0.34-THz Wireless Link Based on High-Order Modulation for Future Wireless Local Area Network Applications

Transcription

1 IEEE TRANSACTIONS ON TERAHERTZ SCIENCE AND TECHNOLOGY, VOL. 4, NO. 1, JANUARY THz Wireless Link Based on High-Order Modulation for Future Wireless Local Area Network Applications Cheng Wang, Bin Lu, Changxing Lin, Qi Chen, Li Miao, Xianjin Deng, and Jian Zhang Abstract This paper presents a 0.34-THz wireless link for future wireless local area networks (WLANs), which is based on high order 16-quadrature amplitude modulation (16QAM). The system adopts super heterodyne transceivers and parallel digital signal-processing techniques. The 0.34-THz transceiver consists of a 0.34-THz subharmonic mixer, a 0.34-THz waveguide H-ladder bandpass filter, and a 0.17-THz multiplier chain. Two 0.34-THz Cassegrain antennas with 48.4-dBi gain have been developed to extend the transmission distance. Based on a 32-way parallel signal processing, we have successfully realized the 3-Gb/s, 16QAM real-time modulator and demodulator. The measured data indicate that the lowest bit error rate of the 0.34-THz, 3-Gb/s data link is over a 50.0-m line-of-sight channel. The maximum received energy per bit to noise power spectral density ratio (E N ) is 23.8 db, while the output power of transmitter is 17.5 dbm and the noise temperature of receiver is 5227 K. In addition, this paper presents a 0.34-THz WLAN prototype based on IEEE b/g protocol. The WLAN prototype, which consists of an access point and two terminal nodes, achieves a transmission data rate of Mb/s over 1.15 m by using rectangular horn antennas. Index Terms Millimeter-wave communication, quadrature amplitude modulation (QAM), wireless local area network (WLAN). I. INTRODUCTION T HE demand for high-data-rate wireless communication is increasing synchronously with the development of military and consumer electronics applications. The analog carrier of digital data link is extending from the microwave band to millimeter and terahertz bands. Terahertz communication is attractive in high-speed satellite data link, ultrashort-range interface of portable device, instantaneous or intermittent link, and wireless local area networks (WLANs). In the last decade, the academia has been striving towards optimized electronic and photonic devices, effective system architectures and high-performance protocols for THz communication. Significant progress has been achieved in recent publications. Manuscript received April 28, 2013; revised July 04, 2013, August 22, 2013, and October 24, 2013; accepted November 15, Date of publication January 02, 2014; date of current version January 17, This work was supported by the Terahertz Research Center, Institute of Electronic Engineering, China Academy of Engineering Physics (CAEP-IEE). The authors are with the Terahertz Research Center, Institute of Electronic Engineering, China Academy of Engineering Physics, Mian Yang, , China ( c-w04@163.com). Color versions of one or more of the figures in this paper are available online at Digital Object Identifier /TTHZ Based on unitraveling carrier photodiode (UTC-PD) and InP high-electron mobility transistors (InP HEMTs), respectively, NTT Microsystem Integration Laboratories in Japan has developed two 120-GHz wireless links [1], whose data rate is 10 Gb/s and maximum transmission distance is 5.8 km [2]. NTT recently realized a 300-GHz 24-Gb/s communication link over laboratory distance also by UTC-PD [3]. Based on a 50-nm InP metamorphic high-electron mobility transistor (InP mhemt), a 220-GHz transceiver has been developed by the Fraunhofer Institute for Applied Solid-State Physics (IAF) in Germany [4], which realized 10-m transmissions of 25-Gb/s on off-keying (OOK) signal and 14-Mb/s 256 quadrature amplitude modulation (QAM) signal [5]. Xi an University has developed a 135-GHz link with 10-Gb/s data rate based on a0.13- m MOSFET switch ASK modulator [6]. Bell Labs has transmitted 2.5-Gb/s signal in the 625-GHz band using duobinary baseband modulation [7]. Most THz communication prototypes give up high-order modulation due to challenges on implementation, including complexity and speed limitation. ASK or OOK schemes seem to be more reasonable. They can achieve tens ofgb/seasily through analog modulators. However, in this paper, we propose a system framework based on super heterodyne Schottky barrier diodes transceiver and parallel digital signal processing (DSP) techniques for 0.34-THz 16QAM transmission. It brings the advantages of: 1) higher spectrum efficiency [8]; 2) the potential of higher output power from vacuum and solid-state amplifier [9] [12]; and 3) higher channel distortion tolerance from digital equalization algorithms. A 0.34-THz wireless link has been developed according to this framework. This link has successfully transmitted a 3-Gb/s, 16QAM signal over a 50.0-m line-of-sight (LOS) channel with 17.5 dbm output power and 5227 K received noise temperature. The maximum energy per bit to noise power spectral density ratio E N is 23.8 db. The minimum bit error rate (BER) is The minimum E N is 13.8 db when 10. Short-range communication is essential for exploiting the terahertz spectrum resources. An overview article paid much attention to short-range high speed communication above 0.3 THz [13]. Furthermore, theoretical and experimental researches on propagation of THz wave in room environment have already been performed [14], [15]. IEEE wireless personal area network (WPAN) workgroup constituted IEEE THz interest group (IG-THz) to promote high-speed protocols for THz communication. In this paper, we have developed a 0.34-THz WLAN prototype based on the IEEE X 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See for more information.

