Improved Performances in Point-to-Multipoint Flexible Optical Transceivers Utilizing Cascaded Discrete Fourier Transform-Spread Inverse Fast Fourier Transform/Fast Fourier Transform-Based Multi-Channel Aggregation/De-Aggregation

Abstract

The previously proposed cascaded inverse fast Fourier transform/fast Fourier transform (IFFT/FFT)-based point-to-multipoint (P2MP) flexible optical transceivers have the potential to equip future intensity modulation and direct detection (IMDD) optical access networks with excellent flexibility, adaptability, scalability and upgradability. However, due to their cascaded IFFT-based multi-channel aggregations, P2MP flexible transceivers suffer high peak-to-average power ratios (PAPRs). To address the technical challenge, this paper proposes a novel P2MP flexible optical transceiver, which uses a cascaded discrete Fourier transformation-spread (DFT-Spread) IFFT/FFT-based multi-channel aggregation/de-aggregation and standard signal clipping to jointly reduce its PAPRs. The upstream performances of the proposed transceivers are numerically explored in a 20 km IMDD upstream passive optical network (PON). The results indicate that the proposed transceiver’s PAPRs are mainly dominated by the size of the last IFFT operation of the multi-channel aggregation, and are almost independent of modulation format and channel count. Compared to conventional cascaded IFFT/FFT-based P2MP transceivers with and without clipping operations, the proposed DFT-Spread P2MP transceivers can reduce PAPRs by 2.6 dB and 3.5 dB, respectively, for a final IFFT operation size of 1024. More significant PAPR reductions are achievable when the last IFFT operation size is increased further. As a direct result, compared to conventional P2MP transceivers adopting clipping operations only, the proposed transceiver can improve upstream receiver sensitivities by >1.9 dB and the aggregated upstream transmission capacities by >14.1%. Such aggregated upstream transmission capacity enhancements are independent of channel count and become more pronounced for longer transmission distances.

1. Introduction

Due to the cost-sensitive nature of mobile access networks, intensity modulation and direct detection (IMDD) passive optical networks (PONs) have been widely considered as a cost-effective solution for practically implementing next generation mobile networks [1]. Existing IMDD PONs utilize conventional point-to-point (P2P) optical transceivers, and each connection is established using a pair of dedicated transceivers operating at the same speeds. Considering the hub-and-spoke network architectures of conventional PONs and mobile networks (e.g., tree network topology), such fixed-speed P2P connections limit network flexibility and network resource utilization efficiency [2]. Such limitations unavoidably become more serious in the 5G-and-beyond era, where numerous new network applications/services, with largely diversified requirements in terms of bandwidth, latency, etc., [3,4] emerge.

To solve the technical challenge cost-effectively, recently, a point-to-multipoint (P2MP) flexible optical transceiver has been discussed, which allows an arbitrary number of low-speed optical transceivers at individual optical network units (ONUs) to concurrently communicate with a high-speed transceiver at the optical line terminal (OLT) [5]. In the P2MP flexible optical transceiver, two enabling techniques are employed, i.e., orthogonal digital filtering and the cascaded inverse fast Fourier transform/fast Fourier transform (IFFT/FFT)-based multi-channel aggregation/de-aggregation. The former technique utilizes frequency division multiplexing (FDM) to allocate different low-speed ONU transceivers at different radio frequency spectral regions for communicating with the high-speed OLT optical transceiver. It potentially equips PONs with excellent upgradability and scalability, and further mitigates the latency and jitter issues of time division multiplexing (TDM) networks [6]. The latter technique ensures that each low-speed ONU transceiver can adaptively establish an arbitrary number of connections (or channels) with the high-speed OLT transceiver according to actual network application/service requirements. For each ONU transceiver, the occupied radio frequency bandwidth, the overall channel count and each channel parameter (bitrate, modulation format, etc.) can be flexibly and dynamically changed by configuring transceiver digital signal processing (DSP) [7].

