WO2014146708A1

WO2014146708A1 - Timing recovery apparatus

Info

Publication number: WO2014146708A1
Application number: PCT/EP2013/055816
Authority: WO
Inventors: Nebojsa Stojanovic
Original assignee: Huawei Technologies Co., Ltd.
Priority date: 2013-03-20
Filing date: 2013-03-20
Publication date: 2014-09-25
Also published as: CN105308895A; EP2976852A1; CN105308895B; EP2976852B1

Abstract

The present invention relates to a timing recovery apparatus, comprising: a grouper for grouping digital signal values of a digital signal to obtain groups of values, each group of values comprising a first digital value and a second digital value; a multiplier for multiplying a first digital value with a conjugated version of a second digital value of each group of values to obtain a first group value representing each group of values, and for multiplying a second digital value of each group of values with a conjugated version of a first digital value of a respectively adjacent group of values to respectively obtain a second group value representing adjacent group of values; a processor for respectively processing first group values respectively representing adjacent groups of values and a second group values representing the adjacent groups of values to obtain a plurality of processed values; and an averager for determining an average value from the plurality of processed values.

Description

DESCRIPTION

Timing recovery apparatus

BACKGROUND

An important goal of long-haul optical fiber systems is to transmit the highest data throughput over longest distances without signal regeneration in optical-electrical-optical regenerators. Given constraints on the bandwidth imposed by optical amplifiers and ultimately by the fiber itself, it is important to maximize spectral efficiency. Most current systems use binary modulation formats, such as on-off keying encoding one bit per symbol.

Advanced modulation formats in combination with coherent receivers enable high capacity and spectral efficiency. Polarization multiplexing, quadrature amplitude modulation (QAM) and coherent detection are seen as a winning combination for the next generation of high- capacity optical transmission systems since they allow information encoding in all the available degrees of freedom. Commercial devices using QAM constellation are already available in 40 and 100 Gb/s optical systems. SUMMARY

It is the object of the invention to enhance digital signal processing.

This object is achieved by the features of the independent claims. Further implementation forms are apparent from the dependent claims, the description and the figures.

In order to describe the invention in detail, the following terms, abbreviations and notations will be used:

CD: chromatic dispersion,

PMD: polarization mode dispersion,

OFE: optical front end,

TIA: transimpedance amplifier, DSP: digital signal processing,

TED: timing error detection,

TEDC: timing error detection characteristics,

AGC: automatic gain control,

ADC: analog/digital converter,

QAM: quadrature amplitude modulation,

VCO: voltage controlled oscillator,

PLL: phase locked loop,

LO: local oscillator,

PMD: polarization mode dispersion,

FFT: fast fourier transformation,

IFFT: inverse fast fourier transformation,

FIR: finite impulse response,

PD: phase detector,

ROF: roll-off factor,

QPSK: quaternary phase shift keying,

PSK: phase shift keying,

QPSK: quadrature phase shift keying, CDU clock distribution unit,

LPF: low pass filter,

Ul: unit interval,

I: in-phase,

Q: quadrature.

According to a first aspect, the invention relates to a timing recovery apparatus, comprising a grouper configured for grouping digital signal values of a digital signal to obtain groups of values, each group of values comprising a first digital value and a second digital value; a multiplier configured for multiplying a first digital value with a conjugated version of a second digital value of each group of values to obtain a first group value representing each group of values, and configured for multiplying a second digital value of each group of values with a conjugated version of a first digital value of a respectively adjacent group of values to respectively obtain a second group value representing adjacent group of values; a processor configured for respectively processing first group values respectively representing adjacent groups of values and a second group values representing the adjacent groups of values to obtain a plurality of processed values; and an averager configured for determining an average value from the plurality of processed values. In a first possible implementation form of the timing recovery apparatus according to the first aspect as such, the grouper is configured to group subsequent N values of the digital signal to obtain a respective group of values.

