Error Vector Magnitude Improvement Research Articles

Machine learning-based methods for piecewise digital predistortion in mmW 5G NR systems

AbstractPiecewise linearization techniques require dividing the signal into multiple pieces each linearized individually. Machine learning (ML) is one of the useful tools to perform the automatic division of these pieces. Complexity reduction in the classification of piecewise digital predistortion is possible through carefully constructing features from both the signal statistics and the power amplifier (PA) characteristics. Our paper introduces two low-complex classical ML-based methods that facilitate the classification of baseband input data into distinct segments. These methods effectively linearize PA behavior by employing tailored Volterra models corresponding to each segment. Moreover, we perform an in-depth analysis of the proposed schemes to further optimize their classification and regression complexities. The two proposed low-complexity approaches are validated by laboratory experiments and show up to 4 dB error vector magnitude (EVM) improvement over the conventional approach for a class A PA at 28 GHz. Similarly, the EVM improvement is up to 2 dB over the vector-switched general memory polynomial scheme. With only one indirect learning architecture iteration, the two proposed schemes obey the 5G new radio standard up to 6.5 dB and 7 dB output backoff, respectively.

Bandwidth-Scalable Digital Predistortion of Active Phased Array Using Transfer Learning Neural Network

This paper proposes a transfer learning neural network (TLNN) approach for digital predistortion (DPD) of mm-Wave active phased arrays (APA) operated under variable signal bandwidth regimes. Compared with the conventional artificial neural network (ANN) method, the proposed approach can achieve similar linearization performance with much lower computational complexity by transferring part of a trained model from one bandwidth to another bandwidth. In the recently introduced 5G, the increased signal bandwidth triggers considerable memory effects in the APA. Moreover, dealing with different signal bandwidths typically requires a time-consuming recalculation of the predistorter parameters. In this paper, the authors propose to have those challenges solved by using a DPD model based on the transfer learning method. The proposed approach was validated with over-the-air (OTA) measurements on an APA excited with signals of varying bandwidth, namely from 20 MHz to 100 MHz. Experimental results show a significant reduction in the training time while ensuring good linearization performance. With the applied TLNN DPD, an 8.5 dB improvement of adjacent channel leakage ratio (ACLR) and 8.6 % points improvement of error vector magnitude (EVM) is achieved. Under the variable bandwidth regime, the complexity of the DPD model in terms of the number of multiplications is reduced from 199168 to 160. The proposed TLNN DPD proved to be robust concerning variation in the bandwidth of the APA excitation signal.

Low-Complexity Adaptive Frequency-Domain Nonlinear Equalization for Analog RoF Mobile Fronthaul Using FFT/IFFT-Assisted Channel Aggregation

Addressing the nonlinearity induced signal degradation is of critical importance for harvesting the transparent benefits of an analog radio over fiber (A-RoF) architecture, to support mobile fronthaul (MFH) with high spectral efficiency and reduced latency. Here, for linearizing the low-latency A-RoF MFH using digital-domain fast Fourier transform /inverse fast Fourier transform (FFT/IFFT) assisted channel aggregation/de-aggregation (CA/DA), we propose and experimentally demonstrate a compatible adaptive frequency-domain (FD) nonlinear equalization scheme. This scheme is characterized by a much lower computational complexity compared with the popular time-domain Volterra nonlinear equalizer while delivering a comparable performance. The proposed equalization scheme is implemented using the FD two-box cascaded Hammerstein structure. It enables simplified coefficients estimation, computationally efficient equalization, FFT/IFFT calculation reusing in channel DA, and shared legacy one-tap FD linear equalization in the OFDM system. Experimental validations of the proposed nonlinear equalization approach are carried out via typical electroabsorption modulator (EAM) and Mach-Zehnder modulator (MZM) based intensity modulation and direct detection (IMDD) A-RoF links. Cost-effective fiber-optic (with EAM) and optical up-conversion enabled V-band millimeter-wave (mm-wave) fiber-wireless (with MZM) transmissions are demonstrated. Experimental results show that the proposed scheme can effectively improve the performances of both EAM and MZM based 10-km fiber-optic A-RoF MFH links, in terms of the (>8 dB) reduction of adjacent channel leakage ratio, (>10 dB) improvement of error-vector magnitude (EVM) and (>5 dB) enhancement of input intermediate-frequency (IF) power range, which retains an acceptable (below 8 %) EVM performance for the 64-QAM modulation. The input IF signal aggregates 8 125-MHz 64-QAM OFDM channels. Furthermore, the proposed scheme enables the transmission of a 5-Gbit/s aggregated IF 64-QAM signal over a 10-km fiber link and 1.2-m 60-GHz wireless channel, wherein a >12 dB improved input power margin is observed.

