In orthogonal frequency division multiple access based passive optical network (OFDMA-PON) uplink, synchronization between optical network units (ONUs) is very important to maintain orthogonality. The synchronization among uplink signals is considered as one of the main challenges in OFDMA-PON due to optical path difference. In this paper, the performance degradation according to timing offset between ONUs is experimentally analyzed. And we propose and demonstrate timing offset effect reduction in asynchronous multiple access by using CP extension and filter bank based multicarrier (FBMC) system in intensity modulation/direct detection (IM/DD) based OFDMA-PON uplink transmission.
© 2015 Optical Society of America
Data traffic have been explosively increased in wireless access network due to proliferation of mobile devices and applications. Moreover, this tendency will be intensified because of the internet of things (IoT) which is one of the technical requirements in 5th generation (5G) wireless communications [1,2]. In order to support a huge data traffic of wireless access, a spectral efficient optical transmission technique is required in optical access network.
Orthogonal frequency division multiplexing (OFDM) is a highly spectral efficient parallel transmission technique, which is based on overlapping among orthogonal subcarriers. It is robust against dispersion and it enables to equalize channel effects by using a simple single tap equalizer. Moreover, it could maximize transmission capacity by using adaptive modulation. Thus, OFDM is standard of the long-term-evolution-advanced (LTE-A) in wireless access. Furthermore, OFDM based optical transmission has been actively researched in optical access network.
In orthogonal frequency division multiple access (OFDMA) based passive optical network (PON), bandwidth is assigned to optical network units (ONUs) by allocating subcarriers. So, frequency and time synchronization are required among ONUs to maintain orthogonal condition. It is relatively simple in downlink because a whole OFDM frame is generated and optically modulated in a single optical line terminal (OLT). However, in case of uplink transmission, signals are independently generated and modulated in several ONUs, and then these signals pass through different optical channel before optical combining at optical distribution network (ODN). In this OFDMA-PON uplink transmission, two important issues arise due to optical path difference; optical beating interference (OBI) and timing offset between ONUs [3–10]. In order to avoid OBI, several techniques were proposed such as spectrum broadening [3,4], coherent receiving , and wavelength separation [6–8], but there was a few research about timing offset reduction [9,10]. In optical communications, optical signals transmit through a longer distance and have a larger bandwidth compared to wireless communications. Thus, a relatively large sample delay could be occurred even with a small optical path difference. Although OFDM is essentially robust against multi-path fading, it is only effective when delay is located within a cyclic prefix (CP) length, or inter symbol interference (ISI) and inter channel interference (ICI) will be generated if delay is larger than CP length. So, a large CP is required in OFDMA-PON uplink transmission to be robust against optical path difference even short range transmission, which decreases spectral efficiency (SE). In PON system, time synchronization at ODN is hard to practically realize. In order to demodulate individual uplink signal with single FFT, timing advance technique [10,11] is required based on additional information about channel delay between ONUs and OLT to receive synchronized signal, which imposes overhead burden on system. In [9,10], OFDMA upstream signals were processed with an individual FFTs for asynchronous reception, which could be avoid ISI. However, ICI still remains leading performance degradation especially in boundary subcarrier. Moreover, performance degradation due to ISI and ICI is varied according to CP length. To avoid this interference, a large frequency guard band is required between ONUs, but it reduces SE, and this degradation will be more severe when the number of multiplexed ONUs is increased.
Filter bank based multicarrier (FBMC) is composed by set of filters respond to OFDM subcarriers . By using FBMC, sinc-shaped sidelobes of OFDM due to rectangular FFT window could be effectively suppressed. The amount of sidelobes suppression is decided by a shape of prototype filter. By comparing characteristics of filters in multicarrier systems , it could be recognized that Mirabbasi-Martin filter has the best spectrum confinement while satisfying orthogonality among subcarriers compared to known filters. In our group, it was experimentally verified that wired/wireless converged optical transmission by employing FBMC based on Mirabbasi-Martin filter for the first time in optical communications . In this work, the improvements were reported by reducing interference between heterogeneous signals. However, because the main observation of this work is an optical downlink transmission, only downlink issues (like RF delay, RF power imbalance) were included, but uplink issue in multiple access (like optical path delay) was not included.
