METHOD AND SYSTEM OF A TIME-FREQUENCY DIVISION MULTIPLEXED RECEIVER

Abstract

A method for extracting a plurality of data streams from a time-frequency division multiplexed (TFDM) signal includes determining a plurality of sub-channels of the TFDM signal, where each of the plurality of sub-channels has a respective one of a plurality of frequency ranges. The method also includes down-converting, based on the plurality of frequency ranges, the TFDM signal into a plurality of down-converted signals, where each down-converted signal corresponds to a respective one of the plurality of sub-channels. The method also includes extracting the plurality of data streams from a respective one of the plurality of down-converted signals.

Claims

1.-19. (canceled)

20. A method, comprising: determining, at an optical line terminal (OLT), a plurality of sub-channels of a TFDM signal received from one or more optical network units (ONU's), each of the plurality of sub-channels having a respective one of a plurality of frequency ranges; calculating, for each of the plurality of sub-channels, a respective one of a plurality of sub-channel powers; determining, within each of the plurality of frequency ranges, an equalized power level of each respective sub-channel of the plurality of sub-channels; and transmitting, from the OLT to the one or more ONU's, a downlink signal including the equalized power level of each respective sub-channel of the plurality of sub-channels.

21. The method of claim 20, said calculating, for each of the plurality of sub-channels, the respective one of the plurality of sub-channel powers comprising estimating the respective sub-channel powers by integrating square of absolute power level values within the respective sub-channel.

22. The method of claim 20, the TFDM signal having a 100 gigabits/second data rate.

23. The method of claim 20, the TFDM signal comprising four sub-channels, each sub-channel having a 25 gigabits/second date rate.

24. The method of claim 20, further comprising determining channel response information and transmitting the channel response information with the downlink signal.

25. The method of claim 24, the channel response information being determined using a constant modulus algorithm (CMA).

26. A method, comprising: transmitting, from an optical network unit (ONU) to an optical line terminal (OLT), a TFDM signal having a plurality of sub-channels of, each of the plurality of sub-channels having a respective one of a plurality of frequency ranges; after the transmitting, receiving, at the ONU from the OLT, a downlink signal defining an equalized power level of each respective sub-channel of the plurality of sub-channels; and transmitting a further TFDM signal according to the equalized power level.

27. The method of claim 26, the equalized power level being determined by the OLT.

28. The method of claim 27, the OLT determining the equalized power level by estimating the respective one of the plurality of sub-channel powers by integrating square of absolute power level values within the respective sub-channel.

29. The method of claim 26, the TFDM signal having a 100 gigabits/second data rate.

30. The method of claim 26, the TFDM signal comprising four sub-channels, each sub-channel having a 25 gigabits/second date rate.

31. The method of claim 26, the downlink signal further including channel response information.

32. The method of claim 24, the channel response information being determined using a constant modulus algorithm (CMA).

33. A system, comprising: an optical line terminal (OLT) communicatively coupled with one or more optical network units (ONU's), the OLT including a processor and memory storing computer readable instructions that, when executed by the processor, control the OLT to: determine a plurality of sub-channels of a TFDM signal received from the one or more ONU's, each of the plurality of sub-channels having a respective one of a plurality of frequency ranges; calculate, for each of the plurality of sub-channels, a respective one of a plurality of sub-channel powers; determine, within each of the plurality of frequency ranges, an equalized power level of each respective sub-channel of the plurality of sub-channels; and transmit, from the OLT to the one or more ONU's, a downlink signal including the equalized power level of each respective sub-channel of the plurality of sub-channels.

34. The system of claim 33, said calculating, for each of the plurality of sub-channels, the respective one of the plurality of sub-channel powers comprising estimating the respective sub-channel powers by integrating square of absolute power level values within the respective sub-channel.

35. The system of claim 33, the TFDM signal having a 100 gigabits/second data rate.

36. The system of claim 33, the TFDM signal comprising four sub-channels, each sub-channel having a 25 gigabits/second date rate.

37. The system of claim 33, comprising further instructions stored in the memory that, when executed by the processor, further cause the OLT to determine channel response information and transmit the channel response information with the downlink signal.

38. The system of claim 37, the channel response information being determined using a constant modulus algorithm (CMA).

Description

BRIEF DESCRIPTION OF THE FIGURES

[0006] FIG. 1A is a schematic of a coherent TFDM passive optical network (PON) system, in an embodiment.

