Open Access
Add to BibSonomy
Abstract
Nyquist wavelength division multiplexing (WDM) using a 40 Gbit/s channel signal is applied to terahertz (THz)-wave communication with a view to increasing its spectral efficiency and capacity. A 2 × 40 Gbit/s Nyquist WDM signal in the THz-band is both generated and demultiplexed by the assistance of optical technology. The optical-domain demultiplexing is adopted mainly due to the difficulty in both the direct THz-wave-domain and the subsequent radio-frequency (RF)-domain demultiplexing. The received THz-wave Nyquist WDM signal at an antenna is divided into two signals, which are down-converted into RF signals with heterodyne detection utilizing different frequency local sinusoidal waves. Each signal in the RF-domain is again transformed into an optical signal through an optical modulator, and each 40 Gbit/s channel is extracted from one of the generated sidebands at each optical modulator output with an optical Nyquist filter. Bit error rates (BERs) on the order of 10−5 and 10−7 (below the first KP4 forward error correction threshold) were obtained for the demultiplexed two 40 Gbit/s channels. In addition to the explanation of the experimental setup and operating principle of the optical-domain demultiplexing method for the THz-wave Nyquist WDM signal, some experimental results are reported to show the validity of our proposed demultiplexing method and factors to limit the BERs.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Terahertz (THz)-wave communication is attractive because it can provide high-speed and short-reach wireless communication of more than 10 Gbit/s utilizing its wide bandwidth [1–4]. Its application areas include building out a fiber-to-THz radio bridge at a location where it is difficult to lay a fiber-optic cable, and laying an occasional wireless link when a fiber-optic cable gets disconnected in a time of disaster [1,2]. The research and development on the single-carrier transmission are vigorously carried out in the THz-band. In addition, when considering the prospective growth in the traffic, it is also significant to investigate multi-carrier communication in the THz-band [5–8], which includes spectrally efficient orthogonal frequency division multiplexing (OFDM) [9] and Nyquist wavelength division multiplexing (WDM) [10], with a view to increasing the capacity of the communication effectively. The symbol rate per channel is equal to the channel spacing in the OFDM and Nyquist WDM.
In this paper, THz-wave Nyquist WDM communication utilizing high-speed channels is investigated and reported. A 2 × 40 Gbit/s on-off keying (OOK) Nyquist WDM signal in the THz-band is generated by mixing an optical Nyquist WDM signal [11] with a local continuous-wave laser diode (LD) at a uni-traveling carrier photo-diode (UTC-PD)-based high-speed photo-mixer [12]. The received THz-wave signal at an antenna is divided into two signals for the subsequent heterodyne down-conversion into radio-frequency (RF) signals using two local sinusoidal waves with different frequencies. The two 40 Gbit/s channels in each down-converted Nyquist WDM signal switch places in the RF-domain. Each RF signal is again transformed into an optical signal through an optical modulator, and each 40 Gbit/s channel is extracted from one of the generated optical sidebands at each optical modulator output with an optical Nyquist filter. Thus, the Nyquist WDM channels are demultiplexed. This optical-domain demultiplexing at the receiver side is adopted due to the difficulty in both the direct THz-wave-domain and the subsequent RF-domain demultiplexing, and high functionality of the optical filter including center frequency tunability and an arbitrary passband shape. As we use high-speed, namely, a 40 Gbit/s channel signal, we cannot set an intermediate frequency (IF) of the RF signal to a sufficiently higher frequency than the channel bit rate due to the bandwidth limits of RF components. The used IF for a demultiplexed channel is only 25 GHz. Therefore, signal processing, namely, envelope detection of the 40 Gbit/s OOK signal in the RF-domain is not possible. While on the other hand, we can process the converted optical signal from the RF signal without difficulty because the optical signal has much higher carrier frequency than the bit rate.
We first explain the experimental setup and operating principle of the optical-domain demultiplexing method for the THz-wave Nyquist WDM signal. We then report experimental results. Measured bit error rates (BERs) were on the order of 10−5 and 10−7 for the demultiplexed two 40 Gbit/s channels, which were below the first KP4 forward error correction (FEC) threshold (2.2 × 10−4) [13]. In addition, from the experimental results, we explain that main factors to limit the BERs of demultiplexed channels are response difference versus a frequency of the UTC-PD, intersymbol interference (ISI) due to incomplete spectral characteristics of the demultiplexed channels, and crosstalk between Nyquist WDM channels.