2 76 IEEE TRANSACTIONS ON TERAHERTZ SCIENCE AND TECHNOLOGY, VOL. 4, NO. 1, JANUARY 2014 Fig. 2. Receiving front end of the 0.34-THz wireless link. Fig. 1. Framework of the 0.34-THz wireless link b/g protocol to validate the character of future THz WLAN. It has realized Mb/s real-time data transmission over 1.15 m, among one access point (AP) and two terminal nodes. To the best of the authors knowledge, this is the first THz WLAN prototype. II. SYSTEM ARCHITECTURE OF HIGH-SPEED WIRELESS DATA LINK AND WLAN A. 3-Gb/s 16QAM Wireless Link Fig. 1 presents the framework of the 0.34-THz 3-Gb/s wireless link. The main reason for using a heterodyne transceiver rather than a zero-if transceiver in the current system is that the latter usually has inferior performance. For instance, flicker noise in zero-if transceiver will interfere with the baseband signal more significantly when compared with the heterodyne transceiver. It is also difficult to eliminate the dc offset and second-order (IM2) and third-order inter-modulations (IM3) in the zero-if transceiver. 16QAM modulation could improve the spectrum efficiency and reduce the occupied bandwidth. It brings many advantages, given as follows. 1) It enlarges the channel capacity. Frequency-division multiplexing (FDM) is commonly adopted to accommodate multiple users in a WLAN system. A 10-Gb/s, 16QAM subband with 3.6-GHz physical bandwidth could support a 100-Gb/s WLAN system (ten subbands) within the 40-GHz bandwidth. However, a system adopting ASK modulation whose bandwidth of a 10-Gb/s signal is 17 GHz can only support a 23-Gb/s WLAN within the same 40-GHz bandwidth. 2) Compared with other high-order modulation schemes, the 16QAM scheme has higher spectrum efficiency than 8PSK, but similar E N versus BER performance. 3) Semiconductor devices with narrow relative bandwidth and higher performance are more accessible and economical. In addition, the measured phase noise in Section V-A indicates that phase noise will not be a significant factor to inhibit the 16QAM demodulation in the 0.34-THz band. However, the lower output power, imperfect signal quality (spectrum purity or phase noise), and higher receiving noise still make the realization of 16QAM scheme in the 0.34-THz band a remarkable challenge. As presented in Fig. 1, the 3-Gb/s modulator based on a field-programmable gate array (FPGA) and 3-Gs/s digital analog converter (DAC) modulates the primal binary signal. The bandwidth of modulated signal could be calculated using where is the data rate, is the rolloff factor, and is the modulation order. The rolloff factor is set to 0.4 in current modulation. The occupied bandwidth of the modulated 3-Gb/s, 16QAM signal is 1.05 GHz. The center frequency is 750 MHz. The modulated signal is then converted to inter-frequency (IF, GHz) by the first-stage converter and finally reaches the 0.34-THz band by the THz transmitter. The ultimate band of operation is GHz GHz. Fig. 2 shows the 0.34-THz receiving front end. The transmitter and receiver depend on the subharmonic mixer (SHM) to perform frequency mixing, which is pumped by the -band 8 multiplier chain. When the SHM is used as an upconverter in the transmitter, the maximum output power is 14.4 dbm, as shown in Fig. 3. The output power is relatively low, but still sufficient for our communication experiment. The variation of output power is 2.3 db in the GHz band. The 1-dB compression point in GHz is 16.6 dbm. The output power of the modulated 3-Gb/s signal at the 1-dB compression point is 17.5 dbm. A solid-state power amplifier (SSPA) can further push the power level to 10 dbm in the future [10]. Compared with fundamental mixer, the frequency of local oscillating frequency (LO) for the SHM is only half of the input THz wave, which eases the difficulty in generating sufficient LO power. The -band 8 multiplier chain consists of three doublers, which operate in the GHz, GHz, and (1)

3 WANG et al.: HIGH-ORDER MODULATION FOR FUTURE WLAN APPLICATIONS 77 TABLE I BUDGET OF 0.34-THZ WIRELESS COMMUNICATION LINK Fig. 3. Output power and downconversion gain of 0.34-THz transceiver GHz bands, respectively. We utilize medium power amplifiers (MPAs) in the GHz and GHz bands between multipliers to provide 20 dbm of driving power. The multiplier chain is driven by a 20.3-GHz dielectric resonance oscillator (DRO) and provides an output of 9.39 dbm in GHz. We use a -band variable attenuator to adjust the LO power for optimizing the conversion loss of SHM. A rectangular waveguide H-ladder bandpass filter is connected between the SHM and antenna to reject the low sideband of the upconverted signal. A high directional Cassegrain antenna is finally connected in front of the transmitter and receiver. The received THz signal is also detected by the SHM. The conversion gain of the SHM is shown in Fig. 3, which is db to db within GHz. The corresponding IF range is GHz. The SHM has perfect noise temperature at room temperature, but also brings significant conversion loss. The SSB conversion loss shown in Fig. 3 is approximately 3 db higher than published data of similar SHMs. The explanation is that common SHMs usually work in low IF ( 100 MHz) and their conversion loss is measured through black body radiation source. In current systems, the SHM works in high IF frequency GHz and is measured by the classical RF method. However, the measured data under the same setting show no significant difference in noise temperature or conversion loss between the two kinds of SHMs. Due to the high conversion loss of SHM, the IF coaxial cable and IF low-noise amplifier (LNA, GHz, 30 db, 1.6 db) contribute remarkable noise to the receiver. The total receiver noise temperature is calculated by where,,and are the noise temperature of SHM, IF coaxial cable, and LNA, respectively. According to the -factor measurement using a black-body radiation