In this paper, in OFDMA-PON uplink transmission, the performance degradation of asynchronous reception based on individual FFT according to timing offset was experimentally evaluated by varying optical path difference between ONUs. It is verified that multiple access interference (MAI) could be reduced in asynchronous reception by extending CP length and minimum required CP depends on optical path difference. Furthermore, FBMC based sidelobes suppression was applied in OFDMA-PON to ensure timing offset robustness by avoiding ISI and ICI in asynchronous reception. Thus, after 23km SMF transmission, performance enhancement was demonstrated in terms of EVM by effectively reducing MAI in OFDMA-PON uplink transmission with optical path difference between ONUs.
Figure 1 illustrates a basic structure of optical OFDMA in IM/DD based PON system. Each ONU is independently modulated at different location with unequal electro-optic (E/O) conversion efficiency. After E/O conversion, the asynchronous optical signals are passing through different optical channels and passively combined at ODN without synchronization.
In OFDMA, each ONU could be allocated a signal bandwidth by assigning subcarriers. Each ONU optically independently modulates signal with their best effort in terms of bias control, modulation depth and polarization control. In this case, the device response and conversion efficiency may not be identical for each user. In order to maximize throughput while ensuring a quality of service (QoS), an adaptive modulation is employed in OFDMA . At first, OFDM without CP is used in experiment to investigate performance degradation due to timing offset excluding CP effect. The timing offset effect is observed by changing RF delay with a small variation. And then, in order to reduce MAI caused by timing offset between ONUs, CP extended OFDM is used in OFDMA-PON uplink transmission. After IFFT process, CP was inserted in OFDM signal, and performance improvement in terms of EVM is investigated by varying CP length in OFDMA-PON uplink transmission including optical path difference. The CP length is selected as 4, 8, 16, 32, 64, and 128 samples, which is corresponding to 1/64, 1/32, 1/16, 1/8, 1/4, and 1/2 of symbol duration, respectively. Moreover, FBMC based digital filtering is used to further reduce timing offset effect by suppressing sidelobes of OFDM, which was employed in our previous work . The Mirabbasi-Martin filter is utilized , which shows the best property in sidelobes suppression among the known filters with a fixed number of coefficients . To realize FBMC based digital filtering, frequency domain oversampling is required, but it increases FFT size by several times according to overlapping factor (K) of filter. The size of FFT is the most computational part of DSP in transceiver. Thus, we employed polyphase network (PPN)  to replace frequency domain oversampling and convolution of filter coefficients by time domain repetition and multiplication of filter coefficients. Moreover, an overlap & sum process  was utilized to maintain comparable symbol duration with OFDM because digital filtering extends symbol duration within same signal bandwidth. Therefore, by using PPN and an overlap & sum process, FFT size and symbol duration could be maintained compared to OFDM case while employing digital filtering. Although FBMC could suppress sinc-shaped sidelobes of OFDM, an orthogonal condition between adjacent subcarriers can be broken. To maintain orthogonal condition between contact subcarriers, an offset QAM (OQAM) was also employed in FBMC generation. In OQAM, each subcarrier is assigned alternatively with real and imaginary part of QAM signal to ensure orthogonal condition. Therefore, orthogonality among subcarriers is totally maintained as like OFDM. However, OQAM modulation reduces transmission capacity and spectral efficiency by half compared to QAM modulation. In order to maintain capacity of OQAM as same as QAM, an offset of the half-symbol-duration was also applied in FBMC. In Fig. 2, τ0 is a symbol-duration. The OQAM has repetitive characteristic in time domain by half-symbol-duration (τ0/2), and it has orthogonality between sine and cosine plane. By using these properties of OQAM, as illustrated in Fig. 2, OQAM signal were mapped inversely every τ0/2 and overlapped to maintain total capacity while ensuring orthogonality between adjacent subcarriers. Thus, by employing these techniques (PPN, an overlap & sum, OQAM, and offset of half-symbol duration), the total capacity and computational complexity of FBMC could be comparable with those of OFDM within same signal bandwidth while suppressing sidelobes. Moreover, by using FBMC, the output signal could be suppressed sidelobes in frequency domain as well as suppressed edge of pulse in time domain. Therefore, FBMC could relax interference between ONUs and have dispersion robustness property even without CP.
