Open Access
Add to BibSonomy
Abstract
Coherent terahertz (THz) wireless communication using silicon photonics technology provides critical solutions for achieving high-capacity wireless transmission beyond 5G and 6G networks and seamless connectivity with fiber-based backbone networks. However, high-quality THz signal generation and noise-robust signal detection remain challenging owing to the presence of inter-channel crosstalk and additive noise in THz wireless environments. Here, we report coherent THz wireless communication using a silicon photonic integrated circuit that includes a dual-parallel Mach-Zehnder modulator (MZM) and advanced digital signal processing (DSP). The structure and fabrication of the dual-parallel MZM-based silicon photonic integrated circuit are systematically optimized using the figure of merit (FOM) method to improve the modulation efficiency while reducing the overall optical loss. The advanced DSP compensates for in-phase and quadrature (IQ) imbalance as well as phase noise by orthogonally decoupling the IQ components in the frequency domain after adaptive signal equalization and carrier phase estimation. The experimental results show a reduction in phase noise that induces degradation of transmission performance, successfully demonstrating error-free 1-m THz wireless transmission with bit-error rates of 10−6 or less at a data rate of 50 Gbps.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
The terahertz (THz) frequency band from 0.1 to 10 THz, which is the unallocated frequency region between infrared and microwaves, has uniquely attractive features: the ability to penetrate clothing, paper, and plastics and be absorbed by water and organic substances [1,2]. THz features play important roles in superior resolution imaging [3,4] and sensing [5,6] within industrial inspections and medical diagnosis, time-resolving capabilities in spectroscopy [7], and significant capacity enhancement of data communication [8,9]. In particular, THz communication has a larger bandwidth than millimeter wave communication, resulting in faster, and larger quantities of data transfer over a limited transmission distance [10]. Accordingly, THz wireless communication exploiting the THz spectrum beyond 275 GHz with low atmospheric loss is crucial to meet the demands of high data throughput of up to 1 Tbps and very low latency of less than 1 ms, which 5G communication has not yet addressed [11]. Furthermore, THz wireless communication enables massive data transmission and ultra-reliable connectivity for human-to-machine and machine-to-machine communications beyond 5G and 6G networks [11–13]. However, there are several technical challenges to overcome before this can fully support communications beyond 5G and 6G networks, such as high free space path loss [14], increased crosstalk between channels caused by inter-channel interference with neighboring users [15,16], ultra-high energy consumption [17], and quality degradation of THz signals, including phase and amplitude distortion [18]. Highly directional dynamic massive multiple-input multiple-output (MIMO) antennas with narrow beam width arrays have been developed to compensate for the free space path loss and inter-channel crosstalk [19]. However, practical implementations of THz wireless communication for various 6G services still require various cost-effective and energy-efficient THz generation and detection methods.
Photonics-based THz communication systems utilizing advanced optical fiber communication technologies can achieve high-speed wireless transmission capacity as well as seamless connectivity between an optical backbone/access and wireless networks [20,21]. For instance, optical heterodyning and advanced optical modulation methods allow THz signal generation with higher-order modulation schemes such as quadrature phase-shift keying (QPSK) [22] and multilevel quadrature amplitude modulation (m-QAM) [23]. Silicon photonics technology can serve as an effective alternative to reduce both the power consumption and cost of THz wireless communications systems [24,25]. Silicon photonic circuits provide dense integration and high-volume production of complex photonic functionalities, including signal generation [26,27], detection [28], phase modulation [29], and intensity modulation [30,31], by using a large-scale photonic-electronic integration platform based on a mature, high-yield fabrication process [32]. High-speed silicon photonic modulators are particularly crucial for fulfilling the enormous bandwidth demands of communication beyond 5G and 6G while lowering the power consumption per bit. However, silicon photonic circuits required for the realization of competitive THz wireless communication still face several challenges associated with high-quality THz signal generation, co-integration of electronic and photonic devices, and reproducible construction.
Coherent THz signal transmission technology is required for 6G fronthaul and backhaul capabilities, with a consistent transmission rate of over 100 Gbps and peak rates reaching 140 Gbps. The coherent THz signal transmission technology adapted from long-haul optical communication systems consists of higher-order modulation and coherent detection combined with digital signal processing (DSP) at the terahertz band [33]. The DSP in coherent detection plays an important role in compensating for free-space transmission impairments from the full complex electric field of the received THz signal, thereby extending transmission capacity [34]. For instance, a 10-m free-space transmission of 10-Gbaud QPSK signals has been demonstrated using an optical sub-harmonic in-phase and quadrature (IQ) mixer and DSP [35,36]. However, the experimental results show a low bit error rate (BER) and IQ imbalance in the constellation diagram. Consequently, the DSP methods are still not suitable for counteracting IQ imbalances or completely recovering the phase information of high-speed THz wireless signals. As a result, there is strong motivation for developing low-loss and modulation-efficient silicon photonic circuits and high-performance DSP, which enable coherent THz wireless communications with a high transmission capacity of 100 Gbps or higher.
In this study, we report high-speed coherent THz wireless communication using a dual-parallel Mach-Zehnder modulator (MZM)-based silicon photonic integrated circuit (DPMZM-SiPIC), successfully demonstrating error-free 1-m THz wireless transmission with bit-error rates of 10−6 or less at a data rate of 50 Gbps. The DPMZM-SiPIC approach improves the modulation efficiency by systematically optimizing the circuit structure and fabrication based on the figure of merit (FOM) while reducing the overall optical loss. Furthermore, advanced DSP resolves the phase noise and IQ imbalance problems of THz wireless signals. The experimental results of coherent THz wireless communication using the DPMZM-SiPIC show high-speed THz signal generation and noise-suppressed signal reconstruction, and hence, significant performance enhancement of coherent THz wireless transmission.
2. Coherent terahertz wireless communication using a DPMZM-SiPIC
Coherent THz wireless communication using a DPMZM-SiPIC is based on advanced optical fiber communication technology. This is well known for achieving a high traffic rate, increased bandwidth, and strong reliability required in optical core networks such as telecommunication centers, campuses, hospitals, and data centers. A conceptual description of coherent THz wireless communication using a DPMZM-SiPIC is shown in Fig. 1. This system consists of two main parts: the first is a THz transmitter (THz TX) responsible for efficient IQ modulation and high-quality THz signal generation, and the second is a THz receiver (THz RX) responsible for the coherent detection of THz signals and the exact recovery of transmitted information.