[0007] FIG. 1B is an example signal spectrum showing the four sub-channels in the TFDM PON system of FIG. 1A.

[0008] FIG. 1C shows an example burst distribution under operation of TFDM PON system of FIG. 1A.

[0009] FIG. 2 is a block diagram of a TFDM PON that illustrate key hardware of ONU and OLT of FIG. 1A.

[0010] FIG. 3A is a schematic of a signal recovery procedure of a coherent PON TFDM burst receiver that implements channel recognition and frequency window calibration (CR-FC), in an embodiment.

[0011] FIG. 3B is a detailed schematic of the CR-FC step of FIG. 3A.

[0012] FIG. 4 shows example plots that accompany the description of the steps of FIG. 3B.

[0013] FIG. 5 illustrates a CPON TFDM burst receiver system that implements the signal recovery procedure of FIG. 3A.

[0014] FIG. 6 shows plots that illustrate power spectra of the signals expected from TFDM burst receiver system of FIG. 5.

[0015] FIG. 7 shows plots illustrating the process of rebalancing and sub-channel power optimization, which may be performed as part of the system of FIG. 5.

[0016] FIG. 8 is a plot showing boundaries of sub-channels recognized by CR-FC of the TFDM burst receiver system of FIG. 5 in an example TFDM spectrum.

[0017] FIG. 9 is a plot comparing bit-error rate (BER) performance of the method used in the TFDM burst receiver system of FIG. 5 to other conventional methods.

[0018] FIG. 10 shows received signal in the example of FIG. 6 in time domain.

[0019] FIG. 11 shows the BER performance plots for with and without power rebalancing step in the measurement shown in FIG. 10

DETAILED DESCRIPTION OF THE EMBODIMENTS

[0020] Coherent passive optical network (PON) using 100-Gbit/s (100G hereinafter) single-wavelength carrier is an attractive solution for next-generation optical access networks to boost the capacity and coverage. For data multiplexing 100G single carrier, conventional time-division multiplexing (TDM) is a mature solution. However, TDM faces a challenge of high scheduling latency when connecting an ultra-large number of optical network units (ONUs) and the insufficient flexibility to simultaneously support high-bandwidth and low-latency services. To overcome the challenges that TDM faces, time and frequency division multiplexing (TFDM), described in Patent U.S. Pat. No. 11,265,099B1 filed on May 18, 2020 and entitled System and methods for time and frequency division passive optical network, which is incorporated by reference, was introduced to provide multiple independent sub-channels, where different network services and ONU groups may be allocated with different bands without the need of contention resolutions. TFDM greatly enhances the flexibility for the coexistence of low-latency mobile and high-speed video streaming services over a converged CPON platform. The polling latency and traffic blocking rate are also greatly improved since multiple parallel scheduling pipelines are provided.

[0021] While a conventional 10-gigabit symmetrical PON (XGS-PON) connects up to 32 optical network units (ONUs), 100G single carrier increases the number of connected users to 256 512 in CPON. However, the increased number of connected users results in significantly higher latency in polling and scheduling in the media access control (MAC) layer. The polling and scheduling delay of single-carrier CPON with 512 ONUs may increase by 16 times: 16k250 s, where k is the number of round trips in one polling cycle. The resulting delay is estimated to be 832 millisecond, which is intolerable for supporting low-latency services. With TFDM, such a latency can be greatly mitigated by scheduling the users independently in multiple sub-channel pipelines (or ONU grouping). Moreover, the TFDM also brings extra flexibility to support network slicing by allocating different bands for different services. For example, one of its sub bands may be used dedicatedly as low-latency transmission without disturbing other regular services.

[0022] However, TFDM also brings some technical challenges. One challenge comes from the burst detection in TFDM upstream. Unlike conventional TDM PON, where the burst can be detected by locating the power envelope in time domain, the burst of TFDM may contain multiple signals from different sub-channels with different power levels. The multiple signals are mixed and require an extra step to recognize how many sub-channels are included and from which ONU each sub-channel originates. Additionally, in conventional TDM burst receivers, to optimize the performance and fully utilize the dynamic range of analog-to-digital converter (ADC), automatic-gain control (AGC) is implemented in trans-impedance amplifier (TIA) to monitor and adjust the burst power. However, adjusting the burst power is difficult to implement in TFDM because the signal components distributed in multiple sub-channels are mixed in time domain. Consequently, instead of passively adjusting the received power, some proactive mechanism may be designed to balance the burst power at the transmitter site.