source, the total DSB noise temperature of the receiver is 5277 K. The noise temperature introduced by the IF coaxial cable 1.0 db and the LNA are, respectively, 688 K and 1490 K. The SHM contributes 3099 K to the front-end. (2) a. The loss introduced by atmosphere is estimated to be 0.5 db. TheIFisrecoveredto750MHzagainbyafirst-stage converter in the receiving front end. The 3-Gb/s, 16QAM demodulator resembles the modulator in the hardware structure, which samples the IF with a 3-Gs/s analog digital converter (ADC). The demodulator applies a 32-way parallel demodulation algorithm based on the frequency-domain implementation of the matched filter and timing phase correction. We proposed innovative timing synchronization, channel equalization, and carrier recovery algorithms and then validated them in the hardware platform. Coding and decoding have been performed to further reduce the BER. Section IV provides details of the modulator and demodulator. Table I shows the link budget. The loss introduced by atmosphere attenuation in the 0.34-THz band is estimated to be 10 db/km. As a result, the 50.0-m transmission will suffer from 0.5 db loss. The calculated E N is 25.4 db, which exceeds the criteria of general digital demodulation. B. WLAN Based on IEEE b/g Short-range WLAN or WPAN is promising for future THz communication. Utilizing sufficient spectrum resources, 0.3-THz band could easily support Gb/s WLAN. The atmosphere attenuation is inconspicuous in room environment. The III V compound semiconductor integrated circuits have achieved 75-mW output power in the 220-GHz band [11]. The output power of IEEE b/g standard WLAN is 100 mw in 2.4 GHz. It is feasible to get comparable power in the 0.34-THz band in the future. However, two backdrops are still inevitable: First, the IEEE b/g WLAN occupies bandwidth 100 MHz. Nevertheless, THz WLAN will occupy bandwidth of GHz, which requires db higher output power using similar architecture. Second, the effective antenna aperture decreases with the increase of carrier frequency as in where is the directivity of antenna and is the wavelength. It can be predicted that, when the carrier frequency increases (3)

4 78 IEEE TRANSACTIONS ON TERAHERTZ SCIENCE AND TECHNOLOGY, VOL. 4, NO. 1, JANUARY 2014 Fig. 4. Architecture of 0.34-THz WLAN prototype. Fig. 7. Circuit of the 0.34-THz SHM based on SBDs. III. CRITICAL COMPONENTS FOR 0.34-THZ TRANSCEIVER Fig. 5. Fig. 6. Schematic of 0.34-THz WLAN node THz WLAN node and detail of the diplex transceiver. from the -band to 0.34 THz, the effective aperture decreases 40 db. Therefore, people must use high-gain antennas in a THz WLAN. Mechanical tunable or phase-array antennas may be possible candidates for this. Fig. 4 shows the architecture of the 0.34-THz WLAN prototype in this paper, which consists of one AP and two terminal nodes. The AP and nodes adopt 25-dBi wideband rectangular horns with a dB main lobe beam width. The nodes A and B exchange data by relaying AP rather than a direct wireless connection. Fig. 5 gives the schematic of 0.34-THz WLAN node. Considering that high-speed WLAN protocol is unavailable and incompatible for the current prototype, the MAC layer and partial physical layer are established through a commercial IEEE b/g wireless module, which operates in GHz. The 0.34-THz front-end is designed to work at diplex mode. A duplexer splits the RF signal from the commercial WLAN module to the 0.34-THz transceiver. The up-converting and receiving SHMs are pumped by LO signal fromthesame20.3-ghzdro.the isolation between upward and downward links is provided by the directivity of the two parallel 25-dBi horn antennas. Fig. 6 is the integrated 0.34-THz WLAN node and detail of the diplex transceiver. According to theoretical calculation, the transceiver could achieve E N of 15.9 db in 3.5-m distance. A THz SHM A0.34-THzSHMactsassignal generator and detector in the transceiver. The circuit of the SHM is shown in Fig. 7. The SHM is based on quartz substrate microstrip line, which is embedded in a metal waveguide cavity. The thickness of the substrate is 50 m and the width is 250 m. The Au layer, with 4- m thickness, acts as conductor of the strip line. The circuit consists of two individual substrates. Anti-parallel Schottky barrier diodes (SBDs) provide current voltage (I V) nonlinearity for the SHM, which are bonded on the main quartz substrate. The LO signal provided by a 8 multiplier chain is coupled to the main substrate by a -band waveguide probe. The received RF signal is also coupled by a waveguide probe and transferred to diode. An RF rejection LPF prevents the propagation of RF signal towards LO and IF ports. The IF signals are filtered by a seventh-order high low impedance filter in the auxiliary quartz substrate. A 18- m gold wire connects the main and auxiliary substrates. The pattern of gold wire is carefully optimized to introduce sufficient inductance to isolate the LO and IF signal. Another 18- m gold wire is soldered near the RF probe to provide dc grounding for SBDs. The adopted circuit structure brings the advantage of broad RF and IF bandwidth. The optimization of the SHM is based on a precise 3-D model of Schottky barrier diodes (SBDs). The parameters of SBDs are as follows: total capacitance 7.5to10.5fF;forward turn-on voltage ; serial resistance ; ideality factor ; saturated current A. The I V nonlinearity of the SBDs is determined by the Schottky junction. The peripheral semiconductor structures, including the cathode pad, the anode finger, the cathode pad, GaAs substrate, and internal layers, bring parasitic parameters. These have prominent influence on high frequency performance. The 3-D model combines a nonlinear time-domain lumped diode model to analog I V behavior and a 3-D frequency-domain electromagnetic model based on finite-element method to analog RF behavior. We have validated this 3-D model in a -band SHM [16]. The difference between simulation and measurement is within 0.5 db. Fig. 8 shows the picture of the 0.34-THz SHM, optimized to work in GHz IF frequency. It also gives the measured