The experimental setup is illustrated in Fig. 3 for optical multiple access uplink transmission in IM/DD based PON system. By using offline process, OFDM and FBMC signals were generated with 1.5GHz signal bandwidth. Under same condition, OFDM and FBMC signals were equally generated where no CP is used in FBMC case. In order to multiplex uplink signals from multiple participants, each ONUs was nulled-out the subcarriers assigned to another participants. The uplink signals were independently modulated at separated ONUs with best effort by using Mach-Zehnder modulator (MZM). Electro-optic (E/O) conversion was individually optimized by controlling bias point, modulation depth and polarization. The modulation efficiency was different between ONUs due to devices characteristics. The ONU1 and ONU2 had optical source of 1546.46 nm- and 1546.86 nm-wavelength, respectively. In order to avoid OBI, optical sources with wavelength difference were used [6–8], which enables to this work focus only on degradation of timing offset according to optical path difference. A further research is in progress when OBI exist in colorless OFDMA-PON structure. After optical modulation, the uplink signals from individual ONU passes through different optical path and then passively combined at optical 3 dB coupler for a 23.2km single mode fiber (SMF) transmission. The experiment was conducted based on IM/DD. It was assumed that the asynchronous uplink signals from multiple ONUs are combined at ODN without synchronization or balancing. In this scenario, the effect on OFDMA was analyzed according to timing offset, which was experimented in terms of RF signal offset before E/O conversion and optical offset after E/O conversion. In order to obtain asynchronous RF signals, AWG output samples of ONU1 were delayed. Moreover, 1m-, 2m- and 5m-length optical fiber were inserted before optical combining, to evaluate performance degradation caused by optical path difference without RF sample delay. The performance was evaluated according to path difference in terms of channel EVM based on preamble. In our experiment, channel EVM is an objective measurer which decides total transmission capacity because adaptive modulation was employed based on a preamble feedback information. The input power of photodetector (PD) was maintained as −8dBm for a fair performance evaluation. The number of subcarriers was 128, and the IFFT/FFT size was 256 for Hermitian symmetry in OFDM signal generation. Before adaptive modulation, the training sequence was modulated by 4 QAM (2bits). Cyclic prefix (CP) was inserted in OFDM and an EVM improvement was analyzed according to CP length when timing offset was occurred. Moreover, in order to further improve EVM, FBMC was employed under same condition with OFDM excepting digital filtering and CP insertion; in FBMC, no CP was inserted, which could more increase SE compared to OFDM. For realizing FBMC, the frequency coefficients of the filter for overlapping factor K = 4 (H0, H1, H2, H3) were 1, 0.971960, 1/21/2, and 0.235147, respectively. In experiment, because PPN based time domain filtering was processed to maintain FFT size, frequency coefficients and convolution were replaced by time coefficients (IFFT of frequency coefficients) and multiplication. At receiver, individual FFT [9,10] based asynchronous reception was used to each ONUs to avoid ISI, which was realized by using offline-process. After FFT, a single tap equalizer was realized to equalize the channel effects based on preamble in both OFDM and FBMC at receiver side.