Hybrid wideband multipath self-interference cancellation with an LMS pre-adaptive filter for in-band full-duplex OFDM signal transmission.

A wideband multipath self-interference cancellation (SIC) system employing both dual-drive Mach-Zehnder modulator-based analog SIC and least mean square (LMS) algorithm-based pre-adaptive filter digital SIC is proposed and demonstrated for the cancellation of multipath self-interference (SI) and facilitation of in-band full-duplex (IBFD) orthogonal frequency-division multiplexing (OFDM) signal transmission. The multipath effect is an unavoidable challenge in SIC due to the dynamic and unpredictable properties in each path, as well as the need for separate matching components for compensating for each path. In this Letter, an LMS algorithm-based adaptive filter is used as a pre-equalizer to adapt and generate the matching signal to the closest approximate of the multipath SI signal. The adaptation is based on the minimization of the error signal generated from the matching signal and multipath SI signal in the LMS algorithm. With the introduction of the LMS adaptive filter to the analog SIC, an additional 9 dB cancellation improvement is obtained, resulting in a total of 32 dB cancellation depth over a cancellation bandwidth of 2.7 GHz at a center frequency of 1.65 GHz. To the best of our knowledge, the achieved performance is by far the widest cancellation bandwidth in a multipath SIC system, which is essential in a large bandwidth and high data rate transmission system. With the help of the proposed LMS adaptive filter digital SIC assisted analog SIC located at the remote node, power-efficient IBFD transmission of an OFDM signal through a 25 km fiber is experimentally demonstrated with a 6 dB bit error rate and 8% error vector magnitude improvements.

Analysis of Spectral Degradation and Error Compensation in 5G NR Digital Polar Transmitters

Cellular standards evolve to support increasingly higher bandwidths which results in strict in- and out-of-band requirements such as lower error vector magnitude (EVM) and/or adjacent channel leakage ratio (ACLR). Digital polar transmitters, which are showing the best performance in terms of power consumption, are challenged to fulfill these requirements. This work will show that even in case of ideal analog components and infinite digital resources, there are principle limitations in terms of out-of-band noise for digital polar transmitters. A typical polar modulation will be compared to a theoretical signal with continuous amplitude modulation. The analysis of the resulting error signal suggests a decomposition into two components which can be partially compensated in the amplitude and phase modulation paths. Based on that decomposition a compensation algorithm has been developed and evaluated with a 400 MHz 5G New Radio (NR) signal for Band n257 on Frequency Range 2 with carrier frequency of 26.5 GHz. The compensation results in an out-of-band noise reduction of 44 dB close to the carrier frequency. Furthermore, EVM improvements for NR signals are demonstrated.

PAPR reduction and non-linearity alleviation using hybrid of precoding and companding in a visible light communication (VLC) system

In this paper, a hybrid technique using a precoder and a µ law compander is proposed to alleviate light emitting diode (LED) nonlinearity as well as peak-to-average power ratio (PAPR) reduction in a VLC system. A µ law compander is used to compress the peak valued signal and enhances small amplitudes to lower the PAPR. The work attains optimal PAPR and improves bit error rate (BER) of the system at low signal to noise ratio values. The nonlinear characteristic of a LED used as the transmitter in a VLC system leads to damaging effects while using OFDM in an optical system. The combination scheme improves LED nonlinearity for low input back off (IBO) power. In the work, nonlinearity is characterized using error vector magnitude (EVM) in percentage. The proposed method shows a 5% EVM improvement for 0 dB IBO. The paper provides a choice for a trade-off among PAPR reduction, BER and reduced nonlinearity.