We previously reported some spectrally efficient THz-wave communications utilizing Nyquist pulses and optical-domain reception signal processing [14–18]. In [14,15], a method for converting a THz-wave signal into an optical signal was proposed for the optical-domain processing, and its effectiveness was shown by carrying out high-speed and single-carrier communication (40 Gbit/s). We then realized first Nyquist WDM (2 × 40 Gbit/s) communication in the THz-band, to our knowledge, by applying the above-mentioned conversion method to optical-domain channel demultiplexing [16]. However, in [16], the receiver could not demultiplex plural Nyquist WDM channels at the same time, and a preliminary operating principle of the receiver was just demonstrated. In this paper, we report on a receiver configuration that can simultaneously process plural channels, and, in addition, show improved characteristics of demultiplexed channels. Portions of this work were already presented in [17]. However, compared to the partial report in [17], this paper provides substantial experimental results including detailed BERs measurement. In [18], we reported variable capacity THz-wave communication whose theme was different from this paper.
2. Experimental setup and demultiplexing method of THz-wave Nyquist WDM signal
Figures 1(a) and 1(b) show experimental setups for photonics-assisted generation and demultiplexing of a THz-wave Nyquist WDM signal, respectively. Schematic diagrams of signal spectral change are inset in Fig. 1. In Fig. 1(a), two lightwaves emitted from LDs 1 (wavelength: 1552.16 nm) and 2 (wavelength: 1552.48 nm) were modulated with different non-return-to-zero (NRZ) 40 Gbit/s OOK data generated at synchronized two pulse pattern generators (PPGs) 1 and 2, respectively. All the used LDs in the experiments were distributed-feedback-type semiconductor lasers. The pseudo-random bit sequence (PRBS) of both data was 27-1. We used this PRBS length because our investigation was at the preliminary stage. We inserted an optical fiber with the length of about 0.3 m, which functioned as a delay line, after a Mach-Zehnder interferometer (MZI)-type optical intensity modulator (IM) 2 in order to minimize the correlation between the two synchronized sequences. The fiber length approximately corresponded to half the cycle length of the sequence. The spectra of produced two 40 Gbit/s OOK signals were individually shaped with root raised-cosine optical Nyquist filters 1 and 2 with a view to generating Nyquist time division multiplexing signals for channels (CHs) 1 and 2, respectively. Each filter was composed of a bulk-type grating, and its measured roll-off factor α was 0.3. A 2 × 40 Gbit/s optical Nyquist WDM signal was generated by combining the two filtered signals with a 3 dB fiber coupler 1. The generated optical Nyquist WDM signal was mixed with a continuous lightwave from a local LD3 (wavelength: 1550.00 nm) by using a UTC-PD [12] so as to produce a 2 × 40 Gbit/s Nyquist WDM signal in the THz-band. The difference between the LD3 frequency fLD3 and the center frequency fWDM of the optical Nyquist WDM signal was 290 GHz, which corresponded to the center frequency fT of the Nyquist WDM signal in the THz-band. The produced THz-wave signal was radiated to a 0.1 m free-space link through a horn antenna 1. This wireless link length was set in order to validate the operational principle and acquire preliminary data.
Fig. 1. Experimental setups for THz-wave Nyquist WDM communication assisted by photonics technology. Setups for (a) generating and (b) demultiplexing THz-wave Nyquist WDM signal.