However, due to the aggregation of multiple independent channels, each consisting of multiple subcarriers, the P2MP-flexible transceiver unavoidably suffers high peak-to-average power ratios (PAPRs). Consequently, the P2MP transceiver requires embedded electrical/optical devices to have large dynamic ranges. In addition, high PAPRs may also lead to transmission performance degradations because of reduced signal-to-noise ratios and enhanced sensitivity to transmission system nonlinearity [8,9,10]. To lower the PAPRs of the P2MP transceivers, conventional signal clipping techniques can be utilized, whose effectiveness is, however, limited due to clipping-induced signal distortions [11]. Other existing PAPR reduction techniques, including selective mapping (SLM) [12] and partial transmit sequences (PTSs) [13] are also not suitable for the P2MP-flexible transceivers. This is because the above techniques were originally designed for static orthogonal frequency division multiplexing (OFDM) transmission systems. Integrating them into the dynamical and flexible multi-channel aggregation operations of the P2MP transceivers can unavoidably lead to a high DSP complexity.

To efficiently address the PAPR issues of the P2MP transceivers, in this paper, a novel P2MP-flexible transceiver is proposed, which utilizes cascaded discrete Fourier transform-spread (DFT-Spread) IFFT/FFT operations to reduce the PAPRs and realize multi-channel aggregation/de-aggregation procedures. In addition, to further reduce the PAPRs, signal clipping operations are also applied after the orthogonal digital filtering operations.

The results of this paper show that the PAPRs of the proposed DFT-Spread P2MP transceivers are determined by the last IFFT operation size at the multi-channel aggregation procedure, and are almost independent of modulation format and channel count. Compared to the cascaded IFFT-based multi-channel aggregation operation adopted in the previously reported P2MP transceivers, the proposed cascaded DFT-Spread IFFT-based multi-channel aggregation operation can effectively reduce the PAPRs by >1.4 dB for the last IFFT operation sizes of >32. When the last IFFT sizes are increased to 4096, a more significant PAPR reduction of >2.2 dB can be achievable.

In 20 km upstream IMDD PONs, the performances of the proposed DFT-Spread P2MP transceivers are explored numerically. The results show that for a given last IFFT operation size of 1024, the proposed transceivers can reduce their PAPRs by ~3.5 dB, of which ~1.4 dB (~2.1 dB) arises from the clipping (cascaded DFT-Spread IFFT-based multi-channel aggregation) operation. Compared to the previously reported P2MP transceivers in which only the clipping operation is adopted, the DFT-Spread P2MP transceiver can further reduce the PAPRs by >2.6 dB. As a direct result, the proposed DFT-Spread transceiver has improved robustness to digital-to-analogue/analogue-to-digital conversion (DAC/ADC) quantization noise and leads to >1.9 dB upstream receiver sensitivity improvements and >14.1% aggregated upstream transmission capacity enhancement. More importantly, the proposed transceiver-induced aggregated upstream transmission capacity enhancements become more significant when the fiber lengths are further increased.

2. Operating Principle of Cascaded DFT-Spread IFFT/FFT-Based P2MP Flexible Transceivers

The diagram of the proposed cascaded DFT-Spread IFFT/FFT-based P2MP flexible transceivers is illustrated in Figure 1, and their corresponding transceiver DSP designs are presented in Figure 2 in detail. Upstream transmissions are considered only.

2.1. Cascaded DFT-Spread IFFT for Multi-Channel Aggregation in Transmitter DSPs

As illustrated in Figure 2a, in each transmitter DSP, a complex baseband signal is produced to convey multiple independent channels. Each channel can change its bitrates adaptively according to traffic requirements.

