Optical spectrum of the optical signals with a frequency separation of 160 GHz for terahertz wave generation, where the blue curve is before modulation and the red curve is after modulation.

Download Full Size | PDF

The terahertz beam is generated by feeding the combined laser into either a D-band uni-traveling-carrier photodiode (UTC-PD) photomixer (NTT D-band photomixer module) or a broadband antenna-integrated InGaAs photodiode (Toptica TeraScan 1550). The UTC-PD can provide ${\sim}0.25\;{\rm mW}$ of terahertz power through the D-band, while the InGaAs photodiode can provide ${\sim}0.1\;{\rm mW}$ in the same band. A horn antenna with 25 dBi gain is connected to the UTC-PD to couple the generated beam into free space, while a silicon lens is used for the InGaAs photodiode.

For free-space terahertz transmission, two plano-convex lenses with a focal length of 50 mm and a diameter of 35 mm are used to collimate the terahertz beam to free space and focus it into the detector (Fig. 1). On the receiver side, another horn antenna is used to couple the signal to a D-band zero-bias Schottky diode (ZBD, Virginia Diodes WR6.5ZBD-F20), which also demodulates the terahertz signal. The DC component in the demodulated baseband signal is then filtered using a Bias-Tee and amplified using a low noise amplifier (LNA) with ${\sim}14\; {\rm dB}$ gain. The amplified signal is finally measured by a real-time oscilloscope (Keysight Infiniium-MXR-Series, with 6 GHz bandwidth).

B. Characterization of the Terahertz Communication System

We measured the received baseband signal strength (in our case, a peak-to-peak amplitude ${V_{{\rm pp}}}$) and bit error rate (BER) for communication quality evaluation, which can be measured by the oscilloscope. To characterize the system performance, here we used a non-return-to-zero (NRZ) pseudo-random binary sequence (PRBS) data with a 5 Gbps bit rate (limited by the bandwidth of AWG) and 400 m ${V_{{\rm pp}}}$ amplitude as the baseband signal (amplitude modulation). First, we measured the ${V_{{\rm pp}}}$ and BER for carrier frequency from 110 to 180 GHz at link distance of 1.2 and 1.5 m, shown in Fig. 3. The bias voltage and photocurrent on the terahertz transmitter are fixed at ${-}{2}\;{\rm V}$ and 7 mA, respectively, which are recommended values by the manufacturer. It can be observed that the 130–165 GHz frequency range has high measured ${V_{{\rm pp}}}$ and low BER, which is consistent with the output and responsivity characterization of the UTC-PD and ZBD, respectively. There is an error-free transmission window, 124–178 GHz, for the 1.2 m link distance; for a 1.5 m link distance, there is no error-free range and the overall received ${V_{{\rm pp}}}$ drops more than 5 mV. We attribute the increase in BER for the longer link distance to the divergence of the terahertz beam, which is partially collected using a 35 mm lenses. The diameter of the terahertz beam ${D_{\rm T}}$ is a function of the link distance, which can be approximated as ${D_{\rm T}} \sim L \cdot \lambda /{D_{\rm L}}$, where $L$ is the link distance, $\lambda$ is the wavelength of the carrier wave, and ${D_{\rm L}}$ is the diameter of the collimating lenses [35]. Therefore, in our case, the terahertz beam diameter at the receiver is 93.6–60.3 mm for 1.2 m and 117–75.4 mm for 1.5 m through D-band, which are more than two times larger than the diameter of the lens (35 mm), leading to only partial signals being received by the terahertz detector. In terms of received terahertz power, if using two identical lenses to collimate the beam, the received power is $P = {P_{0}} \cdot {({D_{\rm L}}/{D_{\rm T}})^2}$ [35], where ${P_{0}}$ is the emitted power from the terahertz transmitter. Therefore, only 13%–34% and 9%–22% of the emitted power is received at the 1.2 m and 1.5 m link distance, respectively. It can be seen that in order to increase the ratio of received power using lenses with larger diameters is a solution. For example, if lenses in the system are replaced by 100 mm diameter lenses, the system is expected to receive all of the emitted terahertz signals for ${\lt}{3.6}\;{\rm m}$ link distances without other modifications of the system. In addition, link distance can be extended by applying terahertz LNAs. If using a terahertz LNA with 30 dB gain (commercially available), the link distance could be extended to 64.8 m with 42 cm diameter lenses.