Dynamic range improvement of six-port receiver through analysis of output DC offset

In this paper, we improved the error vector magnitude (EVM) performance of six-port receiver using a low noise amplifier (LNA) bias control algorithm based on analyzing the correlation between normalized output DC offset voltage and EVM relative to input radio frequency (RF) power variation. In contrast to the other works published in the literature, we consider test setup through which the software impairments of six-port receiver blocks are considered. Therefore, in this work to substantiate the EVM improvement, a designed 26–30 GHz CMOS six-port receiver is used which consists of LNA, coupled line bandpass filter, six-port correlator, diode power detectors and two instrumentation amplifiers for recovering the baseband I/Q data. The EVM improvement process is assessed in terms of 64QAM modulation, and according to obtained results, EVM is reduced from 61% when using constant DC bias to 22% achieved with an LNA bias control algorithm for low input power (− 60 dBm) at the center frequency of 28 GHz. Moreover, the EVM is become less sensitive to input RF power, and also the dynamic range is improved more than 20 dB.

Fine time synchronization and residual ISI influence in SC-FDE systems

We show a novel method for a fine time synchronization in a presence of residual intersymbol interferences (ISIs). The proposed method is compared with a (1) Cramér-Rao lower bound (CRLB) for data-aided timing estimators and with a (2) maximum likelihood (ML) estimator both in (a) ISI-free additive white Gaussian noise (AWGN) channel and (b) in a residual ISI channel. The effect of the residual ISI is introduced via a power-delay profile (PDP) with a delay spread spanning over several samples. We consider a single-carrier system with a frequency domain channel estimation. The proposed method almost reaches the performance of the ML estimator in AWGN, while in the presence of the residual ISI, the proposed method shows superior performance evaluated in terms of error vector magnitude improvement of up to 10%.

Correlation-Based Calibration for Nonlinearity Mismatches in Dual-Channel TIADCs

Mismatches affect the dynamic performance of time-interleaved analog-to-digital converters (TIADCs). Linear mismatches can be calibrated by many mature methods, but if higher performance is required, nonlinearity mismatches have to be suppressed. The background calibration method based on input-free band (IFB) functions poorly for narrow-band signals. This brief proposes a correlation-based calibration method for nonlinearity mismatches in dual-channel TIADCs which behaves well for both wide-band and narrow-band signals. The output samples are calibrated by reducing the residual distortions which are approximated by multiplying the pseudo distortions and the estimated mismatch coefficients. The pseudo distortions are acquired by using a frequency-shifter, a differentiator, and multipliers. The coefficients which indicate the mismatch strength are estimated by eliminating the cross-correlation of the calibrated output samples and the calibrated pseudo distortions at zero lag. Simulations show that the proposed method can improve the SFDR by dozens of dBc for narrow-band input signals, compared with the IFB method. For the 16-QAM signal, the error vector magnitude improvement over the IFB method is 35.48 dB.

A Reliable OFDM-Based MMW Mobile Fronthaul With DSP-Aided Sub-Band Spreading and Time-Confined Windowing

Performance of digital signal processing (DSP) aided sub-band spreading orthogonal frequency division multiplexing (S-OFDM) implemented with the well-defined, time-located window is evaluated and compared with the cyclic prefix (CP) OFDM signal under severe interference circumstances. A four-quadrature amplitude modulation (QAM) mobile fronthaul over 20 km fiber link and 1 m millimeter wave 60-GHz wireless channel is demonstrated experimentally. The bit-error rate performance of applying time windowing on OFDM and S-OFDM are measured in the single channel scenario. In such simplified system, CP-OFDM is found to be a widely adopted waveform. While, as it suffers from the unexpected external interference, the received error vector magnitude (EVM) would be degraded dramatically for typical OFDM signals. By applying the Hanning window, the impact on the EVM performance can be alleviated as the interference is at the edge of the desired signal band and resulting a 2.5% EVM enhancement. However, an irrecoverable 20% data loss is obtained if the interference is totally immersed in the victim signal. In contrast, S-OFDM exhibits a frequency diversity property by grouping OFDM subcarriers into several sub-bands and spreading them into whole available bandwidth. Thus, as S-OFDM suffers from strong interference inside the signal band, about 12.61% and 43.53% EVM improvements of the average and the worst subcarrier performance with respect to the ordinary OFDM can be achieved via applying the spreading codes orthogonality restoring equalizer. All subcarriers in the S-OFDM are generated to meet the forward error correction threshold with reasonably flat performance. Therefore, the S-OFDM shows its superiority in an MMW mobile fronthaul for better reliability in data transmission.