Fig. 1. Conceptual description of coherent terahertz (THz) wireless communication using a dual parallel MZM-based silicon photonic integrated circuit (DPMZM-SiPIC). This system comprises a THz transmitter (THz TX) based on the DPMZM-SiPIC for efficient in-phase and quadrature (IQ) modulation and high-quality THz signal generation, and a THz receiver (THz RX) based on advanced digital signal processing (DSP) for coherent detection and exact information recovery. LD: laser diode, PC: polarization controller, MZM: Mach-Zehnder modulator, PS: phase shifter, ATT: optical attenuator, BPF: bandpass filter, LO: local oscillator, RF: radio frequency, and IoT: internet of things.
The DPMZM-SiPIC in the THz transmitter offers coherent modulation schemes, including QPSK and m-QAM with high spectral efficiency, using a silicon photonic modulator with a dual-parallel MZM structure and relatively low drive voltage. The IQ-modulated optical signal (λSIG) and the continuous wave (CW) optical signal (λLO) are combined using the multi-mode interferometer (MMI) coupler of the DPMZM-SiPIC and then passed to an optical amplifier. The optical amplifier compensates for the insertion loss of the DPMZM-SiPIC, and the optical bandpass filter (BPF) removes the amplified spontaneous emission (ASE) noise of the optical amplifier. The THz wireless signals at the frequency spacing (${f_{THz}} = {f_{LO}} - {f_{SIG}}$) between two independent lasers are generated by heterodyne beating in a photomixer. The structural design and fabrication optimization of the DPMZM-SiPIC plays an important role in achieving high-quality THz signal generation; the process significantly improves the higher-order modulation efficiency and optical signal-to-noise ratio (SNR). At the THz receiver, the electrical field of the THz signals is down-converted to intermediate frequency (IF) signals, including the amplitude and phase information, using coherent detection that combines the IQ-modulated THz wireless signals and an electrical local oscillator (LO) in a sub-harmonic mixer. The intermediate frequency signals are then captured by a real-time scope and digitally down-converted to the baseband signal by advanced offline DSP. Coherent detection by heterodyne demodulation effectively enhances the received THz signal using the LO within the mixing process and reduces the detected noise power by limiting the signal received by the bandwidth of the IF electronics. With wireless transmission over free space, the received THz signals are distorted due to inter-symbol interference and the inherent nonlinearity of electronics such as the sub-harmonic mixer, which degrades transmission performance. Therefore, DSP in coherent THz wireless communication is critical to recover the distorted signal and improve transmission performance. Advanced offline DSP in coherent THz wireless communication using the DPMZM-SiPIC performs accurate down-conversion to the baseband for digital frequency domains, clock phase recovery to estimate the frequency offset, constant-modulus-algorithm (CMA)-based adaptive equalization, carrier phase estimation to eliminate the phase noise, Fourier filtering, and decoding. In particular, Fourier filtering in advanced offline DSP provides additional processing power to remove residual phase noise. Consequently, coherent detection using heterodyne demodulation combined with advanced offline DSP results in high receiver sensitivity, a long wireless transmission distance, and a large transmission capacity for coherent THz wireless communication. Hence, this approach achieves modulation-efficient THz signal generation using the DPMZM-SiPIC and counteracts THz transmission impairment using advanced DSP, resulting in ultra-high-capacity wireless communication and a seamless connection between the optical core and wireless networks.
3. Device characterization of a DPMZM-SiPIC
3.1 Design and fabrication of a dual parallel MZM-based silicon photonic integrated circuit
The DPMZM-SiPIC in the THz transmitter, which consists of a dual-parallel Mach-Zehnder modulator (DPMZM), grating couplers, and multimode interferometer (MMI) couplers, is important for the efficient IQ modulation and low-loss on-chip combination of modulated signals and LO signals (Fig. 2(a)). The DPMZM consists of two carrier-depletion MZMs connected in parallel, and the carrier depletion mode utilizes the change in free carrier concentration due to the depletion of electrons and holes by reversely biasing a p-n junction for phase modulation of optical signals. The thermal phase shifter on one arm (with MZM 2, corresponding to the Q component) generates a stationary π/2 shift between the I and Q components. In addition, on-chip termination resistors for impedance matching at the end of the transmission lines are implemented with doped silicon. The variable optical attenuator was realized utilizing a Mach-Zehnder interferometer structure with a thermal phase shifter in order to adjust the optical LO power level. The DPMZM-SiPIC was constructed on a 200-mm silicon-on-insulator (SOI) wafer with a buried oxide thickness of 2 µm, a silicon device layer thickness of 220 mm, and a silicon substrate thickness of 725 µm in an open-access foundry. An optical microscope image of the DPMZM-SiPIC is shown in Fig. 2(a). The carrier-depletion MZM employs a series push-pull (SPP) driving configuration to reduce losses and improve the modulation bandwidth and linearity (Fig. 2(b)). A p-type silicon substrate with a low resistivity of 16.5 Ohm $\cdot$ cm was implanted with boron. Doping consists of four levels: n++/n/p/p++, where the p++ and n++ heavily doped regions provide ohmic contacts, and the p-and n-doped regions construct embedded p-n junctions in the rib waveguide.
Fig. 2. Design and fabrication of a dual parallel Mach-Zehnder modulator (MZM)-based silicon photonic integrated circuit (DPMZM-SiPIC). (a) Microscope image of the fabricated DPMZM-SiPIC. (b) Electrical driving scheme of the carrier-depletion MZM based on the series push-pull configuration.