4. Results and discussions
In Fig. 4(a), channel EVM of OFDM subcarrier is presented according to timing offset caused by RF signal delay when CP was not inserted. Mean EVM is plotted in Fig. 4(b) by averaging subcarriers around boundary (from 61st to 67th) excepting a boundary subcarrier (64th) as a frequency guard, which is corresponding to Fig. 4(a). Before E/O conversion, RF delay was varied in ONU1 side. In this experiment, one OFDM symbol had 256 samples due to IFFT size unless CP was inserted. Thus, RF delay has cyclic property every 256 samples period. In Fig. 4(a), a bright area means a degraded EVM due to MAI caused by timing offset. If asynchronous uplink signal is received at OLT, MAI would be generated. In ONU1 reception window, a received symbol of ONU2 is not sufficient, i.e. part of symbol(i) and part of symbol(i + 1) of ONU2 are simultaneously received in ONU1 window. This insufficient symbol of ONU2 leads subcarrier spectrum to be stretched as illustrated in Fig. 1(b). Inversely, in ONU2 reception window, the received ONU1 signal is also stretched as illustrated Fig. 1(c), which breaks orthogonality and causes ICI (MAI). On the other hand, without timing offset, no MAI would be generated due to an orthogonality as like in Fig. 1(a). However, as shown in Figs. 4(a) and 4(b), MAI was generated in this work even without RF sample delay. Moreover, a smaller MAI could be observed when ONU1 was delayed about 250 samples compared to other cases. More precisely, the MAI is the smallest when ONU1 was delayed by 252 samples as shown in Fig. 4(c). In other words, uplink signals could be synchronized by bring ONU2 forward by 4 samples due to cyclic property in this work. Even though an intended optical fiber was not inserted, uplink channels practically had optical path difference over several meters unless optical fiber length was finely tuned. In our work, because AWG sampling was 3GS/s due to a 1.5GHz signal bandwidth without oversampling, one sample would be delayed by 10cm optical difference, so the optical path difference could be estimated about 40cm in Figs. 4(a)-4(c) even without RF delay. Moreover, this sample delay would be more severe if signal bandwidth is larger compared to this work; commonly, an enormous signal bandwidth is used in optical communications.
In order to reduce MAI caused by timing offset between ONUs, we inserted CP in OFDM before transmission. Figures 5(a)-5(d) show channel EVM of subcarriers according to CP length when optical path difference was varied by inserting optical fiber before optical combining. It represents some cases, 5(a) no optical fiber insertion, 5(b) 1m fiber insertion in ONU2 side, 5(c) 2m fiber insertion in ONU1 side, and 5(d) 5m fiber insertion in ONU2 side, respectively. In all cases, MAI was generated especially at boundary region. For comparison, EVM of OFDM without MAI was plotted in Fig. 5(a) when EVM of ONU1 was evaluated without ONU2 transmission, and vice versa. Without MAI, EVM of OFDM subcarriers would be only affected by SNR. In case of Fig. 5(a), CP with 8 samples (1/32 of symbol duration) was required to suppress MAI, and there was no performance improvement even though CP was increased as 16 samples (1/16 of symbol). Thus, an optimal required CP length for suppressing MAI caused by timing offset is different in cases 5(a)-5(d). Moreover, CP over optimum length has no effect on performance improvement, rather it decreases spectral efficiency due to increased overhead. Furthermore, EVM is improved only in one ONU while another ONU has almost no improvement even with optimum CP insertion. As illustrated in Fig. 6, if a delay is smaller than CP length, a received symbol of a late ONU is sufficient in an early ONU reception window. Thus, the orthogonal condition could be satisfied in an early ONU reception window. However, in a late ONU window, part of symbol(i) and CP of symbol(i + 1) of an early ONU are received, which leads subcarrier spectrum to be stretched and to generate MAI as illustrated in Fig. 6. From the Figs. 4 results, it could be understood that ONU1 arrived earlier about 4 samples (about 40cm shorter optical fiber) than ONU2. In case of Fig. 5(a), by inserting CP with 8 samples (1/32 of symbol), delay could be located within CP length and MAI could be reduced in ONU1 (an early ONU), which leads EVM of ONU1 to approach to result of without MAI case. With same manner, in Figs. 5(b)-5(d), it could be observed that an optimal CP length is different according to optical path difference. Moreover, only an early ONU could be improved as similar to without MAI case even though optimal CP is inserted with a sacrifice of SE.
