Open Access
Add to BibSonomy
Abstract
Radio-over-fiber (RoF) transmission is a quite reliable technology to support the current and future demands of rapidly progressing broadband wireless network with large capacity and high spectral efficiency. In this paper, we report and demonstrate a digital RoF transmission system using two-dimensional discrete cosine transform with vector quantization (2D-DCT-VQ). By employing the 2D-DCT-VQ technique, the spectral efficiency can be greatly improved, while the system performance is comparable to the traditional approach without compression. The proposed method is experimentally demonstrated in a 20-km 5-Gbaud/λ four-level pulse modulation intensity-modulation/direct-detection optical link. In the orthogonal frequency-division multiplexing -modulated downlink illustrated experimentally, the transmission rate rises by 69.49% on account of the compressed samples by using the proposed method.
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
With rapid developments of broadband wireless network, the spectrum resource in high-frequency band is exploited and utilized. Meanwhile, the evolving massive mobile internet with high connectivity and throughput features higher requirements, such as larger capacity, higher transmission rate, and ultra-reliable low-latency communication [1,2]. To this end, radio-over-fiber (RoF) technology has become a promising and reliable solution to support the progressing demands, due to its inherent advantages, such as large bandwidth, low loss, immune to electromagnetic interference and flexibility [3–5]. Particularly, the base station function of radio access network is reframed and split for fifth generation (5G) new-radio communication network, in which the function split point can be selected by application scenario, and the enhanced common public radio interface (eCPRI) specification has been developed which can employ the RoF mobile fronthaul technology.
Both analogue and digital RoF transmission systems to date have been intensively investigated. The analogue RoF system has characteristics of simple receiver architecture and high bandwidth efficiency. However, the system dynamic range is limited by the nonlinearity of optoelectronic devices, and fiber chromatic dispersion also affects the system performance [6–8]. Digital RoF system offers an attractive alternative to overcome the constraints of analogue RoF system, since it is benefit from the improved performance of digital optical links. Thanks to the mature digitalized-based technology, the digital RoF system is robustness, format-agnostic, multiservice-compatible, etc.
The previous digital RoF technique using the CPRI protocol in mobile fronthaul link employs 15 quantization bits pulse-code modulation (PCM) and uniform quantization, which requires a large transmission bandwidth [9]. Various improved methods are proposed on this issue [10–19], which are mainly classified into two categories. One is based on scalar quantization, such as utilizing delta-sigma modulation [11,12], differential PCM [13–15], fast statistical-estimation (FSE) [16,17] and differential PCM-based entropy coding [18], which can improve the spectral efficiency through reducing the quantization noise. The other is utilizing vector quantization, such as two-dimensional (2D) self-organizing feature map neural network clustering [20], multi-dimensional k-means clustering [21], 2D vector quantization with vector linear prediction [22] and vector quantization-based FSE [23]. Compared to the former ones, the vector quantization which quantizes the input data by groups can achieve better compression.
In this work, we propose and experimentally demonstrate a digital RoF transmission scheme using 2D discrete cosine transform with vector quantization (2D-DCT-VQ). By using the proposed 2D-DCT-VQ method, the spectral efficiency of the transmission system can be considerably improved without increasing the total number of quantization bits for digitizing the baseband orthogonal frequency-division multiplexing (OFDM) symbols. A proof-of-concept experiment is carried out over a 20-km standard single-mode fiber (SMF) with 5-Gbaud/λ four-level pulse modulation (PAM-4) signals to investigate the transmission performance. The experimental results show that the transmission rate of the proposed digital RoF system using OFDM-modulated downlink is improved by 69.49% with comparable performance to the methods without compression.
2. Principle
The digital mobile fronthaul in 5G refers to the communication network between distributed unit (DU) and active antenna unit (AAU). Taking the case of downlink is shown in Fig. 1, where both the digital signal processing flows at the transmitter (Tx-DSP) of the DU and the receiver (Rx-DSP) of the AAU are illustrated. At the transmitter, OFDM modulation is first performed on the baseband signals. Then signals are arranged in frames by time-division-multiplexing (TDM) for carrier aggregation. In following, 2D-DCT and vector quantization are executed. Finally, the processed signals are encoded with PAM-4, and modulated onto an optical source through intensity-modulation. At the receiver, the received signals by direct-detection are processed by an opposite operation before sending to the radio frequency (RF) front-end.