The key performance factors of a DPMZM-SiPIC for high-speed THz wireless signal generation include high modulation efficiency, low insertion loss, high modulation speed, low energy consumption, a large extinction ratio, and a small footprint for high integration density. Furthermore, the doping concentration options of the DPMZM-SiPIC p-n junctions-based phase shifter are critical for enhancing modulation efficiency and reducing optical loss. In coherent modulation, the modulation efficiency is defined as the product of the voltage (${V_\pi }$) applied to the phase shifter to achieve a π phase shift and the length ($L$) of the phase shifter, where a smaller ${V_\pi } \cdot L$ indicates higher modulation efficiency. The modulation efficiency (${V_\pi } \cdot L$) and optical loss ($\alpha $) were numerically calculated by changing the p-and n-doping concentrations. The simulated results clearly show a trade-off between the modulation efficiency and optical loss for p-n junction doping concentrations; that is, as the doping concentrations increases, the modulation efficiency improves, whereas the optical loss increases. Consequently, the trade-off between the modulation efficiency and optical loss of the DPMZM-SiPIC is optimized by numerically calculating the maximum FOM [37] for the p-n junction doping concentrations. The FOM for QPSK modulation depends mainly on the phase shifter cutoff frequency (${f_c})$, energy consumption per bit (${E_{bit}}$), length of the phase shifter to achieve a π phase shift (${L_{ps}}$), and optical loss per length of the p-n junction ($\alpha $) (Eq. (1)). Note that a low ${E_{bit}}$ indicates an energy efficient QPSK modulation. From the relation between the energy consumption per bit and the modulation efficiency, the FOM is also expressed as a function of the phase shifter modulation efficiency and optical loss using the following equation:
Fig. 3. Optimization for doping concentrations of the p-n junction-based phase shifter and the structure of the coplanar stripline (CPS) metal transmission line in the dual parallel Mach-Zehnder modulator-based silicon photonic integrated circuit (DPMZM-SiPIC). (a) The figure-of-merit (FOM) for quadrature phase-shift keying (QPSK) modulation dependence on p and n doping concentrations, optimizing the trade-off between modulation efficiency and optical loss for p-n junction doping concentrations. The ${V_\pi } \cdot L$ decreases as the doping concentrations increase, resulting in enhancement of modulation efficiency, whereas the optical loss increases as the doping concentrations increase. (b) Characteristic impedance dependence on the width and space of the CPS metal transmission line. To prevent electrical reflections and consequent interference, the width and space are chosen to match the characteristic impedance to 50 Ω. (c) Effective refractive index dependence on the width and space of the CPS metal transmission line. The width and space are chosen to be 30 µm and 12 µm, respectively, to ensure the microwave effective refractive index matches the optical group refractive index of 3.8.
The electrode configuration of the DPMZM-SiPIC carrier-depletion MZM is a traveling wave electrode (TWE) with a coplanar stripline (CPS) transmission line. To increase the electro-optic (EO) bandwidth in this scheme, it is critical to terminate the characteristic impedance at 50 Ω and match the microwave effective refractive index to the optical group refractive index of 3.8. The matched impedance prevents electrical reflections and consequent interference with the electrical signal stream. The characteristic impedance and microwave effective refractive index were numerically calculated by changing the width and space of the CPS metal transmission line (Fig. 3(b) and Fig. 3(c)). As a result, the width and space of the microwave transmission metal line were found to be 30 µm and 12 µm, respectively.
3.2 Device characterization of the fabricated DPMZM-SiPIC
The performance of the fabricated DPMZM-SiPIC, including its modulation efficiency, 3-dB bandwidth, and optical loss, is characterized by measuring the small signal EO response and optical transmission spectra. The EO S21 response of MZM 1 in the DPMZM-SiPIC was measured using a lightwave component analyzer (LCA), which offers balanced measurements up to 67 GHz. The radio frequency (RF) signal from the LCA (Agilent Technologies, N4373D) was applied to MZM 1 through a high-speed probe. The transmission line was terminated with an on-chip termination resistor of 50 Ω. The measured EO 3-dB bandwidths of MZM 1 at 0 V and 5 V reverse biases were 13 GHz and 18 GHz, respectively (Fig. 4(a)). The low resistivity silicon substrate can be replaced with a high resistivity (750 Ω $\cdot$ cm) silicon substrate to resolve the limitations of the EO 3 dB bandwidth. In addition, the EO 3 dB bandwidth can be further enhanced by adding a mid-doping region (p+, n+) between the ohmic contact region (p++, n++) and the p-n junction region (p, n). The electrical-electrical (EE) S11 response, which indicates the microwave reflection at the entrance of the TWE due to impedance mismatch, was measured to be less than −12 dB in the frequency range up to 40 GHz (Fig. 4(b)). The experimental results show that the matched impedance of the fabricated DPMZM-SiPIC effectively prevented the electrical reflections. The reflection dips at specific frequencies, such as around 33 GHz and 18 GHz at 0 V and 5 V in the EE S11 response of Fig. 4(b), are caused by destructive interference between the reflected waves from the inlet and the far-end of the MZM. Furthermore, the frequency at which the dip occurs varies according to the bias voltage, since the PN junction capacitance and resistance vary with the bias voltage, changing the microwave effective index [38].
Fig. 4. The electro-optic (EO) response and the electro-electric of the fabricated dual parallel Mach-Zehnder modulator (MZM)-based silicon photonic integrated circuit (DPMZM-SiPIC). (a) The EO S21 response of MZM 1 in the DPMZM-SiPIC. The experimental result exhibits an EO 3 dB bandwidth of 18 GHz at a 5 V reverse bias. The limitation of the EO 3 dB bandwidth can be resolved by replacing the silicon substrate with a high resistivity alternative of 750 Ω ${\cdot} $ cm and adding a mid-doping region (p+, n+) between the ohmic contact region (p++, n++) and the p-n junction region (p, n). (b) The EE S11 response of MZM 1 in the DPMZM-SiPIC. The microwave reflection is measured to be less than −12 dB in the frequency range up to 40 GHz, indicating that the characteristic impedance is effectively matched to 50 Ω.
The total loss of the DPMZM-SiPIC was measured to be 21.4 dB including an on-chip insertion loss of 10.1 dB and 8 dB loss from two grating couplers, and a 3-dB power loss from the MMI coupler for combining IQ-modulated and continuous wave optical signals. In detail, the on-chip insertion loss of the DPMZM-SiPIC originates from a routing waveguide loss of 5.5 dB, an MMI coupler loss of 1.5 dB, a p-n phase shifter loss of 2.1 dB, and a thermal phase shifter loss of 1 dB. The on-chip insertion loss is proportional to the length of the p-n shifter; hence, a reduction in the length of the p-n phase shifter can reduce the on-chip insertion loss. The total loss can be further reduced by replacing the optical combiner from the MMI coupler with a wavelength division multiplexing (WDM) coupler. High-speed photodiode (PD) integration for direct THz generation without connecting to optical fiber links can also significantly reduce the total loss by eliminating the output grating coupler [28,39,40].