As seen in the figure, for the i-th ONU (ONU-i), the channel count is R. For aggregating these R channels, (R − 1) IFFT operations and (2R − 2) DFT-Spread operations are implemented. For aggregating the r-th channel, the (r − 1)-th IFFT operation [14,15] is depicted in Figure 2b. Where two input signals, A = [a₀, a₁, …, a_W−1] and B = [b₀, b₁, …, b_W−1], respectively, represent the (r − 2)-th IFFT operation output and the r-th channel signal. Two DFT operations are implemented, respectively, for the two signals A and B before the IFFT operations. As seen in Figure 2b, assuming the W-point DFT-Spread output signals are E = [e₀, …, e_W−1] and F = [f₀, …, f_W−1], the signal of the (r − 1)-th IFFT operation can thus be given as:

$$S_{\mathrm{IFFT}\_r-1}(n)=\left\{\begin{array}{c}{e}_n+f_n n=0,1,\dots W-1\\ e_{2W-1-n}^{*}-f_{2W-1-n}^{*} n=W,W+1,\dots 2W-1\end{array}\right.$$

(1)

where * represents the conjugate operation. The IFFT size required by the (r − 1)-th IFFT operation is expressed as:

L_{IFFT_(r−1)} = 2W = 2^r−1N

(2)

Here, 2N represents the first IFFT size. After the channel aggregation, following the parallel-to-serial conversion (P/S), a cyclic prefix (CP) insertion is executed. As seen in Figure 1, for each ONU, to accurately position the produced signal at a pre-allocated radio frequency spectral region (namely sub-wavelength), the produced complex-valued baseband signal passes a digital up-sampling (M↑) operation first and then undergoes an orthogonal digital filtering operation. Finally, an electrical-to-optical (E-O) conversion via intensity modulation can be executed to produce an optical signal. Due to the implementation of orthogonal digital filtering, no guard bands are necessary between adjacent radio frequency bands. The intensity modulation produces a double sideband optical signal with two conjugate optical sidebands.

2.2. Cascaded DFT-Spread FFT for Multi-Channel De-Aggregation in Receiver DSPs

As depicted in Figure 1, in the receiver, the signal detection and demodulation procedures include direct detection (DD), ADC, serial-to-parallel conversion (S/P), CP removal, FFT operations and subcarrier identification, as well as ONU channel de-aggregation and signal demodulation.

Following serial-to-parallel (S/P) conversions and CP removals, the FFT operations separate different ONUs’ signals [6]. Subsequently, the targeted ONUs’ subcarriers are identified and transmitted to the corresponding ONU signal demodulation modules, where the subcarrier equalizations are implemented first. Then, the multi-channel de-aggregation operations, as shown in Figure 2c,d, are performed to separate channels for a specific ONU.

It can be found in Figure 2c that for the de-aggregation of R independent channels, the conducting of (R − 2) FFT operations, (2R − 2) DFT-Spread operations and (R − 1) two-signal separation operations is required. As illustrated in Figure 2d, for ONU-i, the separation of the r-th channel is obtained by:

$$\left\{\begin{array}{c}{e}_n^{’}={\displaystyle \frac{1}{2}}\left[d_n+d_{2N-1-n}^{*}\right]\\ f_n^{’}={\displaystyle \frac{1}{2}}\left[d_n-d_{2N-1-n}^{*}\right]\end{array} n=0,1,...W-1\right.$$

(3)

where D = [d₀, …, d_W−1, d_W, …, d_2W−1] is the input signal. At the output, the obtained signal F′ = [fʹ₀, …, fʹ_W−1] and Eʹ = [e′₀, …, e′_W−1] are inputs to the IDFT module for the W-point IDFT operations.

At the output of the IDFT operations, the obtained signal B′ = [b′₀, …, b′_W−1], is the r-th channel signal, while the signal A′ = [a′₀, …, a′_W−1], is delivered to the next two signal separation modules. Detailed operating principles of the cascaded IFFT/FFT-based P2MP flexible transceivers are given in [5].

It is worth pointing out that the receiver does not require an orthogonal digital filter. Comparing the techniques using orthogonal digital filters to separate signals from different ONUs, such digital filter-free receivers, can reduce the signal separation DSP complexity by a factor of >10 [16].