The received signal at a horn antenna 2 in Fig. 1(b) was split into two signals, which were mixed with local sinusoidal waves with different frequencies of 245 GHz and 335 GHz to obtain down-converted RF signals with the same center IF fRF of 45 GHz. The gain of both antennas was 27 dBi. While each RF signal had two channels with the same carrier frequencies (lower carrier frequency fL: 25 GHz, higher carrier frequency fH: 65 GHz), the channel numbers in each RF signal switched places each other in the frequency-domain due to the use of the local sinusoidal waves with different frequencies. The two RF signals were linearly amplified and again transformed into optical signals through IMs 3 and 4. An LD4 with a wavelength of 1547.31 nm supplied us with input lightwaves into the two modulators. The operating bandwidths of each mixer and linear amplifier were 40 GHz and 65 GHz, respectively. The optical bandwidth and half-wavelength voltage of each modulator (IM3 or 4) were around 30.0 GHz and 3.0 V at 40 Gbit/s, respectively. The bias of each modulator was set to a null point. Under this bias setting condition, the RF signal entered into the modulator, whose bandwidth, and amplitude difference between the marked and unmarked levels are less than or equal to double the IF fRF, and 1.74 rad in phase equivalent, respectively, can be converted into the optical signal with small degradation (with an error of less than or equal to 10%) [14]. In the modulator, the optical output amplitude is expressed as a cosine function of half the RF input phase shift [14]. Therefore, the relation between the output and input amplitude shows high linearity when the modulator bias is set to (2m+1)π (m: whole number) and the input amplitude into the modulator is small.
Each modulator generated two main sidebands around the optical carrier frequency f0 (equivalent to the frequency of LD4), whose center frequency difference was 2fRF (90 GHz), under the null bias setting condition [14]. While each lower-frequency carrier channel (frequency: fL) in the RF-domain was converted into the optical signal without distinct attenuation, each higher-frequency carrier RF channel (frequency: fH) considerably attenuated in the optical-domain due to the bandwidth limits of used RF components, and IM3 or 4. The CH1 (after IM3) or 2 (after IM4) was extracted from the longer-wavelength optical sideband with a programmable optical filter 1 or 2, respectively. Each optical filter consisted of a bulk grating and a liquid crystal-type spatial light modulator [19], and was set so that the shape of its output signal spectrum became raised-cosine with the roll-off factor of 0.3. Each filtered optical signal was evaluated. The amplitude of marked CH1 or 2 signal in the RF-domain, which was used to drive IM3 or 4, respectively, was about 1.14 rad in phase equivalent.
In Fig. 1(b), when two laser diodes with different wavelengths are independently assigned to IMs 3 and 4, and two optical signals after the programmable optical filters 1 and 2 are combined with a fiber coupler, the THz to optical conversion relating to the Nyquist WDM signal is achieved. As the setup in Fig. 1(b) does not contain a coherent detection scheme, it can just process amplitude modulation signals.
3. Experimental results
Figure 2 shows transmittance of the root optical Nyquist filter 1 in Fig. 1(a), which was measured with an amplified spontaneous emission light source. The filter loss was 5.1 dB. Figure 3 indicates a frequency response of the UTC-PD. The difference between the maximum and minimum responses was 4.7 dB, 2.1 dB, or 1.1 dB for 250 to 290 GHz (used for CH1), 290 to 330 GHz (used for CH2), or 280 to 320 GHz, respectively. The large degradation of the frequency response in the CH1 band was caused because a cut-off frequency (173.5 GHz) of the UTC-PD output waveguide (WR-3) was adjacent to the CH1 band. The 280 to 320 GHz band was used for the single-channel communication in [14,15] and this manuscript. In this band, the frequency response had the peak at around the center frequency and smoothly decreased with distance from the center frequency without showing distinct degradation of the frequency response. The band of more than 330 GHz, which showed lower response difference than bands used for CHs 1 and 2, could not be utilized in our Nyquist WDM communication due to the operational frequency range limit (220 to 330 GHz) of the mixer in the receiver. Figure 4 shows measured spectra of a 2 × 40 Gbit/s optical Nyquist WDM signal and a local LD3, which were input into the UTC-PD in Fig. 1(a) for producing a THz-wave Nyquist WDM signal. The intensity of the optical signal and local light was both adjusted to 7.5 dBm with a view to generating the THz-wave signal efficiently. The intensity of the generated THz-wave signal from the UTC-PD was -12.2 dBm. Figures 5(a) and 5(b) show spectra measured before the programmable optical filter 1 in Fig. 1(b) when only CH1 and both of the two channels were used for the THz-wave communication, respectively. Green signs at f0±fL and f0±3fL in Fig. 5(a) correspond to the main sidebands and its second-order harmonics, respectively. Red signs at f0±2fL denote unanticipated signals, which appeared because the IM3 bias could not be set to the null point with high precision. In Fig. 5(b), two frequency components indicated by blue lines came into existence due to the combination of the above-mentioned unanticipated signals at f0±2fL and the CH2 signals at f0±fH. Crosstalk components between the two channels, which were induced at the IM3, were small and were not observed in Fig. 5(b).