Lowering the voltage (${V_\pi }$) of the phase shifter is critical for reducing the voltage swing requirement of voltage drivers and fully realizing higher-order modulation in coherent THz wireless communications. A ${V_\pi }$ value of 5 V was applied to the phase shifter to achieve a π phase shift in the two MZM arms. The 3-mm phase shifter in the two MZM arms exhibited a low ${V_\pi } \cdot L$ value of 1.4 $\textrm{V} \cdot \textrm{cm}$ at a 2 V reverse bias. The measured modulation efficiency agreed well with the calculated results. Moreover, the optical transmission spectra of the fabricated DPMZM-SiPIC at 0 V and 5 V reverse biases showed that the modulation efficiency was dependent on the direct current (DC) bias voltage. For instance, a Vπ·L was measured to be 0.97 V·cm at a low DC bias of 1 V, whereas a Vπ·L was increased to 1.55 V·cm at a higher DC bias of 4 V. Low DC bias conditions resulted in high modulation efficiency levels; however, this produced a higher junction capacitance, which is not suitable for high-speed modulation. Conversely, higher DC bias increased the modulation speed, but resulted in lower modulation efficiency values. Therefore, the DC bias voltage needs to be carefully set to find a balance between the modulation efficiency and speed. Consequently, the DPMZM-SiPIC based on FOM-guided systematic optimization resulted in more than two times improvement in both modulation efficiency and phase shifter loss, compared to other on-chip MZM modulators [41–43]. Furthermore, the on-off extinction ratio was measured as high as 30 dB at 0 V bias voltage. The eye diagrams for 20 Gbps and 25 Gbps on a 215−1 non-return-to-zero (NRZ) pseudo random binary sequences (PRBSs) signals were measured at a 4.47 V peak-to-peak driving voltage and 5 V reverse bias (Fig. 5). The measured eye diagram exhibited clear eye openings of 20 Gbps and 25 Gbps.
Fig. 5. The eye diagrams for 20 Gbps and 25 Gbps on a 215−1 non-return-to-zero (NRZ) pseudo random binary sequences (PRBSs) signals measured at a 4.47 V peak-to-peak driving voltage and 5 V reverse bias. The experimental results show clear eye openings for 20 Gbps and 25 Gbps.
4. Experimental demonstration of coherent THz wireless transmission
4.1 Coherent modulation performance of a back-to-back optical link
High-quality IQ modulated signal generation is critical for coherent THz wireless transmission based on silicon photonic integrated circuits. The coherent modulation performance of a back-to-back optical link was experimentally demonstrated by generating QPSK optical signals with modulation speeds from 20 to 50 Gbps using the DPMZM-SiPIC and demodulating optical signals received in a coherent receiver using advanced offline DSP. Before the offline DSP, I/Q skew was removed by using delay function of PPG and DSA. The signal distortion recovery performance in advanced offline DSP, including adaptive equalization, carrier phase estimation, and Fourier filtering, was confirmed by constellation diagrams of 20 Gbps and 50 Gbps QPSK signals (Fig. 6). The experimental results show that the carrier phase estimation, after adaptive equalization, compensates for phase noise, and Fourier filtering effectively suppresses residual phase noise, resulting in a clear constellation for the four phase states of the QPSK signal. The error-free transmission of QPSK signals over a back-to-back optical link for four data rates of 20, 30, 40, and 50 Gbps was measured by error counting the received signal using knowledge of the transmitted pseudorandom pattern; its length was 215−1, demonstrating the high performance of QPSK optical signal generation using the DPMZM-SiPIC in the THz TX and signal recovery using advanced offline DSP in the THz RX. Note that the IQ-modulated optical signal is transmitted to a coherent receiver (Finisar, CPRV1225A, optical input power: Max. 16 dBm for the LO signal and −20∼+6dBm for the modulated signal) after adjusting the received power to 0 dBm using an external variable optical attenuator. On the other hand, a local oscillator (LO) for optical homodyne detection is the same as the laser that generates IQ-modulated optical signals and adjusts the received powers to 14 dBm at the coherent receiver. The experimental results demonstrate that the DPMZM-SiPIC effectively generates high-quality IQ modulated signals, and advanced DSP resolves phase noise and IQ imbalance, further improving the performance of high-speed coherent transmission.
Fig. 6. The coherent modulation performance of a back-to-back optical link for various data rates from 20 to 50 Gbps. The constellation diagrams exhibit the recovery performance of advanced offline digital signal processing (DSP) for signal distortions of 10 Gbaud and 25 Gbaud quadrature phase-shift keying (QPSK) signals on a back-to-back optical link. The phase noise is compensated for by the carrier phase estimation, and the residual phase noise is clearly suppressed by Fourier filtering, resulting in clear constellations for 10 Gbaud and 25 Gbaud QPSK signals.
4.2 Demonstrating coherent THz wireless transmission using the DPMZM-SiPIC
The low-cost and energy-efficient realization of coherent THz wireless communication substantially contributes to the capacity requirement of future outdoor high-speed mobile communication, in which each mobile base station is connected via a THz link in a small-cell mobile network. Coherent THz wireless communication using the DPMZM-SiPIC was experimentally demonstrated by transmitting the THz signal over a 1-m free space and measuring the BERs, error vector magnitudes (EVMs), and constellation diagrams of the received QPSK signals at line rates from 20 to 50 Gbps (Fig. 7). The polarized light from a C-band laser 1 (Santec TSL-510, linewidth: 200 kHz) at a 1547.8 nm center wavelength followed by a fiber polarization controller was amplified to 19 dBm by an EDFA and coupled to the DPMZM-SiPIC through a grating coupler with a transmission loss of 4 dB. Two 215−1 de-correlated PRBSs from a bit pattern generator (SHF 12100 B) were amplified to reach a swing voltage of 7 V with a DC bias of 5 V, and then applied to each phase shifter in the DPMZM-SiPIC to achieve a peak-to-peak phase modulation of π in each arm. A multi-contact DC probe (FormFactor, WPH-910-NS) was connected to the heater, and the phase shift was adjusted to bias the MZM at a quadrature point. The optical QPSK signal from laser 1 with an optical SNR of approximately 40 dB was generated in the DPMZM-SiPIC. The LO signal from C-band laser 2 at a 1550.0 nm center wavelength was separated from the QPSK signal by 2.2 nm, corresponding to 275 GHz. For high-quality THz generation, we adjusted the modulated signal power and the LO signal power using EDFAs and the integrated variable optical attenuator within DPMZM-SiPIC so that there was no power difference between modulated and LO signals. The optical QPSK and optical LO signals were combined using the MMI coupler and coupled to the EDFA via an output grating coupler within the DPMZM-SiPIC. An EDFA and optical bandpass filter (Alnair Labs, BVF-100) compensated for the total loss of the DPMZM-SiPIC and eliminated the ASE noise of the EDFA, respectively. After adjusting the polarization of the optical signal to the unitraveling-carrier photodiode (UTC-PD) using a polarization controller, the UTC-PD (NEL, J-band photomixer) integrated with a WR-3.4 waveguide generated THz signals with a frequency of 275 GHz via the heterodyne beating method. Two THz lenses (Thorlabs TPX 100) were used to compensate free-space path loss. The diameter and focal length of the lenses were 1.5 inches and 100 mm, respectively.