3. PAPR of Cascaded DFT-Spread IFFT-Based Multi-Channel Aggregation

In this section, the PAPRs of the proposed transceiver-adopted cascaded DFT-Spread IFFT-based multi-channel aggregation are investigated numerically. To identify the key elements that determine the PAPRs of the proposed transceivers, the impacts of the last IFFT operation sizes, modulation formats and channel counts on PAPRs are investigated, respectively. In addition, PAPR comparisons between the cascaded IFFT-based multi-channel aggregations with and without DFT-Spread are also made to explore the key factors responsible for PAPR reductions in the proposed transceivers.

It is worth pointing out that, in this section, the PAPRs of the multi-channel aggregation are calculated without involving the digital up-sampling operations, digital filtering operations and clipping operations. The PAPRs of the signals after these operations are presented in the next section.

3.1. The Impact of the Last IFFT Size

The PAPR characteristics of the cascaded IFFT-based multi-channel aggregations with and without DFT-Spread are explored and presented in Figure 3. The aggregated channel count is fixed at four and the modulation format is 16 QAM. Three different sizes of the last IFFT operations, of 32, 256 and 4096, are considered.

It can be found in Figure 3a that a large size for the last IFFT operation leads to a high PAPR, indicating that the PAPR of the proposed transceivers is determined by the last IFFT sizes. The cascaded DFT-Spread IFFT-based multi-channel aggregation-induced PAPR reductions are illustrated in Figure 3b as a function of the size of the last IFFT operation at a probability of 1 × 10⁻³. It can be seen that the cascaded DFT-Spread IFFT-based multi-channel aggregation can reduce PAPRs by >1.4 dB when the last IFFT operation sizes are ≥32. When the last IFFT operation size is further increased to 4096, a more significant PAPR reduction of >2.2 dB can be achievable.

3.2. The Impact of Modulation Formats

For different modulation formats, the PAPRs of the cascaded IFFT-based multi-channel aggregations with and without DFT-Spread are calculated and presented in Figure 4. Four independent channels are aggregated and three different sizes of the last IFFT operations, being 128, 4096 and 16,384, are considered.

As seen in Figure 4, the resulting PAPRs are similar for different modulation formats, implying that for the proposed DFT-Spread P2MP transceivers, the signal PAPRs at the output of the multi-channel aggregation are independent of modulation format, and more importantly, the changes in modulation format have negligible impacts on the PAPR-developing trend of the proposed transceiver-adopted cascaded DFT-Spread IFFT-based multi-channel aggregation.

3.3. The Impact of the Channel Count

For different channel counts, the signal PAPRs at the output of the cascaded IFFT-based multi-channel aggregations with and without DFT-Spread are calculated and presented in Figure 5. The last IFFT operation sizes of 4096, 16,384, and 131,072 and the modulation formats of 16 QAM are considered. The results indicate that for a last IFFT operation size of >4096, the cascaded DFT-Spread IFFT-based multi-channel aggregation can lead to a PAPR reduction of >2.2 dB at a probability of 1 × 10⁻³, which is almost independent of the aggregated channel count.

Comparing the results presented in Figure 3, Figure 4 and Figure 5, it can be found that the last IFFT operation size plays a vital role in determining the PAPRs of the P2MP-flexible transceivers, rather than modulation format. This is because the last IFFT operation size directly determines the subcarrier count of the produced signals. When the last IFFT operation size is increased, the produced signal contains more subcarriers, and thus has higher PAPRs. This is similar to conventional OFDM signals, whose PAPRs are mainly determined by the subcarrier count rather than subcarrier modulation formats [17,18,19].

4. Performances of IMDD Upstream PONs Incorporating DFT-Spread P2MP Flexible Transceivers

In this section, the performances of the proposed DFT-Spread P2MP flexible transceivers are numerically explored in a 20 km upstream IMDD PON. Here, four ONUs are considered. Each ONU produces an optical signal containing four independent channels. The key parameters used in calculating the PON upstream transmission performances are listed in

Table 1. Transceiver and transmission system parameters.