Fig. 1. The mobile fronthaul downlink based on digital RoF. E/O: electrical-optical conversion; O/E: optical-electrical conversion; DU: distributed unit; AAU: active antenna unit; Tx-DSP/Rx-DSP: digital signal processing at the transmitter/ receiver.
Concretely, the 2D-DCT transforms the signals from spatial domain to the frequency domain and goes through with energy compaction. A 2D vector matrix A[SIi;SQi] is constructed by the in-phase and quadrature components of the OFDM samples (Si) as shown in Fig. 2(a). SIi = real(Si) and SQi = imag(Si). Then 2D-DCT of the vector matrix A is performed as follow:
Fig. 2. The performance of 2D-DCT. (a) the constructed 2D vector by in-phase and quadrature components of OFDM samples, (b) the output frequency coefficients by the 2D-DCT.
The process of 2D vector quantization as described above can be mainly divided into two stages: offline codebook design and online real-time codeword search. For a kind of signals with same parameters, the codebook is universality and can be generated offline by performing the 2D k-means clustering algorithm on the input training vector set. After generating the codebook, the untrained vector signals can be simply quantified in real-time by searching the codeword in the codebook with the minimum distortion. Then in the proposed 2D-DCT-VQ method, the DCT is derived by discrete Fourier transform (DFT) which can be performed by fast Fourier transform (FFT) algorithm with the size of input signals of 2M×2N. Therefore, the computational complexity of 2D-DCT is O(N×log(8N)) at the M of 2. Moreover, the computational complexity of the supervised k-means-clustering-based vector quantization is O(k×L) by using full-research algorithm [21], where L is the length of the OFDM signals. Due to the samples reduced by 2D-DCT, the online quantization time can be saved proportionally.
In the digital RoF mobile fronthaul, spectral efficiency is usually determined by the signals’ transmission rate of the optical link. The more effective transmission rate, the higher spectral efficiency. The transmission rate R can be represented as R = fs × QBs, where fs is the sampling rate of original signals in analog-to-digital conversion and QBs is the quantization bits per sample. The R is proportional to QBs as fs is constant. Owing to the significantly reduced samples via using 2D-DCT, the equivalent QBs are lessened. Hence, the effective R is raised. Since the omitted high frequency coefficients are not sensitive to transmission quality, the spectral efficiency can be greatly improved. Besides, the QBs in k-means clustering algorithm is defined as QBs=log2k. For vector quantization, QBs=QBs/D with D is dimensionality. The D is 1 in scalar quantization. Therefore, the spectral efficiency can be further improved.
3. Experimental setup and results
Figure 3 is the experimental setup of the proposed digital RoF transmission system. At the Tx-DSP, the sampling rate per OFDM symbol is 122.88-MSa/s. The number of the subcarriers is 1200 with a subcarrier space of 60-kHz. 2048 inverse fast Fourier transform (IFFT)/FFT points are employed. The length of the cyclic prefix (CP) is 32. The length per ODFM symbol including CP is 16.93-μs. The M-ary quadrature amplitude modulation (M-QAM) orders for OFDM sub-channel vary among 4, 16, 64 and 256. And 13, 10, 8 and 6 72-MHz OFDM signals with 4, 16, 64 and 256-QAM are aggregated by multiplexing different A×C container frames in the time domain, respectively. The 2D-DCT-VQ algorithm detailed in the previous section is executed. The baud rate of the transmitted PAM-4 signals is 5-Gbaud/λ. Afterwards, an arbitrary waveform generator (AWG, 65-GSa/s) is used to download the PAM-4 signals. A light wave emitted by a narrow linewidth laser (∼1550-nm, 10-dBm) is injected into a Mach-Zehnder modulator (MZM, 40-GHz) and modulated by the amplified signals from the AWG. A 20-km standard SMF is utilized to deliver the optical signals. Then the optical signals are detected by a photodetector (PD, 15-GHz). A variable optical attenuator (VOA) placed before the PD is devoted to adjust the received optical power (ROP) for the system performance measurement. The received signals are captured by a digital storage oscilloscope (DSO, 40-GSa/s). At the Rx-DSP, a decision feedback equalizer with 41 feedforward taps and 7 feedback taps is used to recover the PAM-4 signals. The inverse 2D-DCT-VQ algorithm is executed and the signals are de-aggregation by TDM demodulation. The recovery OFDM signals are demodulated to baseband digital signals to analyze the error vector magnitude (EVM) and bit error rate (BER) performance of the system.