Fig. 7. Experimental demonstration of coherent THz wireless transmission using the dual parallel-Mach-Zehnder modulator-based silicon photonic integrated circuit (DPMZM-SiPIC). THz signals with a frequency of 275 GHz via heterodyne beating between the quadrature phase-shift keying (QPSK) signal with an optical signal-to-noise ratio (SNR) of approximately 40 dB and the local oscillator (LO) signal are generated in the unitraveling-carrier photodiode (UTC-PD). After transmitting the THz signal over a 1-m free space, the transmission performance is assessed by measuring the bit error rates (BERs), error vector magnitudes (EVMs), and constellation diagrams of the received QPSK signals at line rates from 20 to 50 Gbps.
In the THz receiver, the THz signals after 1-m free-space transmission were captured at the WR-3.4-based corrugated horn antenna (VDI, frequency coverage: 220-330 GHz) with the same gain as the THz transmitter. The gain of the two horn antennas was 52 dBi (26 × 2 dBi). The received THz signals were frequency-down-converted at the THz mixer (VDI, SAX3.4), amplified by an RF amplifier, and captured using at 50-GS/s digital sampling oscilloscope (Tektronix, DPO 72004) with a 20-GHz electrical bandwidth. Advanced DSP estimates the BERs by comparing the received demodulation pattern with the transmitted PRBS pattern and calculates the error vector magnitude by analyzing the difference between a collection of ideal transmitted symbols and the received symbols in the I-Q plane. Coherent THz wireless communication recovers the transmitted THz signal over free space and obtains the original information by acquiring the entire phase and magnitude of the received THz signals using coherent detection and compensating for transmission impairments using advanced DSP. Hence, advanced DSP in the THz RX is essential for achieving higher bit rates, greater degrees of flexibility, simpler photonic systems, and better transmission performance. After down-conversion to the base band in advanced DSP, resampling and normalization were performed to adapt the fast oscilloscope sampling rate to the DSP data rate. The Viterbi and Viterbi phase recovery algorithm [44] was utilized to estimate the carrier phase from phase noise due to transmission impairment and the frequency offset between the laser of the QPSK signal and the laser of the LO signal. CMA-based adaptive signal equalization [45] minimizes the influence of additive noise, and decision-directed least mean square (DD-LMS)-based carrier phase estimation [46] tracks the carrier phase fluctuation through the adaptive control of the filter tap coefficients in each finite impulse response (FIR) filter. In addition, IQ imbalance due to residual phase noise is effectively resolved by extracting amplitude and phase information using the Fourier transform, orthogonally decoupling the IQ components in the frequency domain, and reconstructing noise-robust signals in the time domain.
The transmission performance of coherent THz wireless communication using the DPMZM-SiPIC was analyzed by measuring the BER against the photocurrent for 1-m wireless transmission of QPSK signals at a carrier frequency of 275 GHz. For lines rates of 20, 30, 40, and 50 Gbps, no errors were measured in the recording lengths of 106 symbols at the UTC-PD photocurrents of 1.0, 1.0, 1.5, and 2.0 mA, respectively. Note that the responsivity of the UTC-PD was 0.22. The EVM, defined as the root-mean-square value of the difference between a collection of ideal transmitted symbols and the received symbols in the I-Q plane, is an important performance factor in measuring the quality of IQ-modulated signals and evaluating optical I-Q transmitters or receivers in coherent wireless communication [47]. The EVM is more appropriate for evaluating the performance at a high SNR, which requires numerous symbols for accurate error counting. The BERs for the 1-m wireless transmission of QPSK signals at various line rates were estimated using the conversion equation from the EVM to BER described in Eq. (2) below. Note that the conversion equation is discussed in detail in [47].
The BERs for the 1-m THz wireless transmission of QPSK signals at line rates of 20, 30, 40, and 50 Gbps were estimated to be less than 10−6 at the UTC-PD photocurrents of 1.0, 1.0, 1.5, and 2.0 mA, respectively (Fig. 8). This exhibits an exact agreement with the experimental BER measurement results by error counting the received signal using knowledge of the transmitted pseudorandom pattern. In the 1-m wireless transmission, the estimated BERs for 20 to 50 GHz were obtained below the forward error correction threshold of 10−3 using 7% hard decision FEC (HD-FEC) overhead codes. It might be noted that the BER of 20 Gb/s is worse than that of 30 Gb/s when photocurrent is higher than 1 mA. This is because the accuracy of the carrier phase estimation function that compensates phase noise in a DSP is improved as the data rate increases, and this effect becomes visible in a situation where BER is mainly limited by phase noise. The constellation diagrams for line rates of 20, 30, 40, and 50 Gbps are also displayed in the insets for the lowest BER values. Experimental results show that the clear constellation for the four phase states of the QPSK signal is maintained, except for a slight thickening as the line rate increases from 20 to 50 Gbps under 1-m coherent THz wireless transmission. This demonstrates the enhanced transmission performance of coherent THz wireless communication using the DPMZM-SiPIC.
Fig. 8. The measured bit error rate (BER) performance and constellation diagrams of coherent THz wireless transmission using the dual parallel Mach-Zehnder modulator-based silicon photonic integrated circuit (DPMZM-SiPIC) after a 1-m THz wireless transmission of quadrature phase-shift keying (QPSK) signals for various data rates including 20, 30, 40, 50 Gbps. No errors are measured in our recording length of 106 symbols at the unitraveling-carrier photodiode (UTC-PD) photocurrents of 1.0, 1.0, 1.5, 2.0 mA, respectively. Consequently, the BERs measured by error counting the received signal are consistent with the BERs estimated using the conversion equation from the measured error vector magnitude (EVM) to BER.