Fig. 4. Measured spectra of 2 × 40 Gbit/s optical Nyquist WDM signal and local LD3, which were input into UTC-PD for producing THz-wave Nyquist WDM signal.
Fig. 5. Spectra measured before programmable optical filter 1 in Fig. 1(b) when (a) only CH1 and (b) both of two channels were used.
Figures 6(a) or 6(b) shows BERs relating to the CH1 or 2, respectively. These figures also include BERs of the single-channel communication, which were measured after the intensity modulator IM1 or 2, after the root optical Nyquist filter 1 or 2, and at the final stage in Fig. 1. The BERs of the demultiplexed channels from the Nyquist WDM signals were measured after EDFA1 in the transmitter and at the final stage. In the transmitter side, the channels were also demultiplexed with the same programmable optical filter as Fig. 1(b). The lowest BERs at the final stage in the single-channel communication were 2.4 × 10−7 and 1.4 × 10−9 for CHs 1 and 2, respectively. The obtained lowest BERs at the final stage in the Nyquist WDM communication were 3.5 × 10−5 and 4.8 × 10−7 for CHs 1 and 2, respectively. In the Nyquist WDM communication, the BERs below the first KP4 FEC threshold (2.2 × 10−4) [13] were achieved for both channels at the final stage. We could achieve the BERs of 10−12 in other cases. Figure 7 indicates measured BERs of the single-channel communication using the 280 to 320 GHz band, which showed the most suitable UTC-PD response for the communication. As we used different PDs in the receiver from that in [14,15], we carried out the single-channel communication again using the new PD. The obtained BERs showed the similar characteristics as [14,15], although the sensitivity was different from that in [14,15] due to the use of the new PD. Figures 8 and 9 show measured spectra and eye diagrams of the demultiplexed two channels at the final stage, respectively, when the lowest BERs were attained. The spectra in Fig. 8 show residual aperture distortion (umbonal spectral shape around the center) in addition to deterioration at the longer-wavelength range, namely, they do not have complete raised-cosine characteristics. The aperture distortion means the difference between the transfer function of the Nyquist filter (ideal raised-cosine characteristics) and the spectrum of the shaped NRZ signal with the Nyquist filter. The bandwidths of the spectra in Figs. 8(a) and 8(b) were 45.2 GHz and 41.2 GHz, respectively. The bandwidths deviated from ideal 40 GHz due to the setting limit of the programmable optical filters in Fig. 1(b). We discuss the effect of this deviation on the BER characteristics later in this section. All the signals at the final stage in Figs. 6 – 9 were extracted from the longer-wavelength optical sidebands generated at the IMs 3 and 4 with the programmable optical filters.
Fig. 8. Measured spectra of demultiplexed (a) CH1 and (b) CH2 signals at final stage when the lowest BERs were attained.
Fig. 9. Measured eye diagrams of demultiplexed (a) CH1 and (b) CH2 signals at final stage when the lowest BERs were attained.
Tables 1 and 2 summarize power penalties in Fig. 6 at the BER of 2.2 × 10−4 (KP4 FEC threshold) and the lowest BERs obtained at the final stage in the Nyquist WDM communication (CH1: 3.5 × 10−5, CH2: 4.8 × 10−7), respectively, which were compared to the BERs of the original NRZ signals. Table 3 summarizes power penalties in Fig. 7 at the BERs of 2.2 × 10−4, 3.5 × 10−5, 4.8 × 10−7, and 1.0 × 10−9. These penalties were also compared to the BERs of the original NRZ signal. The penalties, which arose after the root optical Nyquist filters in the transmitter, were caused by ISI, which originated from generated incomplete Nyquist pulses due to the aperture distortion.