Fig. 3. Experimental setup of the proposed digital RoF transmission system. AWG: arbitrary waveform generator; MZM: Mach-Zehnder modulator; SMF: single-mode fiber; VOA: variable optical attenuator; PD: photodetector; DSO: digital storage oscilloscope; Tx-DSP/Rx-DSP: digital signal processing at the transmitter/ receiver; PC: personal computer.
Firstly, the system EVM performances related to the reserved ratio (ξ) of 2D-DCT coefficients and the rate of the training sequence (η) of 2D-vector quantization are analyzed as shown in Fig. 4 in 16QAM digital RoF system at QBs of 5. It can be seen from Fig. 4(a), when ξ is less than 58% and decreases, the EVM increases sharply because part of discarded frequency coefficients carry OFDM samples information. When ξ is greater than 58.25%, the EVM is almost stable with the BER of baseband digital signal equal to 0. And the more training sequence ratio η, the better EVM. The recovered constellation graphs at η of 0.05, 0.3 and 1 are shown in Figs. 4(b), 4(c) and 4(d) respectively, as ξ is 59%. When the η is more than and equal to 0.3, the EVM varies very slowly with BER of 0. Therefore, the ξ of 59% and η of 0.5 are taken in the experiments. For the ξ of 59%, namely the signal samples are compressed to 59%, thus the effective transmission rate of the system can be improved by 69.49%.
Fig. 4. (a) EVM performances versus the reserved ratio ξ of 2D-DCT coefficients and the rate of training sequence η for 2D-vector quantization. The recovered constellation graphs at (b) η=0.05, (c) η=0.3 and (d) η=1, as ξ=59%.
Then, the performances of the proposed 2D-DCT-VQ compression quantization scheme compared with the existing 2D vector quantization algorithm without compression [21] are researched in 16QAM 20-km standard SMF digital RoF transmission system as shown in Fig. 5. The detailed parameters for only 2D k-means clustering algorithm are the same as that for the proposed 2D-DCT-VQ scheme. From Fig. 5(a), the quantization performances are increasing with the increase of quantization bits per sample QBs. When QBs is greater than or equal to 5, the EVM of the proposed system is comparable to that only using 2D k-means clustering algorithm. But compression ratio of the system using 2D-DCT-VQ is superior to that using 2D k-means clustering as shown in Fig. 5(b). The compression ratio is defined as V/U×100% in which V and U denote the number of QBs with and without a compression method, respectively. The value of U is 15 in the standard of CPRI protocol. In proposed scheme, the value of V is 59%×QBs. Therefore, the spectral efficiency is vastly improved. Moreover, compared with the 2D k-means clustering algorithm without compression, the online quantization time using 2D-DCT-VQ also saves 41% because of the reduced samples.
Fig. 5. The EVM performances versus (a) the quantization bits per sample and (b) compression ratio of the system.
The BER curves of the PAM-4 signals for optical back-to-back (B2B) case and 20-km standard SFM transmission are measured as shown in Fig. 6. The BER of the B2B case is slightly better than that of the optic fiber transmission. The equalized eye diagrams at the ROP of -9 dBm and -14 dBm are shown as the insets (a) and (b) of Fig. 6, respectively. With the decrease of the ROP, the eye diagram is a little blurred. At the Reed-Solomon forward error correction (RS-FEC 528/514) limit of 5×10−4 [16], the ROP is about -13.08 dBm for SMF transmission. When the RS-FEC is used, error free transmission can be achieved for the BER less than 5×10−4.
Fig. 6. The BER curves of the PAM-4 signals versus the ROP for B2B and 20-km standard SMF transmission. Equalized eye diagrams of received PAM-4 signals at ROP of (a) -9 dBm and (b) -14 dBm for optic fiber transmission.