5 WANG et al.: HIGH-ORDER MODULATION FOR FUTURE WLAN APPLICATIONS 79 Fig THz SHM and measured SSB conversion loss 168 GHz. single side band (SSB) conversion loss. The data is taken from back-to-back measurement, rather than the thermal noise method. In back-to-back measurement, we connect RF ports of two SHMs in the opposite direction. One SHM works as up converting and the other as down converting. The conversion loss of the two SHM is assumed to be the same. The SSB conversion loss could then be calculated through the total loss between two SHMs. This method reflects the performance of SHM more directly from the viewpoint of the final application. However, the standing wave of the RF port is rough, which brings the ripple shown in Fig. 8. When the SHM is pumped by LO in 162.5, 168, and 173 GHz, respectively, the lowest SSB conversion loss is 9.67, 9.78, and db, respectively. The 3-dB IF bandwidth is 31, 41, and 41 GHz. The difference between simulation and measurement is within 2 db. Besides the ripple, the difference incorporates errors in simulation, fabrication and measurement. When the SHM is used in GHz IF band, the SSB conversion loss increases to db. The measured lowest double side band (DSB) noise temperature of the SHM by the Y factor method is K at the 1.0-GHz IF point. B THz Multiplier Based on a Schottky Varactor The -band 8 multiplier chain provides the LO signal to pump the SHM, which consists of three doublers and two MPAs. The final stage is a 0.17-THz doubler based on Schottky varactors. Compared with varistors, varactors have better capability of enduring higher driving power and better conversion efficiency. The 0.17-THz doubler and the measured gain compression of the final multiplier chain are shown in Fig. 9. The multiplier is based on 127- m quartz substrate microstrip line. The varactor chip combines four junctions in parallel. The gold band, with a width of 100 m, is bonded between the varactor chip and the copper cavity to realize electrical grounding. It also helps to disperse the heat from diodes. The input driving signal is GHz, 100 mw, with an output range of GHz. The measured output power is 9.39 dbm at GHz. The corresponding conversion efficiency is 8.69%. The gain slope of the multiplier chain is steep. When the variation of primal driving signal is 3 db, the variation of final output power achieves 10 db. It leads to difficulty in precisely tuning the optimal LO power level for the SHM. For instance, the temperature of the devices will increase successively when the dc bias is turned on. It introduces slight variation in the driving signal, but approximate 1dBvariationinthefinal Fig. 9. The GHz doubler based on Schottky varactor and measured gain compress of 8 multiplier chain in GHz. Fig THz rectangular waveguide H-ladder bandpass filter and measured -parameter. LO. This leads the LO level deviating from the optimal value. Ultimately, the issue is solved by introducing a -band variable attenuator connected between the -band doubler and -band MPA. The initial section is then driven to saturated status, which is more stable. The LO level is then turned by the -band attenuator. The -band attenuator also helps to ameliorate the standing wave and uniformity of the chain. C THz Low-Insertion-Loss Bandpass Filter Conventional terahertz instruments adopt quasi-optical mesh filters or photo crystal grids. The advantages are low insertion loss and flexibility in system tuning. However, due to the simple resonant structure, the stopband rejection of mesh filter is imperfect. The volumes of these filters are bulky and seemingly unable to be integrated compactly with the front end. We developed a 0.34-THz H-ladder bandpass filterbasedonametal rectangular waveguide, as shown in Fig. 10. The filter adopts classical fourth-order H-ladder bandpass structure. The conventional choice for the resonators in filter is TE101 mode. However, it brings bad tolerance on the errors of mechanical fabrication. Instead, this filter adopts the resonators in TE102 mode, which could ease the difficulty. To compensate for the high dispersion of the transmission line in Terahertz frequency, we use the following equation is used to slightly shift the center frequency of designed filter: where is wavelength of desired center frequency and and are the wavelengths in upper and lower edges of pass band. The filter is fabricated by high speed CNC mechanical milling. The entire cavity is split to up and down cavities along the center (4)

6 80 IEEE TRANSACTIONS ON TERAHERTZ SCIENCE AND TECHNOLOGY, VOL. 4, NO. 1, JANUARY 2014 Fig THz Cassegrain Antenna and the measured pattern. of the rectangular waveguide. The precision of fabrication is measured to be 2.5 m. Subsequently, they are connected through the pinpointing of metal pins with a gap of 10 m. The simulated data shows that the center frequency is GHz; the 3-dB bandwidth is 19.2 GHz; the minimum in-band return loss is 18.7 db; the rejection at 325 GHz is better than 30.2 db. The data measured by a vector network analyzer (VNA) indicates that the filter is centered at GHz; the 3-dB bandwidth is 18.6 GHz; the lowest insertion loss is 1.41 db; the minimum in-band return loss is db; below GHz (4.75% offset from the center frequency), the stopbandrejectionisbetterthan30db. The difference between the simulation and measurement is mainly caused by errors in milling and gilding. It also should be indicated that the -parameter is measured with a GHz VNA frequency extension. The interface of this instrument is a WR2.2 flange, which causes small mismatch with this filter s WR2.8 flange. Practical results will be better than