EVMs and constellations of QPSK signals after a 1-m THz wireless transmission for various data rates from 20 to 50 Gbps were measured and compared to those after a back-to-back optical link. The removal of residual phase noise using Fourier filtering of advanced offline DSP is more effective in THz wireless links with severe signal distortion due to phase noise than back-to-back optical links (Fig. 9(a)). As the modulation speed of the QPSK signals increased from 20 to 50 Gbps, the EVM for the 1-m THz wireless link increased from 18% to 21%, whereas the EVM increased from 8% to 12% for the back-to-back optical link (Fig. 9(b)). When the data rate increased from 20 to 50 Gbps, the EVM penalties of the 1-m THz wireless link and back-to-back optical link increased by approximately 3% and 4%, respectively. This indicates an increase in BER over two orders of magnitude owing to the limited bandwidth of the RF amplifier and the DPMZM-SiPIC. Furthermore, EVMs after a 1-m THz wireless transmission increased by approximately 10% compared to those after the back-to-back optical link for all data rates. An EVM penalty of 10% between a 1-m THz free-space link and back-to-back optical link is introduced by propagation loss in a wireless link and imperfections in the frequency response and linearity of THz devices. The EVM measurements show that the EVM penalty for the THz wireless link increases by more than two times compared to the penalty for an increase in the data rate from 20 to 50 Gbps. Therefore, coherent THz wireless communication using the DPMZM-SiPIC can achieve a BER of 2 × 10−9 corresponding to a 17% EVM requirement for a wide-area base station according to TS36.104 [48], with compensation of propagation loss along the wireless link by utilizing a low-noise amplifier (LNA) or a sub-harmonic mixer with low conversion loss. In addition to the transmission rate, the wireless transmission distance is also important for evaluating the capacity of coherent THz wireless communication and further applying THz wireless backhauling to small-cell networks. The high system-level performance from the experimental results opens up the possibility of further extension of the THz wireless transmission distance, application of higher-order modulation formats, and/or multi-channel THz wireless transmission based on frequency division multiplexed carriers [49]. The main limitation of the current system is the 3-dB bandwidth of the DPMZM-SiPIC less than 20 GHz owing to problems in the fabrication process of the silicon photonic circuit. Improving this is subject to current research and is expected to dramatically enhance the performance of coherent THz wireless communication. 100-Gbps and beyond high-capacity wireless transmission is currently being studied by applying advanced modulation formats, including m-quadrature amplitude modulation (m-QAM) and polarization division multiplexing (PDM) to coherent THz wireless communication using the DPMZM-SiPIC.
Fig. 9. Error vector magnitudes (EVMs) and constellation diagrams of quadrature phase-shift keying (QPSK) signals at unitraveling-carrier photodiode (UTC-PD) photocurrent of 3 mA after a 1-m THz wireless transmission for various data rates from 20 to 50 Gbps, compared to those after the back-to-back optical link. (a) The constellation diagrams indicating the recovery performance of advanced offline digital signal processing (DSP) for the signal distortion of 25 Gbaud QPSK signals on the 1-m THz wireless link. (b) Comparison of the measured EVMs between the 1-m THz wireless link and back-to-back optical link as the modulation speed of QPSK signals increase from 20 to 50 Gbps. The EVM penalty for the THz wireless link increases by over two times compared to the penalty for the increase in data rate from 20 to 50 Gbps.
5. Conclusions
In conclusion, this study has successfully demonstrated a coherent THz wireless communication system using a chip-scale DPMZM-SiPIC and advanced digital signal processing. We performed the first experimental verification of the FOM-guided systematic optimization method for silicon photonic integrated circuits with IQ modulation to generate high-quality and high-speed THz signals. The experimental results clearly show that the optimization method substantially improves both modulation efficiency and phase shifter loss compared to other on-chip MZM modulators. The microwave reflection is reduced to −12 dB or less in the frequency range up to 40 GHz through the design of metal transmission lines to match characteristic impedance and refractive index. In addition, Fourier filtering of advanced offline DSP significantly suppresses IQ imbalance due to residual phase noise after phase noise compensation by the DD-LMS-based carrier phase estimation. As a result, the experimental results clearly demonstrated that our approach achieves 1-m THz wireless transmission with a BER of 10−6 that is significantly lower than the FEC limit of 10−3 at a data rate of 50 Gbps per carrier, using the chip-scale DPMZM-SiPIC and advanced digital signal processing. The improved system-level performance allows smooth scaling of the overall capacity of THz transmission up to hundreds of GHz, by employing multiple frequency division multiplexed carriers modulated with QPSK signal per carrier. These experimental results provide a cost-effective and energy-efficient THz wireless communication technique for various types of ultra-high-capacity wireless links beyond 5G and 6G networks, which can transfer massive amounts of data and guarantee ultra-reliable connectivity. Moreover, coherent THz wireless communication using the DPMZM-SiPIC can provide an effective route for developing highly sensitive, high-speed, low-cost sensing and networking techniques in the terahertz range.
Funding
Ministry of Science and ICT, South Korea (21ZH1100).
Acknowledgements
This study was supported by Electronics and Telecommunications Research Institute (ETRI) grant funded by the Korean government [21ZH1100, Study on 3D communication technology for hyper-connectivity].
Disclosures
The authors declare no conflicts of interest.
Data availability
The data underlying the results presented in this paper are not publicly available at this time, but may be obtained from the authors upon reasonable request.