Performance Enhancement of Multi-Antenna Relay Link in LTE Het-Net

In this paper, the relay link performance for an Amplify and Forward (AF) multi-antenna relay based heterogeneous LTE (Long Term Evolution) network is enhanced with two novel techniques, the Golden codes and Error Vector Magnitude (EVM) based Constellation Combiner (ECC). The performance evaluation of the proposed co-operative LTE Het-Net is done in a real time indoor environment using Wireless open Access Research Platform (WARP). The proposed techniques for the LTE Het-Net are compared with traditional Alamouti Scheme with Maximal Ratio Combing (MRC). From the obtained results it is seen that the LTE network with proposed techniques gives a 12% improvement in EVM than the conventional LTE network.

Reconfigurable Optical Network Intermediate Node With Full-Quadrature Regeneration and Format Conversion Capacity

In this paper, a reconfigurable optical network intermediate node scheme with format conversion capacity and full-quadrature regeneration functionality is proposed. With a degraded input quadrature phase-shift keying (QPSK), the proposed network node can be dynamically reconfigured based on the network situation to output a QPSK or a four-level pulse amplitude modulation (PAM4), both full-quadrature regenerated. The input QPSK is first converted into the two full-quadrature regenerated binary phase-shift keying tributaries after phase sensitive amplifier-based quadrature de-multiplexing and self-phase modulation-based amplitude regeneration. Then the output regenerated QPSK or PAM4 are generated through vector addition by tuning the relative phase and amplitude between two tributaries. For an input 20 Gb/s QPSK with optical signal-to-noise ratio (OSNR) of 15 dB, the output regenerated QPSK gets 10.4% decrease in error vector magnitude (EVM), 2 dB receiver OSNR improvement at the bit error rate of 10 $^{ - 3}$ , and regeneration factors of 6.4 dB and 3.7 dB in amplitude and phase, respectively. In the case with regenerated PAM4 as the output, compared with the input QPSK, it has 10.4% improvement in EVM and regeneration factors of 7.1 dB and 2.5 dB in amplitude and phase, respectively. The numerical simulation results reveal that the system effectively regenerates input signal in full-quadrature with format conversion capacity. The proposed node scheme can be applied as a full-quadrature QPSK regenerator between optical networks, which both employ QPSK to increase the transmission distance, or an all-optical interconnection node between different optical networks where different modulation formats are used individually.

Hybrid Linearization of Broadband Radio-Over-Fiber Transmission

This letter presents analog and digital pre-distortion hybrid linearization of a broadband radio-over-fiber transmission system. Analog pre-distortion (APD) is used to mainly compensate for out-of-band nonlinear distortion, and digital pre-distortion (DPD) is used to mainly compensate for in-band nonlinear distortion and memory effect. Thus, by combining APD and DPD, all the quiescent distortions and memory effect can be suppressed effectively. The hybrid APD and DPD linearization is evaluated in three different scenarios: two 20-MHz long-term evolution (LTE) signals located at 800 and 900 MHz, and located at 800 and 840 MHz, and three 20-MHz LTE signals located at 800, 850, and 900 MHz. All three experiments show that the hybrid linearization results in better improvement in error vector magnitude, adjacent-channel power ratio, and third-order intermodulation distortion than the APD-only and DPD-only.