Table 1. Power penalties in Fig. 6 at BER of 2.2 × 10−4
Table 2. Power penalties in Fig. 6 at the lowest BERs obtained at final stage in Nyquist WDM communication
Table 3. Power penalties in Fig. 7
We evaluate the factors to limit the BERs in the single-channel communication. In the single-channel communication of CH 1 or 2, large power penalties took place at the final stage. These characteristics were not observed at the single-channel communication in Fig. 7 using the 280 to 320 GHz band. The major factor of the large penalties was attributed to the frequency response degradation of the UTC-PD in the bands used for the CHs 1 and 2 communication, which is shown in Fig. 3. In addition, at the final stage in the single-channel communication, the obtained CH1 BER was worse than the CH2 because the frequency response in the CH1 band became more deteriorated than the CH2 band. In addition, the additional ISI due to the non-ideal spectral shapes of the signals at the final stage increased the power penalties to some extent. The characteristics incompleteness of the demultiplexed spectra primarily originated from discrepancy between set and obtained characteristics of the programmable optical filters in Fig. 1(b), which was caused by factors including frequency setting resolution (about 2.5 GHz), bandwidth setting resolution (about 5.0 GHz), and attenuation setting resolution (about 1.0 dB) of the filters [19]. It was difficult, from the results in Figs. 6 and 7, and Tables 1–3, to discriminate the effect of the UTC-PD frequency response degradation and the additional ISI on the power penalties exactly. However, we derived approximate total power penalties due to both of the two factors by subtracting the power penalties between after the filter at the transmitter and at the final stage in Fig. 7 from the penalties between after the filter and at the final stage in Fig. 6. The estimated penalties were 3.0 dB and 1.0 dB for CHs 1 and 2 at the KP4 FEC threshold, respectively. Also, at the lowest final stage BERs attained in the Nyquist WDM communication, the estimated penalties were 3.6 dB and 1.8 dB for CHs 1 and 2, respectively.
Next, we evaluate the factors to limit the BERs in the Nyquist WDM communication. In the Nyquist WDM communication, as shown in Fig. 6, and Tables 1 and 2, power penalties exist between the single channel and the demultiplexed channel from the Nyquist WDM signal at both the transmitter and receiver sides. The penalties between after the optical Nyquist filter and after EDFA1 in the transmitter were 0.6 dB for both channels at the KP4 FEC threshold. The penalties were 0.7 dB and 0.6 dB for CHs 1 and 2, respectively, at the lowest final stage BERs attained in the Nyquist WDM communication. These penalties were mainly attributed to the original crosstalk between the channels. The crosstalk between the Nyquist WDM channels is inescapable except when the roll-off factors of the channel spectra are unrealistically zero [20]. In the optical Nyquist WDM communication in [20], the Q factor degraded with the increase of the filter roll-off factor when it was larger than 0.2 (0.3 in our experiments). On the other hand, penalties between the final stage signals in the single-channel and Nyquist WDM communications were 0.9 dB and 0.6 dB for CHs 1 and 2, respectively, at the KP4 FEC threshold, and the penalties were 3.4 dB and 2.2 dB for CHs 1 and 2, respectively, at the lowest final stage BERs in the Nyquist WDM communication. These penalties included the original and additional crosstalk between the channels. Therefore, the penalties due to the additional crosstalk were estimated to be 2.7 dB and 1.6 dB for CHs 1 and 2, respectively, at the lowest final stage BERs attained in the Nyquist WDM communication. The larger penalties were observed at the receiver than the transmitter when demultiplexing the Nyquist WDM signals because the spectral shape of the demultiplexed channel became more deteriorated at the receiver than the transmitter due to the frequency, bandwidth, and attenuation setting resolution limits of the programmable filters when compensating for the UTC-PD frequency response degradation. The bandwidths of the demultiplexed CHs 1 and 2 were 45.2 GHz and 41.2 GHz, respectively. The larger bandwidths than ideal 40 GHz were the origin of the additional crosstalk between the Nyquist WDM channels at the receiver, and the CH1 penalty increase due to the additional crosstalk was larger than CH2 because of the larger bandwidth of the demultiplexed channel.