 

Fig. 3. Measured ${V_{{\rm pp}}}$ and BER of a 5 Gbps PRBS baseband signal with a carrier frequency of 110–180 GHz for 1.2 m (solid lines) and 1.5 m (dashed lines) link distances.

Download Full Size | PDF

In short, we described the setup of our photonics-based D-band terahertz system and elaborated on the system’s communication capability in terms of the BER performance and the current limitations in terms of the maximum link distance due to hardware constraints. The data rate can be simply improved by using AWG and an oscilloscope with a higher bandwidth. Moreover, other factors, such as the atmospheric humidity and system cutoff frequency, may also limit the communication performance. We conducted experiments verifying that there is a minimal effect on the transmission due to the humidity increases at the D-band (see Section 1 in Supplement 1 for details), and the cutoff frequency of the system was also measured to be around 50 kHz (see Section 2 in Supplement 1 for details), which will not affect the system performance at the 5 Gbps data rate. In the next section, we will present the experimental validation for the feasibility of integrating the localization functionality into this communication system.

3. TARGET LOCALIZATION USING THE COMMUNICATION SYSTEM

As described in the previous section, the free-space terahertz transmission of this system is highly directional since the beam is collimated via the plano-convex lenses and focused toward a known direction. Thus, the target location can be interpreted easily once the distance to the target is acquired. To achieve the functionality of distance measurement, we adopt the reflected time of flight (ToF) technique widely used for distance measurement [61,62], which measures the time delay from the transmission of a pilot waveform to the reception of the echo signal reflected by the target. Ideally, the ToF approach utilizes the phases of the transmitted signal and its reflected echo to determine the time of flight, while the ASK-modulated communication relies on the amplitude of the transmitted signal to recover the message conveyed. Thus, by transmitting the modulated communication waveform as the pilot waveform for ToF ranging, it is theoretically feasible to convey the message to a target while detecting the range from the target’s reflected echo simultaneously. We realized such a method with our practical system and validated the feasibility via experiments, which will be discussed below.

A. Experimental Setup

Other than the components elaborated in the previous section, we introduce a silicon beam splitter with a split ratio of 50:50 after the lenses of Tx, which can be rotated along the vertical axis without changing the position, and then split and reflect the terahertz beam. This setup enables the measurement of the internal delay of the system and thus the time synchronization between Tx and Rx, as illustrated in Figs. 4(a) and 4(b). We consider the transmitter and receiver as one synchronized system, which can be placed closely in the same chassis for implementation. When the transmitter sends communication signals to the terminal, the receiver can detect the reflected signals from the target in the communication path at the same time. Then, the ToF can be measured to detect the distance from the target. Figure 4(a) demonstrates the configuration for measuring the time delay introduced by the terahertz system, where the beam splitter is rotated for 90°, directly reflecting the signal from the terahertz transmitter to the receiver; Fig. 4(b) demonstrates the target ranging configuration that is commonly used for terahertz reflection spectroscopy [63–65]. A metallic plate perpendicular to the output terahertz beam from the beam splitter is set as the target to emulate the surface of the targeted object. Figure 4(c) shows the photograph of the physical experimental setup in this configuration, where the target distance is the distance between the beam splitter and the metallic plate. As mentioned above, we utilized an NRZ PRBS signal with a 5 Gbps rate and 300 m ${V_{{\rm pp}}}$ amplitude as the pilot waveform, emulating a system with an ongoing communication operation. Figure 4(d) shows an example snapshot of the transmitted pilot waveform (reference) and the reflected waveform received under this configuration, where the reference signal is from the direct AWG output and the received signal is from the terahertz receiver.

 

Fig. 4. Schematic of the experimental setup for reflection measurements, where Tx is the transmitter and Rx is the receiver. (a) Direct reflection for measuring system delay. (b) Normal incident reflection for measuring the distance between the system and a metallic plate. (c) A photograph of the setup of normal incident reflection for distance measurement. (d) Waveforms used for measurements, where the reference signal is the direct output from AWG and the received signal is from the terahertz receiver.