Adaptive Digitization and Variable Channel Coding for Enhancement of Compressed Digital Mobile Fronthaul in PAM-4 Optical Links

The standardization and development of LTE-A and 5G introduced advanced wireless technologies including multiple-input multiple-output and carrier aggregation, which require multiple wireless carriers to be delivered to and from each remote radio head. The common public radio interface (CPRI) as the mainstream standard in mobile fronthaul (MFH) with on-off-keying-based optical links cannot fulfill the capacity and efficiency requirement. Instead, using compressed CPRI in a high-speed pulse-amplitude-modulation-4 (PAM-4) link is actively researched and demonstrated. In this paper, we propose and demonstrate adaptive digitization and channel coding based on these compression and capacity boosting technologies. Depending on the optical link condition, the digitization bits and channel coding rates can be adaptively and dynamically changed to achieve the lowest error vector magnitude (EVM) of wireless carriers. By separating digitization bits into high bits and low bits, the coding overhead can be different between groups, while still keeping the same bit rate per wireless carrier. Based on the existing digital MFH infrastructure, the proposed scheme can significantly improve the capacity and sensitivity in the PAM-4-based compressed digital MFH. Capacity gains from 30% to 68%, sensitivity improvement of 2–9 dB, and significant EVM improvements are demonstrated experimentally, comparing with other compressed CPRI MFH solutions.

Investigation of broadband digital predistortion for broadband radio over fiber transmission systems

In future broadband cloud radio access networks (C-RAN), front-haul transmission systems play a significant role in performance and cost of C-RAN. Broadband and high linearity radio over fiber (RoF) transmission systems are considered a promising solution for the front-haul. Digital linearization is one possible solution for RoF front-haul. In this paper, we investigate RF domain digital predistortion (DPD) linearization for broadband RoF front-haul. The implemented DPD is first investigated in 2.4GHz WiFi over fiber transmission systems at 36Mb/s, and more than 8-dB and 5.6-dB improvements of error vector magnitude (EVM) are achieved in back to back (BTB) and after 10km single mode fiber (SMF) transmission. Further, both WiFi and ultra wide band (UWB) wireless signals are transmitted together, in which the DPD has linearization bandwidth of 2.4GHz. It is shown that the implemented DPD leads to EVM improvements of 4.5-dB (BTB) and 3.1-dB (10km SMF) for the WiFi signal, and 4.6-dB (BTB) and 4-dB (10km SMF) for the broadband UWB signal.

Virtual Carrier-Assisted Direct-Detection MB-OFDM Next-Generation Ultra-Dense Metro Networks Limited by Laser Phase Noise

The impact of the phase-to-intensity conversion of the laser phase noise due to the optical band selector (BS) used for band extraction on the performance of direct-detection (DD) optical multiband orthogonal frequency-division multiplexing (MB-OFDM) networks employing single-sideband (SSB) modulation is assessed. Compared to conventional double-sideband DD systems, in which the amplitude shape of the optical filter is quasi-symmetric relative to the centre of the signal spectrum, ultra-dense optical SSB-MB-OFDM networks employing virtual carrier-assisted DD are differently affected by the phase-to-intensity conversion. This is due to the detuned BS required by these networks, which modify the converted phase noise (CPN) features. A reduction of the power spectral density of the CPN achieving 30 dB can occur along the frequency range in which the BS bandwidth and detuning lead to a filtering attenuation that is similar in both sides of the virtual carrier. For single band, when the ratio between the BS detuning and bandwidth increases from 0% to 28%, an error vector magnitude improvement of 7 dB due to the CPN reduction is observed. The optical signal-to-noise ratio (OSNR) penalty due to the laser phase noise in 112-Gb/s SSB-MB-OFDM networks comprising 12 OFDM bands and employing an ultra-dense band spacing of 3.125 GHz is lower than 1 dB for external cavity lasers and is 1.6 dB for distributed feedback lasers. These OSNR penalties are obtained for an optimized ratio between the BS detuning and bandwidth of 14%.