In a second possible implementation form of the timing recovery apparatus according to the first aspect as such or according to the first implementation form of the first aspect, the grouper comprises an oversampler configured for oversampling an input optical signal to obtain the digital signal.

In a third possible implementation form of the timing recovery apparatus according to the first aspect as such or according to any of the preceding implementation forms of the first aspect, the multiplier comprises a plurality of digital multipliers. In a fourth possible implementation form of the timing recovery apparatus according to the first aspect as such or according to any of the preceding implementation forms of the first aspect, the processor comprises configured for each group of values, a subtractor configured for subtracting first group values representing adjacent groups of values, and a multiplier configured for multiplying the subtraction result with a respective second group of values representing these adjacent group of values.

In a fifth possible implementation form of the timing recovery apparatus according to the first aspect as such or according to any of the preceding implementation forms of the first aspect, the averager is configured to average over real parts of the plurality of processed values to determine the average value.

In a sixth possible implementation form of the timing recovery apparatus according to the first aspect as such or according to any of the preceding implementation forms of thefirst aspect, the timing recovery apparatus comprises a loop filter configured for filtering the average value.

According to a second aspect, the invention relates to a communication receiver apparatus comprising a receiver configured for receiving a communication signal; the timing recovery apparatus of any of the preceding claims configured for timing recovery upon the basis of the communication signal; and a voltage controlled oscillator configured for providing a clock signal, the being voltage controlled oscillator controlled by an output of the recovery apparatus.

In a first possible implementation form of the communication receiver apparatus according to the second aspect as such, the clock signal is provided to the receiver.

In a second possible implementation form of the communication receiver apparatus according to the second aspect as such or according to the first implementation form of the second aspect, the communication receiver apparatus comprises a plurality of receivers, wherein the clock signal is distributed to the plurality of receivers.

In a third possible implementation form of the communication receiver apparatus according to the second aspect as such or according to any of the preceding implementation forms of the second aspect, the communication receiver apparatus further comprises a chromatic dispersion compensation unit arranged downstream the voltage controlled oscillator.

In an possible implementation form of the communication receiver apparatus according to the second aspect as such or according to any of the preceding implementation forms of the second aspect, the communication receiver apparatus is an optical communication receiver.

According to a third aspect, the invention relates to a timing recovery method comprising grouping digital signal values of a digital signal to obtain groups of values, each group of values comprising a first digital value and a second digital value; multiplying a first digital value with a conjugated version of a second digital value of each group of values to obtain a first group value representing each group of values, and multiplying a second digital value of each group of values with a conjugated version of a first digital value of a respectively adjacent group of values to respectively obtain a second group value representing adjacent group of values; respectively processing first group values respectively representing adjacent groups of values and a second group values representing the adjacent groups of values to obtain a plurality of processed values; and determining an average value from the plurality of processed values. In a first possible implementation form of the timing recovery method according to the third aspect as such, the method comprises performing the timing recovery method of claim 13 to provide average values; and controlling the average values to obtain the clock signal.

According to a fourth aspect, the invention relates to a computer program with a program code configured for performing the method according to according to the third aspect as such or according to first implementation forms of the third aspect, when run on a computer.

According to a fifth aspect, a timing recovery apparatus is provided. The timing recovery apparatus comprises a processor configured to perform the method according to the third aspect as such or according to any of the preceding implement forms of the third aspect.

These and other aspects of the invention will be apparent from the embodiment(s) described below. Fig. 1 shows a block diagram of a coherent optical receiver; Fig. 2 shows basic DSP blocks;

Fig. 3 shows a CD compensation block; Fig. 4 shows a frequency and impulse response of raised-cosine filter with various roll-off factors;

Fig. 5 shows quadrature phase-shift keying (QPSK) timing error detector characteristic (TEDC);

Fig. 6 shows further quadrature phase-shift keying (QPSK) timing error detector

characteristics (TEDC);

Fig. 7 shows signal pre-processing;

Fig. 8 shows TED characteristics of a phase detector PD; Fig. 9 shows further TED characteristics of a phase detector PD; Fig. 10 shows further TED characteristics of a phase detector PD; Fig. 1 1 shows further TED characteristics of a phase detector PD; Fig. 12 shows further TED characteristics of a phase detector PD;

Fig. 13 shows Nyquist super-channel clocking; Fig. 14 shows a parallel implementation; and Fig. 15 shows a flow chart of a timing recovery method.