presented data. D THz High-Gain Cassegrain Antenna A high-directivity antenna is helpful for far field transmission. Horn antennas are prevailing in Terahertz range, but highdirectional horn antennas suffer from higher conductor loss. The Cassegrain antenna has been proven to be effective in the 0.14-THz band [17]. A 0.34-THz Cassegrain antenna has been developed in this paper as shown in Fig. 11. The physical aperture of initial feed horn is 3.0 mm 2.5 mm. The diameter of second reflector is 19 mm. The diameter of the main paraboloid reflector is 160 mm. With the increase of working frequency, the precision of fabrication and roughness of paraboloid surface are critical to the antenna. Optical interference has been utilized to monitor the fabrication errors and polish the paraboloid. In this application, the Cassegrain antenna is assembled with the 0.34-THz front-end and posited on a tripod, which would provide pinpointing precision of. The measured antenna pattern is also shown in Fig. 11. It should be indicated that standard antenna is unavailable in 0.34-THz. The measured antenna gain is calculated from the measured link loss. In the setting of measurement, two 0.34-THz Cassegrain antennas were used to transmit and receive the signal in a distance of 50.0 m. The transmitter, whose output power has been calibrated by an Erickson PM-4 power meter, provides a single tone at GHz to the transmitting antenna. A receiver whose conversion loss has been calibrated in the laboratory detects the received signal from receiving antenna. The main lobes of the two antennas pinpoint each other precisely by careful tuning the direction of tripod and monitoring the received signal. Subsequently, the transmission loss between waveguide flanges of two antennas is measured. We calculate the total antenna gain by subtracting the transmission loss with the loss of free space channel propagation and the atmosphere attenuation. We then average the total antenna gain to get single antenna gain. Following this strategy, the measured main lobe gain of the two Cassegrain antennas is 48.4 dbi in GHz point. The 3-dB beam width is The rejection between the main lobe and other lobes is better than 20.0 db. The second lobe is0.62 away from the main lobe. IV. 3-Gb/s MODULATOR AND DEMODULATOR PLATFORM A. 3-Gb/s 16QAM Parallel Modulator Considering the speed of digital devices (FPGA, ADC, DAC, and DSP), the complexity of algorithms and the feasibility of extension in future, 3-Gb/s modulation based on digital processing is incompatible with classical serial processing architecture. In the current system, a 3-Gb/s modulator adopts 32-way parallel 16QAM modulation architecture. Digital IF modulation, high-speed shaping filtering, and quadrature up converting structure are integrated in processing algorithms. The algorithm structure of a 3-Gb/s modulator is shown in Fig. 12(a). The procedure of 3-Gb/s modulation could be described as follows. First, the serial binary bits of 2.8 Gb/s are converted to 32-way parallel data flow with 87.5-Mb/s single-way data rate through serial-to-parallel converter. Second, parallel Reed Solomon RS(255,239) coding is performed on the parallel data flow. The efficiency of coding is 93.3%. Every 239 bits are coded to 255 bits. The output coded flow is 3 Gb/s. Third, the coded flow is mapped to 32-way parallel I/Q data flow with a Mbit/s single way data rate and then filtered by parallel shaping filter with four times the sample rate. Third, shaped data will experience digital up converting. Eventually, the modulated signal will be sampled by 3-Gs/s high-speed DAC. The final IF output signal is centered at 750 MHz, with an occupied bandwidth of 1.05 GHz. High-speed parallel shaping filters and digital quadrature up converting are the critical techniques in this architecture. In the high-speed parallel shaping filter algorithm, the shaping filter is based on a square-root raised cosine filter. It can be decomposed to four multi-phase slave filters. The slave filters are further decomposed through a parallel FIR filtering algorithm, based on an iterative short convolution. In the digital quadrature up converting, parallel implementation is performed through the design of parallel low-speed NCO. The hardware architecture of the 3-Gb/s modulator is shown in Fig. 13(a), which includes the power module, clock module, DAC, FPGA, and peripheral interface. The photograph of the modulator is shown in Fig. 14(a). B. 3-Gb/s 16QAM Parallel Demodulator Compared with modulation, digital demodulation confronts more challenges. The distortion caused by atmosphere (dust, water vapor, molecule absorption, and atmosphere onflow). Also, the nonlinearity of the transmitter (gain compression of

7 WANG et al.: HIGH-ORDER MODULATION FOR FUTURE WLAN APPLICATIONS 81 Fig. 12. Algorithm structure of the 3-Gb/s 16QAM (a) modulator and (b) demodulator. Fig. 13. Hardware architecture and photograph of the 3-Gb/s 16QAM (a) modulator and (b) demodulator. Fig. 14. Photographs of 3-Gb/s 16QAM (a) modulator and (b) demodulator. SHM, phase noise of LO, IM3, and group delay) will introduce phase and amplitude imbalance to the signal. Moreover, although the signal is transmitted through the 50.0-m LOS channel, the received is relatively low due to the addition of Gaussian white noise. As a result, we have proposed a 32-way parallel demodulator based on the frequency-domain implementationofamatchedfilter and timing phase correction, as described in [18]. The architecture of demodulator is shown in Fig. 12(b). The received 3-Gb/s IF signal ( 750 Msymbol/s, located at 750 MHz) will be sampled by a 3-Gs/s ADC with 10-b resolution and then written to 32-way parallel FIFOs at Mb/s single-way data rate. A mixer-free down converter and overlap operation are then performed. In this step, we realize timing frequency offset correction through index controlled read operation of the FIFOs. Subsequently, after the 64-point digital Fourier transition (DFT), the data will be multiplied by the DFT of the 33-point square-root raised cosine