Download Full Size | PDF

With extracted measurement data, the time delay between the received signal and the reference is calculated using the cross-correlation method in a back-end processor running Matlab software. Intuitively, the total measured delay (${\tau _{\rm m}}$) is the summation of the system delay (${\tau _{{\rm sys}}}$) and the ToF for the round trip of the target distance. Thus, the estimated target distance $\hat d$ can then be calculated by

Note that since the calculations for both ${\tau _{\rm m}}$ and ${\tau _{{\rm sys}}}$ are performed with the discrete data samples extracted from the oscilloscope, the step size of the cross-correlation is the sampling period of the data. Thus, the distance detection resolution of this system, in terms of the minimum recognizable distance change, is constrained by the sampling resolution. For data with a sampling rate ${f_{\rm q}}$, the expected sampling resolution of the system for distance detection is $\Delta {d_{\rm Q}} = \frac{c}{{{f_{\rm q}}}}$. Particularly in this system, the oscilloscope provides a sampling frequency of 16 GHz with an $8 \times$ interpolation, resulting in an effective sampling resolution $\Delta {d_{\rm Q}} \approx 1.17\; {\rm mm}$. Since the output of the system can only be discrete values with $1/{f_{\rm q}}$ steps, any distance $d$ has a quantization error ${e_{\rm q}}$, which is the deviation to its nearest quantization level, so

We took measurements with the terahertz transmitter and receiver, lenses, and beam splitter remaining at their respective fixed locations, while the metallic plate (target) moved to various locations (closer or further) along a fixed guide rail. The orientation of the metallic plate remained unchanged. All measurements were taken when the metallic plate was steady for zero-Doppler distance. The ruler on the guide rail provided the ground-truth distances for each measurement, while the distance between the beam splitter and the zero point of the ruler was pre-measured as a reference. We carried out multiple rounds of trials, where the metallic plate was randomly moved to 20–25 different locations on the guide rail with ground-truth distances from 50 to 100 mm.

B. Distance Detection Outcomes

With all the measurements from multiple experimental trials, the outcome of the ToF distance estimation is presented in Fig. 5(a). The horizontal axis shows the ground-truth distances of the target locations, while the vertical axis indicates the estimated distances from the terahertz communication system using Eq. (1). The red circles represent the average estimated distance for each location from multiple measurements, while the error bars indicate the span of the measured results. Meanwhile, the blue triangles mark the corresponding ground-truth values. It is observable that the estimated distances are close to the actual distance, and show the same linear increasing trend, which validates the feasibility of such a distance detection using the D-band communication system. However, a 3–6 mm error exists between the estimated distances and the respective ground truth.

 

Fig. 5. Average distance estimation outcomes with multiple measurements per location (a) before calibration and (b) after calibration.

Download Full Size | PDF

As discussed earlier, we expect the system to deliver distance estimations with a resolution close to the hardware boundary $\Delta {d_{\rm Q}} \approx 1.17\; {\rm mm}$. A drifting error of 3–6 mm observed is severe in terms of the accuracy performance to this scale of sensitivity. Therefore, we took a deep dive into the estimation errors presented by the system, which will be discussed next.

 

Fig. 6. Impact of non-ideal beam splitter on system delay measurement and ToF calculation.

Download Full Size | PDF

C. Distance Detection Accuracy and Calibration

First, we noticed that there was a relatively constant offset due to the imperfect ground-truth reference or system delay computation. By analyzing and tracing the potential sources of error, we determined that one possible root cause for the offset error is the non-ideal effect of the silicon beam splitter. Equation (1), which is used to generate the estimated distance, is based on the assumption of an ideal beam splitter with zero thickness. However, in a practical setup, the thickness of the silicon beam splitter alters the signal propagation from the transmitter due to refraction. It introduces an additional delay (${\tau _{\rm S}}$) when the signal propagates through, as shown in Fig. 6. It is observable that the measured total delay in the system (${\tau _{\rm m}}$) also includes the propagation delay inside the splitter (${\tau _{\rm S}}$) and the corresponding estimated ToF ${\hat \tau _{{\rm ToF}}}$ is

Since the thickness of the beam splitter is consistent, it is possible to directly calibrate the distance estimation result with the offsets mentioned above. Figure 5(b) demonstrated the effect of calibration with all the offsets compensated. It is observable that offset error has been successfully eliminated, while the calibrated estimates are fairly accurate versus the ground-truth distances. Alternatively, in practical applications, this offset error can also be calibrated with a simple linear regression method from two or multiple known ground truth points.

 

Fig. 7. Insights of the estimation errors remnant after calibration: (a) estimation errors for each measurement in mm, (b) the quantization error ${e_{\rm q}}$ in mm, and (c) the random drift ${e_{{\rm dr}}}$ in terms of multiples of $\Delta {d_{\rm Q}}$.