Finally, the BER curves and the EVM performances of OFDM baseband digital signals versus the ROP are shown in Figs. 7(a) and 7(b), respectively, for B2B and 20-km standard SFM transmission. The QBs are set as 4, 5, 6, and 7 for 4QAM, 16QAM, 64QAM and 256QAM, respectively. From the Fig. 7(a), the optic fiber short-reach transmission has a little impact to the system performance compared with the B2B case. And the higher the order of M-QAM, the larger the BER. From the Fig. 7(b), the EVM changes a little when the ROP is more than -11 dBm. As the ROP is less than -11dBm, the EVM is increasing greatly. According to the EVM thresholds of the 3GPP stipulation (17.5%, 12.5%, 8% and 3.5% for 4QAM, 16QAM, 64QAM, and 256QAM, respectively), the ROP for 4QAM, 16QAM, 64QAM and 256QAM in 20-km standard SFM D-RoF system are limited to about -13.71, -13.23, -12.51 and -11.14 dBm, respectively. The limited ROP can be further lowered when the FEC is introduced.
Fig. 7. The BER (a) and the EVM (b) of the OFDM baseband digital signals versus ROP for B2B and 20-km standard SMF transmission.
4. Conclusion
In summary, a digital RoF transmission scheme for mobile fronthaul using 2D-DCT-VQ is proposed to achieve high spectral efficiency. The operation of the proposed scheme is demonstrated experimentally for OFDM-modulated system with four different QAM mapping. In particular, the EVM performance of 256QAM is achieved below 2% at QBs of 7. Also, the system performance is comparable to that of existing approach (e.g., 2D k-means clustering algorithm) without compression. And the effective transmission rate can be improved by 69.49% as the OFDM samples are compressed to 59%. The proposed scheme provides an efficient solution for digital RoF transmission system which has great potential to be used in future fronthaul link of wireless network.
Funding
National Key Research and Development Program of China (No.2019YFB2203200); National Natural Science Foundation of China (No.61860206006, No.62075185); International S and T Cooperation Program of Sichuan Province (2021YFH0013); Wuhan National Laboratory for Optoelectronics (2018WNLOKF016).
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. P. Chanclou, L. A. Neto, K. Grzybowski, Z. Tayq, F. Saliou, and N. Genay, “Mobile fronthaul architecture and technologies: a RAN equipment assessment,” J. Opt. Commun. Netw. 10(1), A1–A7 (2018). [CrossRef]
2. C. Ranaweera, E. Wong, A. Nirmalathas, C. Jayasundara, and C. Lim, “5G C-RAN with optical fronthaul: an analysis from a deployment perspective,” J. Lightwave Technol. 36(11), 2059–2068 (2018). [CrossRef]
3. K. Tanaka and A. Agata, “Next-generation optical access networks for C-RAN,” in proceedings of the 2015 Optical Fiber Communications Conference and Exhibition (OFC), Los Angeles, CA, USA, 2015, paper Tu2E.1.
4. J. Marti and J. Capmany, “Microwave photonics and radio-over-fiber research,” J. Lightwave Technol. 10(4), 96–105 (2009). [CrossRef]
5. D. Liang and J. Bowers, “Recent Progress in Heterogeneous III-V-on-Silicon Photonic Integration,” Light Adv. Manuf. 2, 59 (2021).
6. X. Liu, H. Zeng, N. Chand, and F. Effenberger, “Experimental demonstration of high-throughput low-latency mobile fronthaul supporting 48 20-MHz LTE signals with 59-Gb/s CPRI-equivalent rate and 2-μs processing latency,” in proceedings of the 2015 European Conference on Optical Communication (ECOC), Valencia, Spain, 2015; pp. 1–3.
7. D. Novak, R. B. Waterhouse, A. Nirmalathas, C. Lim, P. A. Gamage, T. R. Clark, T. R. Clark, M. L. Dennis, and J. A. Nanzer, “Radio-over-fiber technologies for emerging wireless systems,” IEEE J. Quantum Electron. 52(1), 1–11 (2016). [CrossRef]
8. T. Pfeiffer, “Next generation mobile fronthaul architectures,” in proceedings of the 2015 Optical Fiber Communication Conference (OFC), Los Angeles, CA, USA, 2015, paper M2J.7.
9. “Common Public Radio Interface (CPRI); Interface Specification,” C. P. R. I. Specification, V7.0, 2015.