8 82 IEEE TRANSACTIONS ON TERAHERTZ SCIENCE AND TECHNOLOGY, VOL. 4, NO. 1, JANUARY 2014 (SRRC) matched filter (MF) and then multiplied by a rotator to carry out timing phase correction (TPC). In the next step, we perform a 64-point inverse digital Fourier transition (IDFT). The demodulator transfers the middle 32 point outputs of IDFT to O&M timing error detector (TED). The timing synchronization algorithm in current architecture adopts a dual feedback structure. After the timing error is calculated by O&M TED, the timing phase offset is then fed back to realize TPC. The timing frequency offset is fed back to control the reading index of FIFOs, based on a delete keep algorithm. Channel equalization is indispensible to overcome amplitude and phase distortion. A parallel adaptive blind equalization module PCMA will process the eight parallel symbols from O&M in the next step. This module is based on a relaxed look-ahead pipelined parallel adaptive CMA equalization algorithm. The demodulator then performs a new phase and frequency detector (PFD)-based parallel DFPLL structure for carrier synchronization. Eight parallel symbols are sent to the DFPLL for carrier recovery. I/Q outputs are then sampled to determine the constellation and data bits. Finally, the demodulator performs RS (255,239) decoding. Parallel-to-serial converter outputs 2.8-Gb/s serial data flow. Systematic simulation indicates that the BER will be lower than, while the is better than 14.3 db without data coding. The hardware architecture of the 3-Gb/s demodulator is shown in Fig. 13(b). The photograph is shown in Fig. 14(b). The demodulator consists of one ADC10D1500 chip and two V6LX240TFF1156 FPGAs. It integrates a PCI-e high-speed interface on broad to transfer binary data between demodulator and personal computer. We calculate the constellation of the demodulated data by the algorithm on broad. The statistic of BER is performed from peripheral instruments. Fig. 15. Photograph of the 0.34-THz, 50-m, 3-Gb/s transmission experiment. Fig. 16. Measured phase noise descending of a 800-MHz single tone GHz transmitted through 0.34-THz communication system. V. MEASURED LINK PERFORMANCE A m Far-Field Transmission The 0.34-THz transmission experiment is shown in Fig. 15. The transmission distance is 50.0 m, measured by a laser range finder. The 0.34-THz transmitter outputs a GHz single-tone signal, given by an 800-MHz IF single-tone signal, which will be transmitted and finally detected by the receiver. Fig. 16 shows the contrast of phase noise between the primal 800-MHz IF single-tone signal and the received 800-MHz IF single-tone signal. For the primal signal, the respective phase noise in 100-Hz, 1-kHz, 10-kHz, 100-kHz, and 1-MHz offset frequency are 106.2, 117.5, 120.8, 118.3, and dbc/hz. For the received signal, the phase noise in the above offset frequency are 53.6, 68.0, 76.3, 77.1, and 93.7 dbc/hz. The degradation of phase noise is db. The increase of phase noise is partially accredited to the frequency multiplying in transceiver, which will raise phase noise by. is the index of multiplying, which equals 16 in the current system. Thus, the phase noise would raise 24.1 db by multiplying. Other parts of the increase in phase noise may be accredited to the phase noise of the LO power chain and first-stage IF converter. In addition, we observe spurs in Hz, Hz, and 4.69-kHz offset, which have not been found in the primal signal. To evaluate the influence of phase noise, we calculate the BER, determined by the phase noise of the received signal and based on the phase-power spectral density (PPSD) [19]. PPSD is gotten from single sideband power density spectrum. The relationship between the two quantities is given by Using linear approximation, we get the rms phase variance from (6). The rms phase variance is calculated based on the received phase noise in Fig. 16. We divide the received phase noise curve to linear approximation with five straight lines. The symbols is the start and stop frequency of each straight line. The symbol represents the slope of straight line in phase noise spectrum as The calculated is In the 16QAM scheme, the allowed phase error is 16.9, as shown in Fig. 17. The BER (5) (6)

9 WANG et al.: HIGH-ORDER MODULATION FOR FUTURE WLAN APPLICATIONS 83 Fig. 17. Allowed phase error of 16QAM scheme. Fig. 19. Relationship between, received and transmitted power. Fig. 18. Spectrum of primal and received 3-Gb/s 16QAM IF signal after 50.0-m transmission (RF is centered at GHz). introduced by phase noise could be obtained based on the probability of Gaussian distribution, as in The calculation tells us that the BER introduced by phase noise is, regardless of other distortions from RF channel and digital signal processing. After being corrected by decoding, the BER introduced by phase noise will decrease significantly. The corrected BER is shown in Fig. 21. The original BER is between to and is determined by complex factors. The primal 3-Gb/s IF spectrum and received 3-Gb/s IF spectrum after 50.0-m transmission are shown in Fig. 18. The of primal signal from the 3-Gb/s modulator is better than 40 db. The measured 99.99% power bandwidth is GHz. The gain compression and IM3 raise accessory shoulder in the received spectrum, which is 20 dbc lower than the main signal. Fig. 19 gives the measured relationship among the, the received and transmitted RF power. The maximum received is 23.8 db with the output power of 17.5 dbm. The corresponding received power is 51.4 dbm. The minimum received is 9.8 db under 32.4-dBm output power. The corresponding received power is 66.2dBm.Thedataisobtained by changing the IF input power of the SHM