DETAILED DESCRIPTION OF EMBODIMENTS OF THE INVENTION

Fig. 1 shows a block diagram of a coherent optical receiver 100. Since digital signals are mapped into both polarization, a 90°hybrid 101 is used to mix input optical signal with a local oscillator (LO) signal resulting in four output signals (two signals per polarization). The output optical signals of the 2Pol hybrid 101 are converted into electrical signals via a plurality of optical front ends (OFEs) 103 consisting of photo diodes, e.g. single PIN or balanced, and a transimpedance amplifier (TIA). As the power of the converted electrical signals may vary over time, a plurality of fast automatic gain control (AGC) blocks 105 may be used to compensate for signal power variations. There are exampled four AGC blocks 105 which can also be an integral part of the OFE blocks 103. Due to realization complexity, a pair of AGC blocks 105 is controlled by one voltage control signal, such as VX_AGC for X polarization and VY_AGC for Y polarization in Fig. 1. However, each of the four AGC blocks 103 can also be controlled by one independent voltage control signal. Output signals of the AGC blocks 105are quantized by a plurality of analog-to-digital converters (ADCs) 107 to produce four quantized digital signals.

Four quantized digital signals are further processed in a digital signal processing (DSP) block 109 that is divided into two parts, a fast DSP hardware part 1 1 1 and a slow DSP software part 1 13. In the DSP block 109, the quantized digital signals are compensated for chromatic dispersion (CD), polarization mode dispersion (PMD), polarization rotation, nonlinear effects, LO noise, LO frequency offset, etc. Estimation of slow processes, such as LO frequency offset and CD, can be done in the software part 1 13 of the DSP block.

The DSP block may be in the form of a DSP circuit.

The a hybrid 101 , optical front ends (OFEs) 103, automatic gain control (AGC) blocks 105 and analog-to-digital converters (ADCs) 107 are within the x analogue part of the receiver 100, whereas the digital signal processing (DSP) block 109, DSP hardware part 1 1 1 and DSP software part 1 13 are within the Rx digital part.

Fig. 2 shows basic DSP blocks 101 of Fig 1 , shown as 200 in this figure. After offset and gain correction 201 the four signals are equalized for chromatic dispersion in frequency domain using two fast Fourier transformation (FFT) blocks 203. Frequency offset is removed in a frequency recovery block 205.

Polarization tracking, PMD compensation and residual CD compensation are done in time domain using finite impulse response (FIR) filters 207 arranged in butterfly structure. Both residual frequency offset and carrier phase recovery are done in a carrier recovery block 209. When differential decoding is applied at the transmitter side, a differential decoder is used in a decoding and frame detection block 21 1.

CD is efficiently compensated in FFT blocks 203. The compensation CD function is

where λ₀ is the signal wavelength, f_s is the sampling frequency, N is the FFT size, c is the speed of light, n is the tap number, L is the fiber length, and D is the dispersion coefficient.

Fig. 3 shows a CD compensation block 203 of Fig. 2. Due to complexity reasons, only one FFT block 301 with a complex input is applied to each polarization. The inverse FFT (IFFT) identical to the FFT although real and imaginary parts are swapped at input and output.

In digital communication systems the heart of each receiver is a clock recovery circuit that extracts frequency and phase from an incoming optical signal and forces a local clock source to control the sampling rate and the sampling phase of the ADC. In over-sampled systems data processing blocks are less sensitive to the sampling phase.

Phase detectors (PD) are used in practical systems. In phase detectors a timing error detector characteristic, ie PD output over symbol interval, is very similar to a sinusoidal function. One exception is a„bang-bang" phase detector in which TEDC also has sinusoidal shape in the presence of noise. Some PDs work with one sample per symbol, while others are used with two-fold oversampling.