Parameter	Value
ONU count	4
Channel count for each ONU	4
IFFT/FFT sizes for multi-channel aggregation	256/512/1024
Data-bearing subcarriers in each channel	128/128/256/512
Modulation formats	16 QAM
Cyclic prefix	12.5%
DAC and ADC sampling rate	12.5 GS/s
ONU optical launch power	8 dBm
PIN detector sensitivity	−19 dBm
PIN responsivity	0.8 A/W
ONU wavelength frequency	193.1 THz
Fiber attenuation	0.2 dB/km
SSMF dispersion parameter at 1550 nm	16 ps/nm/km
Orthogonal digital filter length	32
DAC/ADC resolution	8 bits
Digital up-sampling factor (M)	M = 8
Clipping ratio (With DFT-S/Without DFT-S)	10 dB/11 dB
Kerr coefficient	2.35 × 10−20 m2/W

. For the considered simulation setups, the transceiver DSPs are programmed using Matlab, while for all other components, including fibers, VPItransmissionMaker-embedded models are employed.

4.1. Optimal Clipping Ratio and Quantization Bit

To improve the transmission performances of the proposed DFT-Spread P2MP transceivers, in the considered back-to-back (B2B) upstream IMDD PONs, the optimal clipping ratios and DAC/ADC quantization bits are identified and presented in Figure 6a,b, respectively. Only the ONU-1 bit error rate (BER) performances are plotted in the figure for simplicity. In addition, for the conventional cascaded IFFT/FFT-based P2MP transceivers without DFT-Spread, the optimal clipping ratio and DAC/ADC quantization bits are also explored and presented in the same figures for performance comparisons.

In the calculation of Figure 6a, an 8-bit DAC/ADC quantization resolution and a received optical power of −3 dBm are taken into account. It can be found that for the proposed DFT-Spread P2MP transceivers (the conventional transceivers without DFT-Spread), the optimum clipping ratios are 10 dB (11 dB). This implies that the proposed DFT-Spread P2MP transceivers can utilize smaller clipping ratios to further reduce their PAPRs.

For the proposed DFT-Spread P2MP transceivers, when the clipping ratios are <10 dB, decreasing the clipping ratio results in significant signal distortions, thus leading to non-negligible transmission performance degradations. On the other hand, when the clipping ratios exceed 10 dB, insufficient signal clipping results in high PAPRs, which can reduce the effective signal-to-noise ratio (SNR) and degrade the BER performance. In Figure 6a, for the proposed transceivers, ONU 1’s four channels have slightly different BER performances due to the channel crosstalk effects [5].

Figure 6b illustrates the BER performances of ONU-1’s four channels for different DAC/ADC quantization bits. In obtaining the figure, for the proposed DFT-Spread P2MP transceivers (the conventional transceivers without DFT-Spread), the identified optimal clipping ratios of 10 dB (11 dB) are employed. The received optical power is set at −3 dBm. As shown in Figure 6b, comparing the two involved transceivers with and without DFT-Spread, the proposed DFT-Spread P2MP transceivers can significantly improve the transmission system tolerance to ADC/ADC quantization noise. In the following sections, a quantization bit of eight is used for comparing the transmission performances between the proposed DFT-Spread transceivers and the conventional transceivers without DFT-Spread.

In Figure 6c, to highlight the advantages of the proposed transceivers in terms of PAPR reductions, the PAPRs of the proposed transceivers and the conventional cascaded IFFT/FFT-based P2MP transceivers (without DFT-Spread) with and without a clipping operation are presented. It can be found that, for the considered last IFFT operation size of 1024, in comparison with the conventional P2MP transceivers without clipping operations, the proposed transceivers can reduce the PAPRs by ~3.5 dB, of which ~2.1 dB (~1.4 dB) arises from the cascaded DFT-Spread IFFT-based multi-channel aggregation (the clipping operation) operation. Compared to the conventional P2MP transceivers adopting the clipping operation only, the proposed transceivers can further reduce the PAPRs by >2.6 dB.