in the transmitter. The integration bandwidth in measurement is 1.1 GHz. (7) Fig. 20. Measured Constellation under different. The measured is 1.6 db lower than the link budget in Table I, which is mainly caused by the error of antenna gain by pointing, the uncertainty of atmosphere attenuation, and the uncertainty of instrument. If the RF power of transmitter is set to relatively higher level, the system holds better. However, the quality of signal will decrease due to gain compression. If the RF power decreases, the will be lower, but the system linearity will be improved. Fig. 20 shows the demodulated constellation of the 16QAM signal under different, which helps us to interpret this relationship. When the transmitter works in maximum output power 17.5 dbm, the corresponding received 23.8 db. The constellation tends to converge to the center, which brings with it remarkable error bits.asthe decreases to 20.6 db by reducing the output power of transmitter, the constellation converges to the four phases and four amplitudes (16 points). The interference between nearby points is inconspicuous. The corresponding BER is.whenthe falls to 12.9 db, the constellation will present acute dispersion and the relative BER degrades to. The measured BER under different is shown in Fig. 21. The lowest BER is while 18.6 db. When the is lower than 16.8 db, the BER will increase as the decreasing of by db per decade. At the lowest 9.8 db, the BER reaches. Generally speaking, a communication system with is sufficient in pragmatic application. The minimum required is 13.8 db, while. A cascaded MMIC amplifier will help us to achieve higher

10 84 IEEE TRANSACTIONS ON TERAHERTZ SCIENCE AND TECHNOLOGY, VOL. 4, NO. 1, JANUARY 2014 TABLE II SUMMARY OF RECENTLY PUBLISHED MMW/THZ COMMUNICATION LINKS Fig. 21. Measured BER under varied. Fig. 22. Photograph of the 0.34-THz WLAN experiment. with tolerable distortion in the future. Only rudimentary RS (255,239) coding is integrated in current system; a more complex interleaved code will improve the performance of the system. B THz WLAN Based on IEEE Fig. 22 shows the 0.34-THz WLAN experiment. The 0.34-THz WLAN AP is put in the opposite side of the 0.34-THz WLAN nodes A and B. As a result, the main lobe of the rectangular horn antennas of the AP could cover the lobe of horns in nodes. Two PCs are connected to nodes A and B. The AP could identify the wireless nodes with IEEE b/g protocol and act as an agent to relay the data between nodes A and B. The system operation is stable. In the experiment, a data file is transmitted from the computer of node A to that of node B by relaying of AP. The measured distance of transmission is 1.15 m. The transmission data rate is 817 kbyte/s (or Mbit/s), within 30 min. The BER of the WLAN prototype has not been measured. Because the protocol of the IEEE b/g has a strong correction mechanism to detect error bits and correct them, through coding and repeating transmission, which ensures no error bits, eventually. VI. CONCLUSION In this paper, a 0.34-THz wireless data link is presented. The wireless data link is based on a super heterodyne transceiver and 16QAM parallel digital modulation. A 0.34-THz transceiver consisted by SHM, multiplier chain, band pass filter and high directivity antenna, is described. The algorithm and hardware platform of a 3-Gb/s 16QAM modulator and demodulator are also elaborated. The experimental data indicates that the link has achieved a lowest BER of as transmitting a 3-Gb/s signal over 50.0 m. Compared with the other published THz communication research in Table II, the presented work successfully realizes high-order modulation above 0.3 THz. Both the transmission distance and spectrum efficiency are better than those achieved in other works. Furthermore, a 0.34-THz WLAN prototype is realized. Despite the fact that the physical layer of the prototype (transceiver, modulation and coding) has been completed, the media access and control (MAC) layer and high-speed communication protocol are still unexplored. Wehopetodevelopa0.34-THz WLAN that will combine both low-frequency wireless area network and high-speed THz data link in the future. ACKNOWLEDGMENT The authors would like to thank C. Zhou, J. Yao, W. Su, S.Xiao,X.Kang,P.Chen,B.Cui,andJ.JiangofCAEP-IEE for their support of this work. REFERENCES [1] A. Hirata, T. Kosugi, H. Takahashi, J. Takeuchi, H. Togo, M. Yaita, N. Kukutsu,K.Aihara,K.Murata,Y.Sato,T.Nagatsuma,andY.Kado, 120-GHz-band wireless link technologies for outdoor 10-Gbit/s data transmission, IEEE Trans. Microw. Theory Techn., vol. 60, no. 3, pp , Mar [2] A. Hirata, T. Kosugi, H. Takahashi, J. Takeuchi, K. Murata, N. Kukutsu, Y. Kado, S. Okabe, T. Ikeda, F. Suginosita, K. Shogen, H. Nishikawa, A. Irino, T. Nakayama, and N. Sudo, 5.8-km 10-Gbps data transmission over a 120-GHz-band wireless link, in Proc. IEEE Int. Wireless Inform. Technol. Syst. Conf., 2010, pp [3] H. J. Song, K. Ajito, Y. Muramoto, A. Wakatsuki, T. Nagatsuma, and N. Kukutsu, 24 Gbit/s data transmission in 300 GHz band for future terahertz communications, Electron. Lett., vol. 48, pp , [4] I. Kallfass, J. Antes, T. Schneider, F. Kurz, D. Lopez-Diaz, S. Diebold, H. Massler, A. Leuther, and A. Tessmann, All active MMIC-based wireless communication at 220 GHz, IEEE Trans. Terahertz Sci. Technol., vol. 1, no. 2, pp , Nov [5] I. Kallfass, J. Antes, D. Lopez-Diaz, S. Wagner, A. Tessmann, and A. Leuther, Broadband active integrated circuits for terahertz communication, in Proc. 18th Eur. Wireless Conf., 2012, pp [6] Z. Bo, X. Yong-Zhong, W. Lei, and H. Sanming, A switch-based ASK modulator for 10 Gbps 135 GHz communication by 0.13 m MOSFET, IEEE Microw. Wireless Compon. Lett., vol. 22, no. 8, pp , Aug