A real part of a PD timing error detector characteristic (TEDC) can be described for complex signals for example as:

TEDC(x ) = E[real(x(kT - T I 2 +x

* (kT +x )- x * (kT - T +x )))] (2) where T is a symbol interval, x is an input optical signal , τ is a sampling instant (between 0 and 7), £ is an expectation, and * denotes a complex conjugate operation. The imaginary part of a PD can be easily translated in FFT domain as

where A/ is a FFT-size, and is a FFT of X(/ 7/2+T)), / =0,1 ,..Λ/-1 . The received signal oversampled with two samples per symbol.

Nyquist transmission based on Nyquist pulses is used to frequency limit channel bandwidth.

This enables better channel packaging and automatically higher spectral efficiency. The raised-cosine filter is an implementation of a low-pass Nyquist filter, i.e. a filter having the property of vestigial symmetry. This means that its spectrum exhibits odd symmetry about 1/2 T, where T is the symbol-period of the communications system. Its frequency-domain description is given by:

0≤β < 1 (4) and characterized by two values: β - the roll-off factor and T - the sampling period. The impulse response of such a filter is given by:

in terms of the normalized s/nc-function.

The roll-off factor β is a measure of the excess bandwidth of the filter, ie the bandwidth occupied beyond the Nyquist bandwidth of 1/27. Fig. 4 shows frequency and impulse response of raised-cosine filter with various roll-off factors. Minimum signal bandwidth is achieved for roll-off factor equal to 0.

Decreasing roll-off factor (ROF) destroys clock tone quality. The TEDC becomes small and generates large and uncontrolled jitter. TEDC simulation results of PD for QPSK modulation format at Eb/N0=3dB show clock tone degradation for small ROF values. Sinusoidal TEDC is acceptable for ROF values greater than 0.3.

Fig. 5 shows a quadrature phase-shift keying (QPSK) timing error detector characteristic (TEDC) for ROF from 0 to 1 with a step width of 0.1 .

Fig. 6 shows QPSK TEDCs for ROF=0 and the clock recovery problem. In this case simulated TEDCs (one per 512 symbols) for ROF=0 are very small and desynchronized. Fig. 7 shows signal pre-processing. In Fig. 7 power operation in PAM systems is used to generate the clock tone at Baud rate. This approach uses specific pre-filter and narrowband filter to filter out the clock tone. It enables the clock extraction for small ROF values but fails for higher ROF values. The complete system is realized in the analog domain where is no limitation in terms of sampling frequency and signal digitalizing before the clock extraction.

The invention describes a method for the clock extraction in Nyquist systems with the benefits that the method enables clock extraction independently on ROF value unlike other known techniques. While present PD detectors often use two complex multiplications per symbol, the proposed PD uses additionally 1 .5 complex multiplications per symbol, i.e. totally 3.5 multiplications. The method is independent from the order of QAM and PSK modulation formats and can tolerate a large frequency offset.

The enhanced PD works with complex modulation formats. The received signal is twice over- sampled, ie two samples per symbol. Let us denote samples within one symbol interval n by A(n) and B(n). Then, the TEDC is calculated using equation

where C values are derived as C(n - \) = A(n - \)conj[B(n - \)]

C{n) -- = B(n - \)conj[A(n)] (7)

C(n + \) = A(n)conj[B(n)]

The TEDC has a sinusoidal shape with positive zero crossing indicating steady-state

(sampling instant; see Fig. 8). Using imaginary part in equation (5) results in TEDC having the constant value over one unit interval (Ul) symbol interval. Such a TEDC cannot be used for clock extraction.

Fig. 8 shows TED characteristics of the proposed PD with 4QAM (4-PSK), Eb/N0=3dB, and OF=0. In simulation results jitter performance is investigated. Each TEDC curve is derived on 512 symbols. 64 curves are shown in each figure. The amount of jitter can be estimated from the width of positive zero crossing area. In all cases the peak-to-peak jitter does not cross 4% of Ul (sampling period; unit interval).