Analysis of Dual Wavelength Linearization Technique for Radio-Over-Fiber Systems With Electro-Absorption Modulator

The dual wavelength linearization (DWL) technique using two lasers with different wavelengths is for the first time comprehensively studied theoretically and experimentally to improve output signal power and suppress second- and third-order nonlinearities in radio-over-fiber (RoF) systems using an electro-absorption modulator (EAM). Spurious-free dynamic ranges (SFDRs) are calculated and measured. Both the calculation and experiments show that not only the RF signal power is significantly increased, but also both the second- and third-, as well as fifth-order nonlinearities are suppressed. For our considered EAM, when the second-order nonlinearity is suppressed maximally, the SFDRs with respect to second-order harmonic distortion and second-order intermodulation distortion are improved by 11.5 and 8.5 dB, respectively. In the mean time, the SFDRs with respect to third-order harmonic distortion (HD3) and third-order intermodulation distortion (IMD3) are improved by 1.8 and 1.3 dB, respectively. When the third order is suppressed maximally, the SFDRs with respect to HD3 and IMD3 are improved by 8.1 and 20.4 dB, respectively. The fifth-order intermodulation distortion is also suppressed at the same time. The RoF transmissions for WiFi signals are also verified. As a result, 3.5 dB at 2.4-GHz improvement and 2.8 dB at 5-GHz improvement of error vector magnitude are obtained for an RoF system by using the DWL technique.

Broad baseband nonlinear distortion mitigation using digital pre- and post-distortion in OFDM-based WDM LR-PON.

We propose a new technique for mitigation of nonlinear distortion of broadband signals due to electrical/optical conversions and fibre transmission. This technique uses memory polynomials on signals on their original format. The performance improvement is assessed experimentally for an orthogonal frequency division multiplexing signal, with 750 MHz of bandwidth (centred at 1.5 GHz) and carrying 1.25 Gbps, transmitted along a wavelength division multiplexing long-reach passive optical network. It is shown that, in back-to-back, an error vector magnitude improvement of 4.1 dB and 11.8 dB can be achieved with digital pre-distortion and digital post-distortion, respectively. With 125 km of single-mode fibre transmission, these improvements are reduced to about 1 dB and 3 dB, respectively.

Improved nonlinear tolerance in ultra-dense WDM OFDM systems

This paper demonstrates and experimentally validates the Zadoff–Chu matrix transform (ZCT) precoding as a peak-to-average power ratio (PAPR) reduction technique for coherent optical orthogonal frequency division multiplexing (OFDM) over ultra-dense wavelength division multiplexing (UDWDM). The work provides performance comparison between conventional and the precoded UDWDM–OFDM system. The system use case was implemented by 16-QAM–OFDM modulated channels among 3.5GHz, 7GHz, and 10GHz channel spacings scenarios. The ZCT precoding approach improves the nonlinear tolerance in UDWDM–PON systems. The performance in terms of error vector magnitude (EVM) and bit error rate (BER) is presented after transmission over 40km of standard single mode fiber (SSMF). The results show a significant EVM improvement considering the aforementioned channel spacings and for different number of UDWDM channel groups in the nonlinear regime operation. The implemented ZCT precoding system outperforms the non precoded approaches in terms of EVM and BER results.

Real-time ultra-wideband video streaming in long-reach passive optical networks with wireless radiation in the 10 and 60 GHz Bands

Real-time video streaming using ultra-wideband (UWB) technology is experimentally demonstrated along long-reach passive optical networks (LR-PONs) with different wired and wireless reaches. Experimental tests using external and direct modulation with UWB wireless radiation in the 10- and 60-GHz bands are performed. An ultra-bendable fiber is also considered for a last-mile distribution. The video quality at the output of the optical fiber infrastructure of the LR-PON is assessed using the error vector magnitude (EVM), and the link quality indicator (LQI) is used as a figure of merit after wireless radiation. An EVM below -17 dB is achieved for both externally and directly modulated LR-PONs comprising up to 125 km of optical fiber. EVM improvement is observed for longer LR-PONs when directly modulated lasers (DMLs) are used because of the amplitude gain provided by the combined effect of dispersion and DML's chirp. Compared with optical back-to-back operation, the LQI level degrades to the maximum around 20% for LR-PONs ranging between 75 and 125 km of fiber reach and with a wireless coverage of 2 m in the 10-GHz UWB band. The same level of LQI degradation is observed using the 60-GHz UWB band with a LR-PON integrating 101 km of access network, a last-mile distribution using ultra-bendable fiber, and a 5.2-m wireless link.