[0023] Additionally, at each ONU, an independently operated laser diode introduces random frequency drifts up to a few GHz, detuning the signal from the original sub-channel window and causing failure of the conventional carrier frequency offset (CFO)-estimation algorithms. Thus, the carrier frequency of the sub-channels must be precisely estimated before digital down conversion. To overcome these challenges, methods for channel detection and frequency-window rectification are described herein for TFDM based CPON burst receiver. With these methods, the burst receiver may efficiently differentiate the sub-channels from mixed signal, and the system tolerance for CFO is significantly increased for each sub-channel in the coherent detection process.

[0024] FIG. 1A is a schematic of a coherent TFDM PON system 100. The PON system 100 includes an optical line terminal (OLT) 120, a power splitter 140, and ONUs 122. ONUs 122 includes up to N ONUs: ONU 1 122(1) to ONU N 122(N). In the example implementation of TFDM PON system shown in FIG. 1A, the uplink transmission includes four sub-channels: CH 1 to CH 4. Each ONU 122(i), where i is a positive integer up to N, may use any number of available sub-channels. For example, ONU 1 112(1) uses all four sub-channels while ONU 2 122(2) uses sub-channels CH 1 and CH 4 for uplink transmission.

[0025] FIG. 1B is an example signal spectrum 102 showing four sub-channels in TFDM PON system 100. Each sub-channel occupies a 10-GHz window 112 with 6.25-GHz bandwidth 114 centered at 15, 5, +5, and +15 GHz. Signal spectrum 102 also shows a guard band 116 of 3.75 GHz reserved between adjacent sub-channels. Modulated by 25-Gbit/s dual-polarization (DP) quadrature phase shift keying (QPSK) signal in each channel, the total uplink data rate reaches 100 Gbit/s. Depending on the bandwidth demand, service requirements, or hardware options, each ONU 122(i), may use one to four sub-channels with the data rate varying from 25 to 100 Gbit/s with a balance in capacity and power budget. While the example shown in FIG. 1B has four sub-channels of 25 Gbits/s with a total data rate of 100 Gbits/s, the total data rate may be higher or lower than 100 Gbits/s, and/or the total number of sub-channels may be any positive integer without departing from the scope of this disclosure.

[0026] FIG. 1C shows an example burst distribution under operation of TFDM PON system 100. The burst distribution from different ONUs is dynamically allocated in different sub-channels, leading to a highly flexible architecture. For example, to accommodate the low-latency services with minimal interruptions to the normal service, one of the sub-channels can be configured (e.g., CH 1 in FIG. 1C) to allow consecutive short bursts with reduced average data queueing time.

[0027] FIG. 2 is a block diagram of a TFDM PON 200 that illustrate key hardware of ONU 122 and OLT 120 of FIG. 1A. Functions of the key hardware shown in FIG. 2 may also be performed by one or more processors. TFDM PON 200 is an example of TFDM PON system 100. ONU 122 includes a hardware 232 that performs channel pre-equalization and a hardware 234 re-balances transmitter power. OLT 120 includes a hardware 236 that performs sub-channel recognition and frequency window calibration (CR-FC) and a hardware 238 that performs burst detection and coherent digital signal processing (DSP). The ONU transmitter initialization, including channel pre-equalization and power rebalancing, may compensate the power difference of received bursts thus improving the fairness and better utilizing the quantization range of ADC. The channel recognition algorithm detects the active sub-channels on a large range, estimates and calibrates its channel frequency windows, and thus detects dynamic burst.

[0028] FIG. 3A is a schematic of a signal recovery procedure 300 of a CPON TFDM burst receiver that implements channel recognition and frequency window calibration (CR-FC). The signal recovery procedure 300 includes steps 303, 304, 310, 306, 308, and 309. An example of a corresponding hardware for the burst receiver is shown in FIG. 5. Step 303 receives an uplink signal using a coherent receiver. Any of the steps 304, 310, 306, 308 and 309 may be performed by a processor, The CR-FC step 310 is performed between ADC step 304 and sub-channel down conversion step 306. FIG. 3B is a detailed schematic of the CR-FC step 310 of FIG. 3A. The CR-FC step includes steps 312, 314, 316, 318, and 320. For clarity of description, FIG. 4 shows example plots 410, 420, and 430 that accompany the description of the steps of FIG. 3B. FIGS. 3A, 3B and 4 are best viewed together in the following description.