References
1. H. Elayan, O. Amin, B. Shihada, R. M. Shubair, and M.-S. Alouini, “Terahertz Band: The Last Piece of RF Spectrum Puzzle for Communication Systems,” IEEE Open J. Commun. Soc. 1, 1–32 (2020). [CrossRef]
2. P. H. Siegel, “Terahertz technology,” IEEE Trans. Microwave Theory Tech. 50(3), 910–928 (2002). [CrossRef]
3. R. I. Stantchev, X. Yu, T. Blu, and E. Pickwell-MacPherson, “Real-time terahertz imaging with a single-pixel detector,” Nat Commun 11(1), 2535 (2020). [CrossRef]
4. S. C. Chen, L. H. Du, K. Meng, J. Li, Z. H. Zhai, Q. W. Shi, Z. R. Li, and L. G. Zhu, “Terahertz wave near-field compressive imaging with a spatial resolution of over lambda/100,” Opt. Lett. 44(1), 21–24 (2019). [CrossRef]
5. R. I. Stantchev, D. B. Phillips, P. Hobson, S. M. Hornett, M. J. Padgett, and E. Hendry, “Compressed sensing with near-field THz radiation,” Optica 4(8), 989–992 (2017). [CrossRef]
6. R. I. Stantchev, B. Sun, S. M. Hornett, P. A. Hobson, G. M. Gibson, M. J. Padgett, and E. Hendry, “Noninvasive, near-field terahertz imaging of hidden objects using a single-pixel detector,” Sci. Adv. 2(6), e1600190 (2016). [CrossRef]
7. J. Neu and C. A. Schmuttenmaer, “Tutorial: An introduction to terahertz time domain spectroscopy (THz-TDS),” J. Appl. Phys. 124(23), 231101 (2018). [CrossRef]
8. M. Tonouchi, “Cutting-edge terahertz technology,” Nat. Photonics 1(2), 97–105 (2007). [CrossRef]
9. T. Kürner and S. Priebe, “Towards THz Communications - Status in Research, Standardization and Regulation,” J. Infrared, Millimeter, Terahertz Waves 35(1), 53–62 (2014). [CrossRef]
10. I. F. Akyildiz, J. M. Jornet, and C. Han, “Terahertz band: Next frontier for wireless communications,” Physical Communication 12, 16–32 (2014). [CrossRef]
11. M. H. Alsharif, M. A. M. Albreem, A. A. A. Solyman, and S. Kim, ““Toward 6G Communication Networks: Terahertz Frequency Challenges and Open Research Issues,” Computers,” Materials & Continua 66(3), 2831–2842 (2021). [CrossRef]
12. K. B. Letaief, W. Chen, Y. Shi, J. Zhang, and Y.-J. A. Zhang, “The Roadmap to 6G: AI Empowered Wireless Networks,” IEEE Commun. Mag. 57(8), 84–90 (2019). [CrossRef]
13. S. Chen, Y.-C. Liang, S. Sun, S. Kang, W. Cheng, and M. Peng, “Vision, Requirements, and Technology Trend of 6G: How to Tackle the Challenges of System Coverage, Capacity, User Data-Rate and Movement Speed,” IEEE Wireless Commun. 27(2), 218–228 (2020). [CrossRef]
14. J. M. Jornet and I. F. Akyildiz, “Channel Modeling and Capacity Analysis for Electromagnetic Wireless Nanonetworks in the Terahertz Band,” IEEE Transactions on Wireless Communications 10(10), 3211–3221 (2011). [CrossRef]
15. R. Chataut and R. Akl, “Massive MIMO Systems for 5G and beyond Networks—Overview, Recent Trends, Challenges, and Future Research Direction,” Sensors 20(10), 2753 (2020). [CrossRef]
16. Z. Hossain, S. H. Vedant, C. R. Nicoletti, and J. F. Federici, “Multi-user Interference Modeling and Experimental Characterization for Pulse-based Terahertz Communication,” in 3rd ACM International Conference on Nanoscale Computing and Communication (ACM NANOCOM'16) (2016).
17. J. M. Jornet and I. F. Akyildiz, “Joint Energy Harvesting and Communication Analysis for Perpetual Wireless Nanosensor Networks in the Terahertz Band,” IEEE Trans. Nanotechnol. 11(3), 570–580 (2012). [CrossRef]
18. K. Tekbıyık, A. R. Ekti, G. K. Kurt, and A. Görçin, “Terahertz band communication systems: Challenges, novelties and standardization efforts,” Physical Communication 35, 100700 (2019). [CrossRef]
19. A. H. Naqvi and S. Lim, “Review of Recent Phased Arrays for Millimeter-Wave Wireless Communication,” Sensors 18(10), 3194 (2018). [CrossRef]
20. T. Nagatsuma, G. Ducournau, and C. C. Renaud, “Advances in terahertz communications accelerated by photonics,” Nat. Photonics 10(6), 371–379 (2016). [CrossRef]
21. A. J. Seeds, H. Shams, M. J. Fice, and C. C. Renaud, “TeraHertz Photonics for Wireless Communications,” J. Lightwave Technol. 33(3), 579–587 (2015). [CrossRef]
22. X. Yu, R. Asif, M. Piels, D. Zibar, M. Galili, T. Morioka, P. U. Jepsen, and L. K. Oxenlowe, “400-GHz Wireless Transmission of 60-Gb/s Nyquist-QPSK Signals Using UTC-PD and Heterodyne Mixer,” IEEE Trans. Terahertz Sci. Technol. 6(6), 765–770 (2016). [CrossRef]
23. A. Stöhr, M. F. Hermelo, M. Steeg, P.-T. Shih, and A. Ng’oma, “Coherent Radio-over-Fiber THz Communication Link for High Data-Rate 59 Gbit/s 64-QAM-OFDM and Real-Time HDTV Transmission,” in Optical Fiber Communication Conference (OFC 2017), Tu3B.2 (2017).