By careful design of PLL loop this jitter can be kept in acceptable limits. Based on

publications and experience, such jitter performances are within applicable limits.

Fig. 9 shows further TED characteristics of a phase detector PD with 4QAM and Eb/N0=3dB.

Fig. 10 shows further TED characteristics of a phase detector PD with 16QAM and

Eb/N0=6dB.

Fig. 1 1 shows further TED characteristics of a phase detector PD with 64QAM and

Eb/N0=10dB.

Fig. 12 shows further TED characteristics of a phase detector PD with 4QAM and

Eb/N0=3dB.

4QAM modulation format is simulated for ROF=0. The frequency offset varies from 0 to 1 GHz in step of 100MHz. Baud rate is 28GB. In Fig. 12 results indicate immunity with respect to a large frequency offset. Even a frequency offset of 1 GHz does not influence TEDC, i.e. the curve with smallest maximum absolute value. Fig. 13 shows Nyquist super-channel clocking. In coherent optical communication spectral efficiency is increased using higher modulation formats, both polarizations and dense channel packaging, like OFDM, Nyquist, etc. However, a group of N transmitters is integrated to save power, size and price.

Instead of N VCOs a super-channel can share one VCO for all transmitters. In Fig. 13 one VCO supplies the clock distribution unit (CDU) that clocks all N transmitters. The receiver side gains from channel integration. Instead of the number of N only one PD is implemented to support clock extraction. One PD connected to data of the first receiver or any other receiver extracts clock tone information that is later filtered by low-pass filter (LPF) and sent to VCO. PD gain, LPF parameters and VCO gain control the timing performance, such as loop bandwidth, dumping, etc.

Fig. 14 shows a parallel implementation of a timing recovery apparatus. The timing recovery apparatus comprises a grouper 1401 for grouping digital signal values of a digital signal to obtain groups of values, each group of values comprising a first digital value and a second digital value; a multiplier 1403 for multiplying a first digital value with a conjugated version of a second digital value of each group of values to obtain a first group value representing each group of values, and for multiplying a second digital value of each group of values with a conjugated version of a first digital value of a respectively adjacent group of values to respectively obtain a second group value representing adjacent group of values; a processor 1405 for respectively processing first group values respectively representing adjacent groups of values and a second group values representing the adjacent groups of values to obtain a plurality of processed values; and an averager for determining an average value from the plurality of processed values.

High symbol rates require parallel implementation of DSP algorithms including the clock extraction algorithm. Data are grouped in blocks of N symbols. When two-fold oversampling is employed the total number of samples per block is 2N. In Fig. 14, two samples within one symbol interval are denoted by a and b.

Maximum number of PD outputs per data block is equal to N. As shown in Fig.14, the last output requires first two samples from next data block. In real applications, not all data have to be processed by PDs. Usually, some portion of data are unused that might sacrifice clock performance. High modulation formats require higher signal-to-noise ratios, and the reducing the number of data to be processed in the clock extraction engine does not significantly influence timing performance.

PD outputs denoted by v are summed up and filtered by a loop filter (low-pass filter) 1407 that can have a proportional-integral structure. Filter 1407 outputs control VCO frequency and phase. The VCO output then clocks the subsequent ADC circuit.

Fig. 15 shows an example timing recovery method, wherein digital signal values of a digital signal are grouped 1501 to obtain groups of values, each group of values comprising a first digital value and a second digital value. Then a first digital value is multiplied 1503 with a conjugated version of a second digital value of each group of values to obtain a first group value representing each group of values, and a second digital value of each group of values is multiplied with a conjugated version of a first digital value of a respectively adjacent group of values to respectively obtain a second group value representing adjacent group of values. Moreover, first group values respectively representing adjacent groups of values and a second group values representing the adjacent groups of values are processed 1505 to obtain a plurality of processed values. Further, an average value is determined 1507 from the plurality of processed values.