[0029] The first step 312 of CR-FC is to periodically extract data samples from a received stream with the data samples forming a monitoring data slice. In step 314, Fast Fourier transform (FFT) converts the data slice into the frequency domain. Plot 410 shows a frequency spectrum of an example data slice after the FFT step 314 is performed. In step 316, channel boundary search is performed. Step 316 includes smoothening the frequency spectrum of step 314. In an example of step 316, plot 420 shows the result of smoothening of plot 410. Step 316 also includes searching for channel boundaries. In an example of channel boundary search, plot 430 of a binary channel distribution map is generated with 0 and 1 indicating whether the signal components in plot 420 fall below or above, respectively, a threshold 428.

[0030] Additional rules may be applied to eliminate the outlier noisy peaks, optical carriers, and harmonics in this step. In step 318, frequency windows are calibrated, and sub-channel carrier frequencies are calculated. In step 320, the power of each channel is estimated by integrating the square of the signal absolute values within the sub-channel window. After step 306 of down conversion and baseband digital filtering, step 308 of burst detection is performed in each sub-channel to locate and synchronize the bursts before a step 309 of coherent DSP is applied to recover the payload signals.

[0031] Additionally, information including the estimated power from step 320 of CR-FC and channel response information obtained from constant modulus algorithm (CMA) may feed back to the ONU in a step 382. Step 382 transmits a downlink signal that includes the estimated power to the ONU to perform pre-equalization (Pre-EQ) and power rebalancing (PR).

[0032] FIG. 5 illustrates a CPON TFDM burst receiver system 500 that implements the signal recovery procedure of FIG. 3A. System 500 includes an OLT 520 and two independently operated ONUs 522(1) and 522(2) to emulate the bidirectional point-to-multi-point transmission. OLT 520 is an example of OLT 120, and ONUs 522(1) and 522(2) are examples of ONU 122. FIG. 6 shows plots 610, 620, and 630 that illustrate power spectra of the signals expected from TFDM burst receiver system 500. For clarity, FIGS. 5 and 6 are best viewed together in the following description. In this example of a TFDM burst receiver system with four channels, the downstream is achieved by broadcasting a single-carrier 100-Gbit/s DP-QPSK signal, as shown in plot 610, from OLT 520 to ONUs 522(1) and 522(2) with the wavelength at 1565.5 nm. For the four channels in the TFDM uplink, ONU 1 522(1) is allocated with CH 1 and CH 3 as shown in plot 620. ONU2 is allocated with CH 2 and CH 4 as shown in plot 630. The two TFDM signals at each ONU are generated by two 80-GSa/s arbitrary waveform generators (AWG) and modulated onto the light through dual-polarization IQ modulators (DP-IQM) with the wavelength at 1553.3 nm. The driving voltages of the modulators are tuned and stabilized to suppress the frequency image due to IQ imbalance. At the OLT site, a homodyne coherent receiver is used to receive the combined uplink optical signal and offline DSP is applied to demodulate the data.

[0033] ONUs 522(1) and 522(2) may also perform a sub-channel pre-equalization step to enable a reliable and accurate power control utilizing step 382, FIG. 3A, from OLT 520. The pre-equalization step is achieved by extracting from the downlink signal of step 382 the estimated power and the channel response information from CMA. FIG. 7 shows plots 710, 720, and 730 illustrating the process of rebalancing and sub-channel power optimization, which may be performed as part of the system 500, FIG. 5. Plots 710, 720, and 730 show the power spectra of the combined TFDM signals in the process of equalization. Inset drawings in the plots 710, 720, and 730 show exaggerated power spectra to highlight the steps. Plot 710 is a TFDM signal spectrum prior to any pre-processing. Plot 720 shows the TFDM signal spectrum after pre-equalization is applied to the spectrum of plot 710. After pre-equalization, each sub-channel signal is flattened in frequency domain. Then, based on the information of power levelling, the transmitter at each ONU adjusts its signal amplitude to achieve the fully rebalanced power distribution as shown in plot 730. Plot 730 shows the TFDM signal spectrum after power re-balancing is applied to the spectrum of plot 720.