4.2. 20 km Upstream Transmission Performances

The 20 km upstream transmission performances of all four of the involved ONUs adopting the proposed DFT-Spread P2MP transceivers are explored numerically. The key parameters are listed in Table 1. Because all these ONUs present similar transmission performances, only ONU-1 BER performances are plotted in Figure 7 for simplicity. In addition, the B2B transmission performances of these ONUs are also given in the figure. For performance comparisons, the B2B and 20 km upstream transmission performances of the previously reported cascaded IFFT/FFT-based P2MP transceivers without DFT-Spread are also calculated and plotted in the same figure. The channel bitrates of each ONU are listed in

Table 2. Each ONU’s bit rate (Gbit/s).

ONU1	ONU2	ONU3	ONU4	Total
5.55	5.55	5.55	5.55	22.2

.

As shown in Figure 7, for the proposed DFT-Spread P2MP transceivers, all channels exhibit similar transmission performances, and 20 km upstream transmissions can only lead to <1 dB performance degradations at a BER of 1 × 10⁻³ for all the considered channels. In comparison with the previously reported P2MP transceivers without DFT-Spread, the proposed DFT-Spread P2MP transceivers can improve receiver sensitivity by >1.9 dB, because of the low PAPR-induced effective SNR improvements. Here, the receiver sensitivity is defined as the minimum receiver optical power for achieving BERs at the forward error correction (FEC) limits of 1 × 10⁻³.

4.3. Maximum Achievable Aggregated Upstream Signal Transmission Capacity

For the proposed cascaded DFT-Spread IFFT/FFT-based P2MP transceivers and the previously reported cascaded IFFT/FFT-based P2MP transceivers without DFT-Spread, the maximum achievable aggregated signal transmission capacities for different fiber lengths are explored and illustrated in Figure 8a, utilizing the parameters outlined in Table 1 and adaptive bit-loading techniques [20]. The received optical powers of −3 dBm are set.

As shown in Figure 8a, for the considered two transceivers, an increase in upstream transmission distance can lead to relatively low upstream signal transmission capacity because long fiber transmissions can result in relatively large signal distortions and channel interferences [21]. However, for all the considered fiber transmission distances, the proposed DFT-Spread P2MP transceivers can lead to larger aggregated upstream signal transmission capacities compared to the previously reported P2MP transceivers. More importantly, the proposed DFT-Spread P2MP transceiver-induced upstream signal transmission capacity improvements are more pronounced for longer transmission distances, as verified by Figure 8b. For a 20 km (50 km) upstream transmission distance, the proposed transceivers can enhance the maximum aggregated upstream signal transmission capacities by >4.7 Gbit/s (6.6 Gbit/s), which corresponds to >14.1% (26.3%) aggregated upstream signal transmission capacity enhancements with respect to previously reported transceivers.

To vigorously verify the proposed DFT-Spread P2MP transceivers’ superiority in improving the signal transmission capacity, the maximum achievable aggregated upstream signal transmission capacities of the proposed transceivers and the previously reported transceivers for different channel counts are illustrated in Figure 9. The last IFFT operation size is fixed at 1024. The upstream transmission distance is fixed at 20 km. The received optical powers are fixed at −3 dBm.

The results show that for a given clipping ratio, the proposed transceiver-induced improvements in the maximum aggregated upstream signal transmission capacities are independent of aggregated channel count. It also implies that the proposed transceivers support flexible channel count variations for each ONU without considerably compromising its transmission performance.

5. Conclusions

A novel P2MP flexible optical transceiver has been proposed, which uses a cascaded DFT-Spread IFFT/FFT-based multi-channel aggregation/de-aggregation and standard clipping operations to jointly reduce the PAPRs. The upstream transmission performances of the proposed transceiver are numerically explored in a 20 km IMDD PON. The results show that the proposed transceiver’s PAPRs are mainly related to the last IFFT operation size of the multi-channel aggregation and are almost independent of modulation format and channel count. In comparison with the previously reported cascaded IFFT/FFT-based P2MP flexible optical transceivers, the proposed DFT-Spread P2MP transceivers not only have lower PAPRs but also possess better upstream transmission performances and higher aggregated upstream signal transmission capacities. More importantly, such superiority of the proposed transceivers becomes more pronounced as the last IFFT operation size is increased.