In order to further reduce MAI while maintaining SE, FBMC was employed in OFDMA-PON uplink transmission. In Figs. 7(a) and 7(b), RF spectrum of OFDMA and FBMC-MA are represented when ONU2 was modulated with 4 QAM without ONU1 transmission. In case of OFDM, sinc-shaped sidelobes of ONU2 is obviously observed as shown in Fig. 7(a). The sidelobes of OFDM degrade SNR of ONU1 due to a raised noise floor and generate interference unless ONUs are perfectly synchronized in terms of time and frequency. By using FBMC, sidelobes could be effectively suppressed as shown in Fig. 7(b). Theoretically, the amount of sidelobes attenuation of FBMC is very large about 60dB. In Fig. 7(b), sidelobes of FBMC is located under noise floor; thus, the spectrum of FBMC could be more improved if the transmission channel has a low noise.
In Fig. 8(a), channel EVM of FBMC subcarriers is presented according to timing offset caused by RF signal delay. Experimental condition was same with that of OFDM. In case of FBMC, because an observed channel EVM was similarly maintained although changing RF sample delay, only some cases were presented in this paper. By comparing Figs. 4(a) and 8(a), MAI (a bright region) could be effectively suppressed at boundary subcarriers, and overall channel EVM could be improved because of a lower noise floor due to sidelobes suppression when FBMC was applied in optical uplink transmission. Moreover, Fig. 8(b) represents a corresponding mean EVM of subcarriers located around boundary region, which is plotted with Fig. 4(b) for comparing with OFDMA case. In case of FBMC-MA, it has a flat result in terms of mean channel EVM even though RF delay was changed, which means that it is more robust against RF timing offset compared to OFDMA.
Channel EVM of FBMC subcarrier is shown in Fig. 9 according to timing offset caused by varying optical path difference. Along with the above OFDM experiment, 1m-, 2m-, and 5m-length optical fiber was inserted before optical coupler to make timing offset caused by optical path. Unlike the OFDM experiment, CP length was not varied because FBMC doesn’t have to insert CP. As shown in Fig. 9, a resultant performance is nearly flat in a whole signal bandwidth, and there is no significant difference among the observed cases. The result of FBMC has a similar feature to the OFDMA without MAI case presented in Fig. 5(a), excepting a high frequency subcarriers. In case of OFDM, a high frequency subcarriers were degraded by ICI caused by sampling error because an oversampling was not used in synchronization process in this work. On the other hand, EVM in high frequency subcarriers could be improved by using FBMC, because it could be robust against sampling error due to filtering effect. By using FBMC based MAI suppression, EVM performance is further improved in both an early ONU and a late ONU (in Fig. 9), which is clearly differ from CP extended OFDMA case (in Figs. 5(a)-5(d)). Again, in case of OFDMA, only an early ONU could be improved even with CP extension, which waste efficiency of spectrum utilization. The channel improvement of FBMC enables subcarriers to allocate more bits when water-filling algorithm based adaptive modulation is employed, which could increase total transmission capacity and SE. Furthermore, no CP is required in FBMC-MA, which more increases SE compared to OFDMA. Moreover, in contrast with a CP extended OFDM, an optimal CP length is no longer a significant consideration in optical uplink transmission. Thus, FBMC-MA is more robust against optical timing offset even without CP compared to a CP extended OFDMA.