Other factors of the BERs degradation at the final stage were channel crosstalk increment due to frequency fluctuation of used LDs 1 and 2 in Fig. 1(a) (about 2.5 GHz) and crosstalk between channels caused by non-linear responses of the IM3 or 4 in Fig. 1(b). The spacing narrowing between the LDs frequencies due to their fluctuation increases the crosstalk between the Nyquist WDM channels [21]. In [21], when the channel spacing between the optical Nyquist channels was narrowed by only 5% compared to the ideal spacing (equal to each channel symbol rate), the serious error floor was observed. However, from the measured results, we did not observe significant penalty caused by the above-mentioned two factors.
From our experimental results, we concluded that, under the condition of used experimental parameters, the frequency response degradation of the UTC-PD, the ISI, and the crosstalk between the Nyquist WDM channels were factors for determining the BER limits at the receiver.
4. Conclusion
2 × 40 Gbit/s Nyquist WDM communication in the THz-band was reported, whose signal generation and reception were both assisted by optical technology. The THz-wave Nyquist WDM signal was produced by mixing a 2 × 40 Gbit/s optical Nyquist WDM signal with a local lightwave at a UTC-PD. A received THz-wave Nyquist WDM signal at an antenna was split into two signals, which were both down-converted into RF signals by using two local sinusoidal waves with different frequencies. Each RF signal was transformed into an optical signal with an optical modulator, and one of generated two main sidebands at each optical modulator output was extracted with a programmable optical filter with a view to demultiplexing Nyquist WDM channels. This optical-domain demultiplexing method was adopted because of the difficulty in both the direct THz-domain and the subsequent RF-domain demultiplexing of channels, and high functionality of the optical filter including flexible center frequency tunability and an arbitrary passband shape. After explaining the experimental setup and the operating principle of the optical-domain demultiplexing, we reported experimental results on the THz-wave Nyquist WDM communication. Bit error rates on the order of 10−5 and 10−7 were obtained for the demultiplexed two 40 Gbit/s channels. These error rates were below the first KP4 forward error correction threshold (2.2 × 10−4). We also confirmed that, from our experiments, the frequency response degradation of the UTC-PD, the ISI, and the crosstalk between the Nyquist WDM channels were factors to limit bit error rates at the receiver. We think that the demonstrated scheme is one of candidate technologies to realize high-speed THz-wave communication with high spectral efficiency.
Funding
Support Center for Advanced Telecommunications Technology Research Foundation; Japan Society for the Promotion of Science (19H02144).
Disclosures
The authors declare no conflicts of interest.
References
1. T. Nagatsuma, G. Ducournau, and C. C. Renaud, “Advances in terahertz communications accelerated by photonics,” Nat. Photonics 10(6), 371–379 (2016). [CrossRef]
2. S. Koenig, D.-L. Diaz, J. Antes, F. Boes, R. Henneberger, A. Leuther, A. Tessmann, R. Schmogrow, R. 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,” Nat. Photonics 7(12), 977–981 (2013). [CrossRef]
3. A. Kanno, P. T. Dat, N. Sekine, I. Hosako, T. Kawanishi, Y. Yoshida, and K.-I. Kitayama, “High-speed coherent transmission using advanced in terahertz bands,” Trans. Inst. Electron., Inf. Commun. Eng., Sect. E 98(12), 1071–1080 (2015). [CrossRef]
4. C. Wang, J. Yu, X. Li, P. Gou, and W. Zhou, “Fiber-THz-fiber link for THz signal transmission,” IEEE Photonics J. 10(2), 1–6 (2018). [CrossRef]