[0034] For performance evaluation of CR-FC in the TFDM burst receiver system 500, FIG. 8 is a plot 800 showing boundaries of sub-channels recognized by CR-FC of the TFDM burst receiver system 500 in an example TFDM spectrum. In plot 800, sudden changes in power are recognized at points 812. The points 812 indicate the sub-channel boundaries 814. Even for difficult cases when CH 3 has drifted away from the original center frequency by 2.2 GHZ, the sub-channel boundaries 814 are successfully recognized. In general, by varying the frequency detuning of CH 3 from 2.4 to +2.4 GHz, bit-error rate (BER) performance may be measured. FIG. 9 is a plot 900 comparing BER performance of the method used in the TFDM burst receiver system 500 to other conventional methods. The three methods compared in plot 900 are: (i) the conventional fourth-power carrier estimation (FPCE) at baseband (BB) DSP (e.g., coherent BB DSP step 309, FIG. 3) after down conversion; (ii) frequency-pilot-tone-based coarse CFO estimation first with inserted pilot tones at 20, 10, 10, 20 GHZ, and applying FPCE in the BB DSP, and (iii) CR-FC of the TFDM burst receiver system 500 and FPCE in the BB DSP. Among the three methods, method (i), FPCE-only method, shows the worst detuning tolerance since the frequency detection range is reduced by four times due to fourth-power operation. Method (ii), pilot-tone based method, performs better than method (i), but the method fails under a larger frequency offset when the pilot tones enter the adjacent channels. Method (iii), CR-FC of TFDM burst receiver system 500, shows the best frequency-detuning tolerance because of the larger detection range covering over 75% of the 3.125-GHz frequency gap between adjacent sub-channels.

[0035] In another example of applying the signal recovery procedure 300, FIG. 3A, in the TFDM burst receiver system 500, FIG. 10 shows received signal in the example of FIG. 6 in time domain. FIG. 10 includes plots 1010, 1012, 1014, 1016, and 1018. Plot 1010 shows a combined signal, or as received from the two ONUs 552(1) and 552(2), from all sub-channels from CH1 to CH4. In plot 1010, the asynchronous bursts in four sub-channels and two ONUs are mixed together forming a quasi-continuous-wave distribution. After the frequency windows are calibrated by CR-FC step 310, FIG. 3A, the four sub-channels, CH1, CH2, CH3, and CH4 are down converted in down conversion step 306 in respective plots 1012, 1014, 1016, and 1018. Burst power detection step 308 for each sub-channel is applied to locate the start and end point of each burst 1052 before the payload signals are demodulated by coherent BB DSP step 309.

[0036] FIG. 11 shows the BER performance plots 1110, 1120, and 1130 for with and without power rebalancing step in the measurement shown in FIG. 10. In plot 1110, when power rebalancing is applied, the initial 4-dB power budget difference among the sub-channels is mitigated and the overall BER is improved. Plot 1120 shows the BER performance with and without 80-km fiber transmission, where no significant fiber-transmission penalty is observed for different ONU and sub-channels. Plot 1130 shows the BER performance with and without burst operation, where burst receiving results in no power penalty. Because of the guard time between bursts, the average received power is slightly reduced, leading to a small shift of the curves. The BER curve of the 100-Gbit/s downlink transmission is provided as a reference. When considering the case of four TFDM sub-channels from two ONUs with aggregated 100-Gbit/s data rate, the receiver sensitivity is around-39.2-dBm at 4.5E-3 BER threshold with no significant degradation compared with 100-Gbit/s single-carrier downlink.

[0037] In summary, coherent burst receiving technologies for TFDM uplink transmission in 100G CPON are described. Embodiments disclosed include a power control system that includes pre-equalization and power rebalancing and a frequency control system that includes sub-channel recognition and frequency-window calibration are detailed. Using these systems, a TFDM coherent burst receiver may achieve 2.4-GHz frequency detuning tolerance and 39.2-dBm sensitivity under 80-km uplink transmission with four sub-channels.

[0038] Changes may be made in the above methods and systems without departing from the scope hereof. It should thus be noted that the matter contained in the above description or shown in the accompanying drawings should be interpreted as illustrative and not in a limiting sense. The following claims are intended to cover all generic and specific features described herein, as well as all statements of the scope of the present method and system, which, as a matter of language, might be said to fall therebetween.