Although two ONUs were transmitted as uplink signals in this experiment, practically the number of multiple ONUs will be large. In both FBMC and OFDM, complexity will be linearly increased according to number of ONUs because individual FFT based asynchronous reception is requires independent process for multiple ONUs. However, signal performance of OFDM will be severely degraded because MAI increased according to number of ONUs while maintaining system conditions, such as, total signal bandwidth, power spectral density, and number of subcarriers . Thus, without a tight synchronization process, signal-to-interference ratio (SIR) will be degraded in OFDM, but not in FBMC. In order to maintain signal performance of OFDMA as good as that of FBMC even with multiple ONUs, additional interference cancellation technique is required in OLT. However, this process will be very complex due to iterations while FBMC has only moderate complexity in filtering without iterations. Therefore, FBMC-MA will be more effective compared to OFDMA for the large number of asynchronous ONUs.
In summary, the performance degradation according to timing offset between uplink signals was experimentally investigated by varying both RF and optical delay in IM/DD based OFDMA-PON system uplink transmission. Timing offset between ONUs causes MAI in OFDMA-PON uplink transmission even using individual FFT based asynchronous reception. In order to reduce MAI caused by timing offset, CP extended OFDM was employed and MAI reduction effect was evaluated according to CP length. In a CP extended OFDM, MAI could be reduced with an optimal CP length at the sacrifice of SE. However, only an early ONU was improved and an over length CP has no effect on channel improvement. To further reduce MAI while maintaining a high SE, FBMC was proposed in OFDMA-PON uplink transmission. By using FBMC-MA, the MAI of both an early ONU and a late ONU could be effectively reduced even without CP, which could improve overall channel EVM. This enables subcarriers to allocate more bits, which could increase transmission capacity and SE. It is experimentally verified that FBMC-MA is more robust against timing offset between uplink signals even without CP compared to CP extended OFDM. Thus, FBMC based asynchronous reception would be more effective in timing offset reduction in OFDMA-PON uplink transmission.
This work was supported by the ICT R&D programs of MSIP/IITP, Republic of Korea. [R0101-15-0086, Research on 25Gbps optical access network based on discrete multitoned enabling dynamic network resource management].
References and links
1. G. Wunder, P. Jung, M. Kasparick, T. Wild, F. Schaich, Y. Chen, S. Brink, I. Gaspar, N. Michailow, A. Festag, L. Mendes, N. Cassiau, D. Kténas, M. Dryjanski, S. Pietrzyk, B. Eged, P. R. Vago, and F. Wiedmann, “5GNOW: Non-orthogonal, asynchronous waveforms for future mobile applications,” IEEE Commun. Mag. 52(2), 97–105 (2014). [CrossRef]
2. G. Wunder, M. Kasparick, S. ten Brink, F. Schaich, T. Wild, I. Gaspar, E. Ohlmer, S. Krone, N. Michailow, A. Navarro, G. Fettweis, D. Ktenas, V. Berg, M. Dryjanski, S. Pietrzyk, and B. Eged, “5GNOW: Challenging the LTE design paradigms of orthogonality and synchronicity,” http://arxiv.org/abs/1212.4034 [CrossRef]
3. S. M. Jung, S. M. Yang, K. H. Mun, and S. K. Han, “Optical beat interference noise reduction by using out-of-band RF clipping tone signal in remotely fed OFDMA-PON link,” Opt. Express 22(15), 18246–18253 (2014). [CrossRef] [PubMed]
4. H. D. Jung, Y. Y. Won, and S. K. Han, “Optical beat noise reduction in WDM-SCM access networks using modulated optical pulse train,” IEEE Photon. Technol. Lett. 17(10), 2215–2217 (2005).