5. H. Shams, T. Shao, M. J. Fice, P. M. Anandarajah, C. C. Renaud, F. V. Dijk, L. P. Barry, and A. J. Seeds, “100 Gb/s multicarrier THz wireless transmission system with high frequency stability based on a gain-switched laser comb source,” IEEE Photonics J. 7(3), 1–11 (2015). [CrossRef]
6. X. Yu, S. Jia, H. Hu, M. Galili, T. Morioka, P. U. Jepsen, and L. K. Oxenløwe, “160 Gbit/s photonics wireless transmission in the 300-500 GHz band,” APL Photonics 1(8), 081301 (2016). [CrossRef]
7. K. Takiguchi, “Sub-carrier channel demultiplexing of terahertz-wave OFDM signal assisted by optical technology,” Electron. Lett. 53(16), 1123–1125 (2017). [CrossRef]
8. N. Oshima, K. Hashimoto, S. Suzuki, and M. Asada, “Terahertz wireless data transmission with frequency and polarization division multiplexing using resonant-tunneling-diode oscillators,” IEEE Trans. Terahertz Sci. Technol. 7(5), 593–598 (2017). [CrossRef]
9. R. W. Chang and R. A. Gabby, “A theoretical study of performance of an orthogonal multiplexing data transmission scheme,” IRE Trans. Commun. Syst. 16(4), 529–540 (1968). [CrossRef]
10. H. Nyquist, “Certain topics in telegraph transmission theory,” Trans. Am. Inst. Electr. Eng. 47(2), 617–644 (1928). [CrossRef]
11. G. Bosco, A. Carena, V. Curri, P. Poggiolini, and F. Forghieri, “Performance limits of Nyquist-WDM and CO-OFDM in high-speed PM-QPSK systems,” IEEE Photonics Technol. Lett. 22(15), 1129–1131 (2010). [CrossRef]
12. T. Ishibashi, Y. Muramoto, T. Yoshimatsu, and H. Ito, “Unitraveling-carrier photodiodes for terahertz application,” IEEE J. Sel. Top. Quantum Electron. 20(6), 79–88 (2014). [CrossRef]
13. N.-P. Diamantopoulos, W. Kobayashi, H. Nishi, K. Takeda, T. Kakitsuka, and S. Matsuo, “Amplifierless PAM-4/PAM-8 transmissions using a directly modulated laser for optical data-center interconnects,” Opt. Lett. 44(1), 9–12 (2019). [CrossRef]
14. K. Takiguchi, “Method for converting high-speed and spectrally efficient terahertz-wave signal into optical signal,” Opt. Express 27(5), 6598–6606 (2019). [CrossRef]
15. K. Takiguchi, “Seamless connection between high-speed THz-wave and optical signals with high spectral efficiency,” in Proc. Conference on Lasers and Electro-Optics (CLEO) (OSA, 2018), paper STu3D.3, San Jose.
16. K. Takiguchi, “Nyquist wavelength division multiplexing communication in terahertz-band assisted by optical technology,” in Proc. Photonics West (SPIE, 2019), paper 10917–10987, San Francisco.
17. K. Takiguchi, “Spectral-efficient and high-speed THz-wave communication using 40 Gsymbol/s channel-based Nyquist WDM,” in Proc. CLEO (OSA, 2019), paper SW4F.7, San Jose.
18. K. Takiguchi and N. Nishio, “Flexible data rate THz-wave communication using Nyquist pulses and optical-domain reception signal processing,” in Proc. Optical Fiber Communication Conference (OFC) (OSA, 2020), paper M4I.5, San Diego.
19. M. A. F. Roelens, S. Frisken, J. A. Bolger, D. Abakoumov, G. Baxter, S. Poole, and B. J. Eggleton, “Dispersion trimming in a reconfigurable wavelength selective switch,” J. Lightwave Technol. 26(1), 73–78 (2008). [CrossRef]
20. S. Shimizu, G. Cincotti, and N. Wada, “Demonstration of no-guard-interval 6 × 25 Gbit/s all-optical Nyquist WDM system for flexible optical networks by using CS-RZ signal and optical Nyquist filtering,” in Proc. Optoelectronics and Communications Conference/Australian Conference on Optical Fiber Technology (OECC/ACOFT) (Engineers Australia, 2014), paper MO1B-1, Melbourne.
21. K. Igarashi, Y. Mori, K. Katoh, and K. Kikuchi, “Bit-error rate performance of Nyquist wavelength-division multiplexed quadrature phase-shift keying optical signals,” in Proc. OFC (OSA, 2011), paper OMR6, Los Angeles.