Download Full Size | PDF

In addition to the offset error elaborated above, there are other error factors involved, as illustrated in Fig. 7. Figure 7(a) indicates the remaining estimation errors after calibration ($e = {\hat d_{{\rm calibrated}}} - d$), which can be further decomposed into two components:

Both the quantization error and the random drift error mentioned above are inevitable. Thus, they become the limiting factors affecting the distance detection resolution. Next, we will elaborate in detail.

D. Distance Detection Resolution

We experimentally verified the impact of each error factor on the detection resolution. As discussed previously, it is reasonable to consider the detection resolution on distance in the form of the minimum separation distance between two distinct locations that are distinguishable by the system with adequate accuracy. In other words, the detection is beyond the system resolution if the actual distance difference between two measurements $\Delta d = {d_{1}} - {d_{2}}$ is too small for the system to adequately distinguish; that is, the system recognizes their estimated distances ${\hat d_{1}} = {\hat d_{2}}$. To evaluate the resolution of the system, three data groups were formed with pairs of measurements collected of $\Delta d = 1\; {\rm mm}$, 2 mm, and 3 mm, respectively. For each measurement pair, the difference in estimated distances $\Delta \hat d = {\hat d_{1}} - {\hat d_{2}}$ was evaluated, as shown in Fig. 8.

 

Fig. 8. Distributions of the difference of estimated distances for measurement pairs with 1 mm, 2 mm, and 3 mm actual distance gaps.

Download Full Size | PDF

For each group of measurement pairs, the pin locations on the horizontal axis indicate the corresponding $\Delta d$ and $\Delta \hat d$, while the height of the pin on the vertical axis shows the percentage of the samples among the respective group. Intuitively, since the two target locations are distinct ($\Delta d \gt 0$) for each measurement pair involved, it is expected that $\Delta \hat d \gt 0$ for any successful distinction between the two estimated distance (“Good Estimation”), as shown by blue pins with circles. On the other hand, an unsuccessful distinction can be marked with $\Delta \hat d \le 0$, which contradicts with the ground truth $\Delta d \gt 0$ (“Bad Estimation”), as shown by red pins with crosses. As indicated, for $\Delta d = 1\;{\rm mm}$, 40% of the measurement pairs end up as bad estimation, which is equivalent to a 60% accuracy in distinguishing two distances with a 1 mm difference. When $\Delta d$ increases to 2 mm, the accuracy is boosted to 91.6%, and further increased to 94.7% for $\Delta d = 3\;{\rm mm}$. The accuracy is 100% when $\Delta d$ is 4 mm and above. The observations are aligned with the discussions earlier that the highest achievable resolution is $\Delta {d_{\rm Q}} = 1.17\;{\rm mm}$, which is the boundary due to the hardware capability. However, the random drift error of another 1.17 mm due to the ambiguity can occur, which leads to certain estimation errors appearing in a 2 mm resolution evaluation. Thus, empirically it is reasonable to state that the system has a 100% accuracy performance in detecting a distance with a 4 mm resolution, with an adequate safe margin. Note that the accuracy performance displayed in Fig. 8 is limited by the number of samples available (about 10–20 samples per group). The exact resolution might be deferred if more data is available or evaluated with a different accuracy threshold to define the resolution.

4. CONCLUSION

We demonstrated a D-band photonics-based communication system designed to investigate wireless communication and integrated localization and sensing. We showed a 5 Gbps communication link for a 1.5 m distance. Atmospheric humidity and the cut-off frequency limit of the system have been experimentally investigated to have a minimal effect. The data transfer rate and link distance can be simply improved by using AWG and an oscilloscope with a higher bandwidth and larger diameter lenses, respectively. The link distance also can be improved by using a terahertz low-noise amplifier with a higher gain. We then utilized the same communication signal to measure the target location. We demonstrated that millimeter-order range resolution (${\lt}{4}\;{\rm mm}$) can be achieved simultaneously with the same system and discussed the system calibration due to the thickness of the silicon beam splitter and the potential sources of the errors (quantization and random drift errors). This revealed that the error due to the thickness of the beam splitter can be eliminated, but the quantization error and the random drift error are inevitable and are the limiting factors of the resolution achieved. Alternatively, when there is no communication data transmitted in the system, other waveforms such as a squared wave can be actively sent for distance measurements. (See Section 3 in Supplement 1 for details.)

To conclude, we demonstrate target localization and communication with the same signal simultaneously, which achieves millimeter-order range resolution. Future works may include an investigation on the relationship between the localization performance and communication data rate. This work paves the path toward the implementation of photonics-based D-band integrated localization and communication. Furthermore, this system is easily expandable to any terahertz communication band.