5. D. Qian, N. Cvijetic, J. Hu, and T. Wang, “A novel OFDMA-PON architecture with source-free ONUs for next-generation optical access networks,” IEEE Photon. Technol. Lett. 21(17), 1265–1267 (2009). [CrossRef]
6. L. A. Neto, G. B. Farias, N. Genay, S. Menezo, B. Charbonnier, P. Chanclou, and C. A. Berthelemot, “On the limitations of IM/DD WDM-FDMA-OFDM PON with single photodiode for upstream transmission,” in Optical Fiber Communication Conference and Exposition, and the National Fiber Optic Engineers Conference 2012, OFC/NFOEC Technical Digest (Optical Society of America, 2012), paper OW4B.1.
7. W. Q. Jin and J. M. Tang, “Experimental investigations of wavelength spacing and colorlessness of RSOA-based ONUs in real-time optical OFDMA PONs,” J. Lightwave Technol. 30(16), 2603–2609 (2012). [CrossRef]
8. I. Cano, M. C. Santos, V. Polo, F. X. Escayola, and J. Prat, “Dimensioning of OFDMA PON with non-preselected independent ONUs sources and wavelength-control,” Opt. Express 20(1), 607–613 (2012). [CrossRef] [PubMed]
9. J. von Hoyningen-Huene, H. Griesser, M. H. Eiselt, and W. Rosenkranz, “Experimental demonstration of OFDMA-PON uplink-transmission with four individual ONUs,” in Optical Fiber Communication Conference and Exposition, and the National Fiber Optic Engineers Conference 2013, OFC/NFOEC Technical Digest (Optical Society of America, 2013), paper OTh3A.2. [CrossRef]
10. J. von Hoyningen-Huene, H. Griesser, M. H. Eiselt, and W. Rosenkranz, “Asynchronous signal reception in OFDMA-PON-uplink,” in Advanced Photonics 2013, Advanced Photonics Congress (Optical Society of America, 2013), paper SP4D.2. [CrossRef]
11. E. Dahlman, S. Parkvall, and J. Skold, 4G: LTE/LTE-Advanced for Mobile Broadband, 1st ed. (Academic, 2011).
12. M. Bellanger, D. LeRuyet, D. Roviras, M. Terré, J. Nossek, L. Baltar, Q. Bai, D. Waldhauser, M. Renfors, T. Ihalainen, A. Viholainen, T. H. Stitz, J. Louveaux, A. Ikhlef, V. Ringset, H. Rustad, M. Najar, C. Bader, M. Payaro, D. Katselis, E. Kofidis, L. Merakos, A. Merentitis, N. Passas, A. Rontogiannis, S. Theodoridis, D. Triantafyllopoulou, D. Tsolkas, D. Xenakis, M. Tanda, T. Fusco, M. Huchard, J. Vandermot, A. Kuzminskiy, F. Schaich, P. Leclair, and A. Zhao, “FBMC physical layer: a primer,” http://www.ict-phydyas.org.
13. A. Sahin, I. Guvenc, and H. Arslan, “A Survey on multicarrier communications: prototype filters, lattice structures, and implementation aspects,” IEEE Comm. Surv. and Tutor. 16(3), 1312–1338 (2014). [CrossRef]
14. S. Y. Jung, S. M. Jung, and S. K. Han, “AMO-FBMC for asynchronous heterogeneous signal integrated optical transmission,” IEEE Photon. Technol. Lett. 27(2), 133–136 (2015). [CrossRef]
15. E. Giacoumidis, J. L. Wei, X. L. Yang, A. Tsokanos, and J. M. Tang, “Adaptive-modulation-enabled WDM impairment reduction in multichannel optical OFDM transmission system for next-generation PONs,” IEEE Photon. J. 2(2), 130–140 (2010). [CrossRef]
16. R. Fantacci, D. Marabissi, and S. Papini, “Multiuser interference cancellation receivers for OFDMA uplink communications with carrier frequency offset,” in Proceedings of IEEE Global Telecommunications Conference 2004 (IEEE Communications Society, 2004), vol. 5, pp. 2808–2812. [CrossRef]