24. J. Xie, W. Ye, L. Zhou, X. Guo, X. Zang, L. Chen, and Y. Zhu, “A Review on Terahertz Technologies Accelerated by Silicon Photonics,” Nanomaterials 11(7), 1646 (2021). [CrossRef]
25. R. Sabella, “Silicon Photonics for 5G and Future Networks,” IEEE J. Sel. Top. Quantum Electron. 26(2), 1–11 (2020). [CrossRef]
26. G. Carpintero, S. Hisatake, D. de Felipe, R. Guzman, T. Nagatsuma, and N. Keil, “Wireless Data Transmission at Terahertz Carrier Waves Generated from a Hybrid InP-Polymer Dual Tunable DBR Laser Photonic Integrated Circuit,” Sci Rep 8(1), 3018 (2018). [CrossRef]
27. S. Bae, J. Park, S. Han, B. G. Kim, M. Kim, K. Yu, and Y. C. Chung, “A Cost-Effective 2-Channel OTDM System Implemented With Sinusoidally Modulated Light Source,” IEEE Access 8, 157504–157509 (2020). [CrossRef]
28. T. Harter, S. Muehlbrandt, S. Ummethala, A. Schmid, S. Nellen, L. Hahn, W. Freude, and C. Koos, “Silicon–plasmonic integrated circuits for terahertz signal generation and coherent detection,” Nat. Photonics 12(10), 625–633 (2018). [CrossRef]
29. S. Ummethala, T. Harter, K. Koehnle, Z. Li, S. Muehlbrandt, Y. Kutuvantavida, J. Kemal, P. Marin-Palomo, J. Schaefer, A. Tessmann, S. K. Garlapati, A. Bacher, L. Hahn, M. Walther, T. Zwick, S. Randel, W. Freude, and C. Koos, “THz-to-optical conversion in wireless communications using an ultra-broadband plasmonic modulator,” Nat. Photonics 13(8), 519–524 (2019). [CrossRef]
30. R. Degl’Innocenti, S. J. Kindness, H. E. Beere, and D. A. Ritchie, “All-integrated terahertz modulators,” Nanophotonics 7(1), 127–144 (2018). [CrossRef]
31. S.-R. Moon, S. Han, S. Yoo, H. Park, W.-K. Lee, J. K. Lee, J. Park, K. Yu, S.-H. Cho, and J. Kim, “Demonstration of photonics-aided terahertz wireless transmission system with using silicon photonics circuit,” Opt. Express 28(16), 23397–23408 (2020). [CrossRef]
32. T. Pinguet, S. Denton, S. Gloeckner, M. Mack, G. Masini, A. Mekis, S. Pang, M. Peterson, S. Sahni, and P. D. Dobbelaere, “High-Volume Manufacturing Platform for Silicon Photonics,” Proc. IEEE 106(12), 2281–2290 (2018). [CrossRef]
33. K. Endo, Y. Hashimoto, T. Tanaka, T. Takamichi, K. Fukuchi, M. Toyoshima, and Y. Takayama, “Development and evaluation of a digital signal processing for single polarization QPSK modulation format,” Proc. SPIE 8246, 82460A (2012). [CrossRef]
34. S. Koenig, D. Lopez-Diaz, J. Antes, F. Boes, R. Henneberger, A. Leuther, A. Tessmann, R. Schmogrow, D. Hillerkuss, R. Palmer, T. Zwick, C. Koos, W. Freude, O. Ambacher, J. Leuthold, and I. Kallfass, “Wireless sub-THz communication system with high data rate enabled by RF photonics and active MMIC technology,” 2014 IEEE Photonics Conference, 414–415 (2014).
35. A. Kanno, T. Kuri, I. Morohashi, I. Hosako, T. Kawanishi, Y. Yoshida, and K. Kitayama, “Coherent MMW/terahertz signal transmission with frequency-reconfigurable RoF transmitter based on an optical frequency comb,” in IEEE Global Communications Conference (GLOBECOM'13), 2686–2691 (2013).
36. A. Kanno, T. Kuri, I. Morohashi, I. Hosako, T. Kawanishi, Y. Yoshida, and K.-I. Kitayama, ““Coherent Terahertz Wireless Signal Transmission Using Advanced Optical Fiber Communication Technology,” J Infrared Milli Terahz Waves 36(2), 180–197 (2015). [CrossRef]
37. J. Witzens, “High-Speed Silicon Photonics Modulators,” Proc. IEEE 106(12), 2158–2182 (2018). [CrossRef]
38. C.-M. Kim and Y.-T. Han, “Optical Response of Traveling-Wave Optical Modulator with 3-Section Phase Reversal,” J. Opt. Soc. Korea 5(3), 76–82 (2001). [CrossRef]
39. L. Vivien, A. Polzer, D. Marris-Morini, J. Osmond, J. M. Hartmann, P. Crozat, E. Cassan, C. Kopp, H. Zimmermann, and J. M. Fédéli, “Zero-bias 40Gbit/s germanium waveguide photodetector on silicon,” Opt. Express 20(2), 1096–1101 (2012). [CrossRef]
40. E. Lacombe, C. Belem-Goncalves, C. Luxey, F. Gianesello, C. Durand, D. Gloria, and G. Ducournau, “10-Gb/s Indoor THz Communications Using Industrial Si Photonics Technology,” IEEE Microw. Wireless Compon. Lett. 28(4), 362–364 (2018). [CrossRef]
41. A. Samani, M. Chagnon, D. Patel, V. Veerasubramanian, S. Ghosh, M. Osman, Q. Zhong, and D. V. Plant, “A Low-Voltage 35-GHz Silicon Photonic Modulator-Enabled 112-Gb/s Transmission System,” IEEE Photonics J. 7(3), 1–13 (2015). [CrossRef]
42. A. Samani, V. Veerasubramanian, E. El-Fiky, D. Patel, and D. V. Plant, “A Silicon Photonic PAM-4 Modulator Based on Dual-Parallel Mach–Zehnder Interferometers,” IEEE Photonics J. 8(1), 1–10 (2016). [CrossRef]
43. H. Sepehrian, J. Lin, L. A. Rusch, and W. Shi, “Silicon Photonic IQ Modulators for 400 Gb/s and Beyond,” J. Lightwave Technol. 37(13), 3078–3086 (2019). [CrossRef]
44. Q. Zhang and C. Shu, “Viterbi and Viterbi Algorithm based Phase Recovery for Probabilistically Shaped Signals,” J. Lightwave Technol. 39(5), 1364–1370 (2021). [CrossRef]
45. I. Fatadin, D. Ives, and S. J. Savory, “Blind Equalization and Carrier Phase Recovery in a 16-QAM Optical Coherent System,” J. Lightwave Technol. 27(15), 3042–3049 (2009). [CrossRef]
46. M. S. Faruk, “Blind equalization and carrier-phase recovery based on modified constant-modulus algorithm in PDM-QPSK coherent optical receivers,” Opt. Quantum Electron. 48(1), 3 (2016). [CrossRef]
47. M. Hudlička, C. Lundström, D. A. Humphreys, and I. Fatadin, “BER estimation from EVM for QPSK and 16-QAM coherent optical systems,” in IEEE 6th International Conference on Photonics (ICP), 1–3 (2016).
48. “LTE; Evolved Universal Terrestrial Radio Access (E-UTRA); Base Station (BS) radio transmission and reception,” 3GPP TS 36.104 (2017).
49. S. Jia, X. Yu, H. Hu, J. Yu, T. Morioka, P. U. Jepsen, and L. K. Oxenløwe, “120 Gb/s Multi-Channel THz Wireless Transmission and THz Receiver Performance Analysis,” IEEE Photon. Technol. Lett. 29(3), 310–313 (2017). [CrossRef]