11 WANG et al.: HIGH-ORDER MODULATION FOR FUTURE WLAN APPLICATIONS 85 [7] L. Moeller, J. Federici, and S. Ke, THz wireless communications: 2.5 Gb/s error-free transmission at 625 GHz using a narrow-bandwidth 1 mw THz source, in Proc. XXXth URSI Gen. Assemb. Sci. Symp.,2011, pp [8] V. Dyadyuk, J. D. Bunton, J. Pathikulangara, R. Kendall, O. Sevimli, L. Stokes, and D. A. Abbott, A multigigabit millimeter-wave communication system with improved spectral efficiency, IEEE Trans. Microw. Theory Techn., vol. 55, no. 12, pp , Dec [9]J.H.Booske,R.J.Dobbs,C.D.Joye,C.L.Kory,G.R.Neil,P. Gun-Sik, P. Jaehun, and R. J. Temkin, Vacuum electronic high power terahertz sources, IEEE Trans. Terahertz Sci. Technol., vol. 1, no. 1, pp , Sep [10] L. A. Samoska, An overview of solid-state integrated circuit amplifiers in the submillimeter-wave and THz regime, IEEE Trans. Terahertz Sci. Technol., vol. 1, no. 1, pp. 9 24, Sep [11] V.Radisic,K.M.K.H.Leong,S.Sarkozy,X.Mei,W.Yoshida,L. Po-Hsin,andR.Lai, A75mW210GHzpoweramplifier module, in Proc. IEEE Compound Semiconduct. Integr. Circuit Symp., 2011, pp [12] J. C. Tucek, M. A. Basten, D. A. Gallagher, and K. E. Kreischer, 220 GHz power amplifier development at Northrop Grumman, in Proc. IEEE 13th Int. Vacuum Electron. Conf., 2012, pp [13] H.-J. Song and T. Nagatsuma, Present and future of terahertz communications, IEEE Trans. Terahertz Sci. Technol., vol. 1, no. 1, pp , Sep [14] T. Kleine-Ostmann, C. Jastrow, S. Priebe, M. Jacob, T. Kurner, and T. Schrader, Measurement of channel and propagation properties at 300 GHz, in Proc. Conf. Precision Electromagnetic Meas., 2012, pp [15] C. Jastrow, K. Munter, R. Piesiewicz, T. Kurner, M. Koch, and T. Kleine-Ostmann, 300 GHz channel measurement and transmission system, in Proc. 33rd Int. Conf. Infrared, Millimeter and Terahertz Waves, 2008, pp [16] C. Wang, X. Deng, L. Miao, and J. Yan, GHz sub-harmonic mixer based on Schottky barrier diodes, in Proc. Int. Microw. Millim. Wave Technol. Conf., 2012, pp [17] C. Wang, L. Changxing, C. Qi, D. Xianjin, and Z. Jian, 0.14 THz high speed data communication over 1.5 kilometers, in Proc. 37th Int. Infrared, Millim., Terahertz Waves Conf., 2012, pp [18] C. Lin, J. Zhang, and B. Shao, A multi-gigabit parallel demodulator and its FPGA implementation, IEICE Trans. Fundam. Electron., Commun. Comput. Sci., vol. E95-A, pp , [19] W. F. Egan, Practical RF System Design. New York, NY, USA: Wiley Interscience, pp Cheng Wang was born in Suining, China, on March 8, He received the B.S.degreeinengineering physics from Tsinghua University, Beijing, China, in 2008, and the M.S. degree in radio physics from China, in He joined the Institute of Electronic Engineering, China, in His current research involves millimeter-wave//terahertz communication system, mixers and multipliers based on Schottky diodes, and terahertz passive waveguide components. Changxing Lin was born in Chongqing, China, on January 7, He received the B.S. degree in engineering physics and Ph.D. degree in nuclear science and technology from Tsinghua University, Beijing, China, in 2007 and 2012, respectively. He was a Research Assistant from June 2009 to December 2009 with the European Organization for Nuclear Research (CERN). He joined the Institute of Electronic Engineering, China Academy of Engineering Physics, Mianyang, China, in His current research involves algorithm and implementation of high-speed demodulation for communication and terahertz wireless local area network. Qi Chen was born in Chongqing, China, on November 3, He received the B.S. degree in remote sensing techniques and instrument and M.S. degree in electromagnetic and microwave techniques from Xidian University, Xian, China, in 2003, and 2007, respectively. He is currently working toward the Ph.D. degree in radio physics at the China Academy of Engineering Physics, Mianyang, China. He joined the Institute of Electronic Engineering, China in His research interests include mmw/ Terahertz antennas, photo crystal and meta-materials. Li Miao was born in Mianyang, China, on September 27, She received the B.S. and M.S. degrees in electromagnetic and microwave techniques from Southwest Jiao Tong University, Chengdu, China, in 2009, and 2012, respectively. She joined the Institute of Electronic Engineering, China, in Her research includes terahertz nonlinear circuits and passive components. Xianjin Deng wasborninanyue, China, on June 11, He received the B.S. degree in electronic engineering from Xidian University, Xian, China, in 1998, and the M.S. degree in electronic science and techniques from University of Electronic Science and Technology of China, Chengdu, China, in He joined the Institute of Electronic Engineering, China, in His current research involves millimeter-wave (mm-wave)/terahertz communication system, mm-wave solid-state power combining, and microwaveactivecircuits. Bin Lu was born in Chongqing, China, on August 18, He received the B.S. degree in electronic engineering from Fudan University, Shanghai, China, in 2004, and the M.S. degree in communication and information system from China Academy of Engineering Physics, Mianyang, China, in He has been an Engineer since 2011 with the Institute of Electronic Engineering (IEE), China Academy of Engineering Physics (CAEP), Mianyang, China, where he is engaged in research on terahertz communication and radar systems, millimeter-wave passive filters and digital processing algorithms. Jian Zhang was born in Sichuan, China, on November 27, He received the B.S. degree in electronic techniques from the National University of Defense Technology, Changsha, China, in 1989, the M.S. degree in communication engineering from the China Academy of Engineering Physics, Mianyang, China, in 1994, and the Ph.D. degree in electrical engineering from Chongqing University, Chongqing, China, in He joined the Institute of Electronic Engineering, China, in his current research involves electronic systems, wireless communication, and terahertz science and technology.