According to another perspective, a timing recovery apparatus embodiment is provided, which comprises a processor configured to perform the method embodiment as shown in Fig. 15.

Many alternatives, modifications, and variations will be apparent to those skilled in the art in light of the above teachings. Of course, those skilled in the art readily recognize that there are numerous applications of the invention beyond those described herein. While the present inventions has been described with reference to one or more particular embodiments, those skilled in the art recognize that many changes may be made thereto without departing from the scope of the present invention. It is therefore to be understood that within the scope of the appended claims and their equivalents, the inventions may be practiced otherwise than as specifically described herein.

Claims

CLAIMS:

1 . Timing recovery apparatus, comprising: a grouper (1401 ) configured for grouping digital signal values of a digital signal to obtain groups of values, each group of values comprising a first digital value and a second digital value; a multiplier (1403) configured for multiplying a first digital value with a conjugated version of a second digital value of each group of values to obtain a first group value representing each group of values, and configured for multiplying a second digital value of each group of values with a conjugated version of a first digital value of a respectively adjacent group of values to respectively obtain a second group value representing adjacent group of values; a processor (1405) configured for respectively processing first group values respectively representing adjacent groups of values and a second group values representing the adjacent groups of values to obtain a plurality of processed values; and an averager (1405) configured for determining an average value from the plurality of processed values.

2. The timing recovery apparatus of claim 1 , wherein the grouper (1401 ) is configured to group subsequent N values of the digital signal to obtain a respective group of values.

3. The timing recovery apparatus of claim 1 or 2, wherein the grouper (1401 ) comprises an oversampler configured for oversampling an input optical signal to obtain the digital signal.

4. The timing recovery apparatus of any of the preceding claims, wherein the multiplier (1403) comprises a plurality of digital multipliers.

5. The timing recovery apparatus of any of the preceding claims, wherein the processor (1405) comprises, for each group of values, a subtractor configured for subtracting first group values representing adjacent groups of values, and a multiplierconfigured for multiplying the subtraction result with a respective second group of values representing these adjacent group of values.

6. The timing recovery apparatus of any of the preceding claims, wherein the averager (1405) is configured to average over real parts of the plurality of processed values to determine the average value.

7. The timing recovery apparatus of any of the preceding claims, further comprising a loop filter (1407) configured for filtering the average value.

8. Communication receiver apparatus, comprising: a receiver configured for receiving a communication signal; the timing recovery apparatus of any of the preceding claims configured for timing recovery upon the basis of the communication signal; and a voltage controlled oscillator configured for providing a clock signal, the being voltage controlled oscillator controlled by an output of the recovery apparatus.

9. The communication receiver apparatus of claim 8, wherein the clock signal is provided to the receiver.

10. The communication receiver apparatus of claim 8 or 9, comprising a plurality of receivers, wherein the clock signal is distributed to the plurality of receivers.

1 1 . The communication receiver apparatus of claims 8 to 10, further comprising a chromatic dispersion compensation unit arranged downstream the voltage controlled oscillator.

12. Timing recovery method, comprising: grouping (1501 ) digital signal values of a digital signal to obtain groups of values, each group of values comprising a first digital value and a second digital value; multiplying (1503) a first digital value with a conjugated version of a second digital value of each group of values to obtain a first group value representing each group of values, and multiplying a second digital value of each group of values with a conjugated version of a first digital value of a respectively adjacent group of values to respectively obtain a second group value representing adjacent group of values; respectively processing (1505) first group values respectively representing adjacent groups of values and a second group values representing the adjacent groups of values to obtain a plurality of processed values; and determining (1507) an average value from the plurality of processed values.

13. Method for generating a clock signal from a received communication signal, the method comprising: performing the timing recovery method of claim 12 to provide average values; and controlling the average values to obtain the clock signal.

14. A computer program with a program code for performing the method according to claim 12 or 13 when run on a computer.

15. A timing recovery apparatus, comprising a processor configured to perform the method of any of claims 12 to 13.