10. S. H. Kim, H. S. Chung, and S. M. Kim, “Experimental demonstration of CPRI data compression based on partial bit sampling for mobile front-haul link in C-RAN,” in proceedings of the 2016 Optical Fiber Communications Conference and Exhibition (OFC), Anaheim, CA, USA, 2016, paper W1H.5.
11. J. Wang, Z. Yu, K. Ying, J. Zhang, F. Lu, M. Xu, and G. K. Chang, “Delta-Sigma modulation for digital mobile fronthaul enabling carrier aggregation of 32 4G-LTE / 30 5G-FBMC signals in a single-λ 10-Gb/s IM-DD channel,” in Proceedings of the 2016 Optical Fiber Communications Conference and Exhibition (OFC), Anaheim, CA, USA, 2016, paper W1H.2.
12. L. Breyne, G. Torfs, X. Yin, P. Demeester, and J. Bauwelinck, “Comparison between analog radio-over-fiber and sigma delta modulated radio-over-fiber,” IEEE Photonics Technol. Lett. 29(21), 1808–1811 (2017). [CrossRef]
13. L. Zhang, X. Pan, O. Oskars, U. Aleksejs, S. Richard, W. Urban, J. Gunnar, P. Sergei, W. Lena, and S. Xiao, “Digital mobile fronthaul employing differential pulse code modulation with suppressed quantization noise,” Opt. Express 25(25), 31921–31936 (2017). [CrossRef]
14. H. Li, X. Li, and M. Luo, “Improving performance of differential pulse coding modulation based digital mobile fronthaul employing noise shaping,” Opt. Express 26(9), 11407–11417 (2018). [CrossRef]
15. M. Xu, Z. Jia, J. Wang, L. A. Campos, and G. Chang, “Statistical data compression and differential coding for digital radio-over-fiber-based mobile fronthaul,” J. Opt. Commun. Netw. 11(1), A60–A71 (2019). [CrossRef]
16. M. Xu, F. Lu, J. Wang, L. Cheng, D. Guidotti, and G.-K. Chang, “Key technologies for next-generation digital RoF mobile fronthaul with statistical data compression and multiband modulation,” J. Lightwave Technol. 35(17), 3671–3679 (2017). [CrossRef]
17. M. Xu, X. Liu, N. Chand, F. Effenberger, and G.-K. Chang, “Fast statistical estimation in highly compressed digital RoF systems for efficient 5G wireless signal delivery,” in proceedings of the 2017 Optical Fiber Communications Conference and Exhibition (OFC), Los Angeles, CA, USA, 2017, paper M3E.7.
18. L. Zhang, A. Udalcovs, R. Lin, O. Ozolins, X. Pang, L. Gan, R. Schatz, M. Tang, S. Fu, D. Liu, W. Tong, S. Popov, G. Jacobsen, W. Hu, S. Xiao, and J. Chen, “Toward terabit digital radio over fiber systems: architecture and key technologies,” IEEE Commun. Mag. 57(4), 131–137 (2019). [CrossRef]
19. H. Li, M. Luo, X. Li, and S.H. Yu, “Spectrally efficient digital mobile fronthaul with discrete cosine transform and multi-band quantization,” Opt. Lett. 43(20), 5130–5133 (2018). [CrossRef]
20. J. Ye, Y. Guo, P. Li, L. Yan, W. Pan, and X. Zou, “2-D quantization scheme utilizing SOFM neural network clustering for a D-RoF system,” Opt. Lett. 43(19), 4663–4666 (2018). [CrossRef]
21. L. Zhang, X. Pang, O. Oskars, U. Aleksejs, P. Sergei, S. Xiao, W. Hu, and J. Chen, “Spectrally efficient digitized radio-over-fiber system with k-means clustering-based multidimensional quantization,” Opt. Lett. 43(7), 1546–1549 (2018). [CrossRef]
22. J. Ye, Y. Guo, P. Li, L. Yan, W. Pan, and X. Zou, “Improving performance of digital mobile fronthaul employing 2-D vector quantization with vector linear prediction,” IEEE Photonics J. 11(2), 7201911 (2019). [CrossRef]
23. H. Ren, P. Yu, J. Lu, M. Gao, Z. Qin, S. Guo, and W. Hu, “Multidimensional vector quantization-based fast statistical estimation in compressed digitalized radio-over-fiber systems,” Appl. Opt. 58(13), 3418–3425 (2019). [CrossRef]
