RECEIVER FOR COHERENT OPTICAL TRANSPORT SYSTEMS BASED ON ANALOG SIGNAL PROCESSING AND METHOD THEREOF

Abstract

The present invention discloses a receiver for coherent optical transport systems based on analog signal processing and the method of recovering transmitted data by processing signals in electronic domain. In the present invention, high-speed electrical signals obtained from optical-to-electrical converters which carry transmitted data information in a coherent transport system are jointly processed in analog domain itself without converting these signals to the digital domain using high speed ADCs. Different processing steps which may include carrier phase & frequency offset recovery and compensation, polarization mode dispersion and/or chromatic dispersion, clock & data recovery and deserialization may be performed while keeping the information signals in analog domain itself.

Description

FIELD OF THE INVENTION

The present invention generally relates to receivers for coherent optical data transmission systems and more particularly to a method of recovering transmitted data by processing signals in electronic domain.

BACKGROUND OF THE INVENTION

Optical communication links have to employ polarization multiplexing and coherent modulation & detection for increasing data rates in optical transport systems. Coherent links carrying more than 100-Gbps over single optical carrier have been demonstrated till date. Typically, receivers for coherent, polarization multiplexed transmission systems use ultra-fast ADCs (analog-to-digital converters) that convert electrical signals to the digital domain. These signals are then jointly processed using digital signal processors (DSPs) to recover the transmitted data.

In prior art, digitization of such high-speed signals is extremely difficult. Once the digitization is done, processing of these signals is also very challenging, since massive amount of computation is required to meet the extremely high throughput rates for real-time operation. Digitization of the signals obtained at the receivers of high-speed coherent optical links is extremely difficult because of the large bandwidth required for this purpose. High-speed ADCs add significant overheads in terms of power consumption, design complexity, chip area and cost.

In the view of the foregoing, there is a need for a receiver for processing data signals in coherent transmission system, in analog domain itself, without first converting them to digital domain using high-speed ADCs.

OBJECTS OF THE INVENTION

• • 1. It is the primary object of the present invention to provide a receiver for coherent optical transmission system.
  • 2. It is another object of the present invention to provide a method for recovering transmitted data in data transmission system.
  • 3. It is another object of the present invention to provide a system and method for processing signals in analog domain.
  • 4. It is another object of the present invention to provide a method for joint equalization of high speed signals.
  • 5. It is another object of the present invention to provide a method of carrier phase recovery & compensation, and clock & data recovery from the high speed signals.
  • 6. It is another object of the present invention to provide a receiver which consumes significantly less power.
  • 7. It is another object of the present invention to provide a receiver which requires less chip area.
  • 8. It is another object of the present invention to provide a receiver with significantly lower cost.

SUMMARY OF THE INVENTION

The present invention discloses a receiver for coherent optical transport systems based on analog signal processing and the method of recovering transmitted data by processing signals in electronic domain. In the present invention, high-speed electrical signals obtained from optical-to-electrical converters which carry transmitted data information in a coherent transport system are jointly processed in analog domain itself without converting these signals to the digital domain using high speed ADCs. Different processing steps which may include carrier phase & frequency offset recovery and compensation, polarization mode dispersion and/or chromatic dispersion, clock & data recovery and deserialization may be performed while keeping the information signals in analog domain itself. The receiver in accordance with the present invention consumes significantly less power and has less chip area hence requires less space with significantly lower cost.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is block diagram of a receiver for transmission system employing coherent modulation and polarization multiplexing and analog signal processing in accordance with an aspect of the present invention.

FIG. 2 is block diagram of a feed forward equalizer (FFE) for joint equalization of the signals received in a polarization multiplexed coherent optical link in accordance with an aspect of the present invention.

FIG. 3 is block diagram of a decision feedback equalizer (DFE) for joint equalization of the signals received in a polarization multiplexed coherent optical link in accordance with an aspect of the present invention.

FIG. 4 is block diagram of costas loop based carrier phase recovery and correction architecture for QPSK signals in accordance with an aspect of the present invention.

FIG. 5 is an illustration of a practical implementation of carrier phase recovery and equalizer blocks combined together in accordance with an aspect of the present invention.

FIG. 6 is block diagram of analog coherent receiver for short distance links in accordance with an aspect of the present invention.

FIG. 7 is block diagram of analog coherent receiver with LMS based equalizer for long distance links in accordance with an aspect of the present invention.

FIG. 8 is block diagram of analog coherent receiver with CMA based equalizer for long distance links in accordance with an aspect of the present invention.

DETAILED DESCRIPTION OF THE INVENTION

The present invention discloses a receiver for coherent optical transport systems based on analog signal processing and the method of recovering transmitted data by processing signals in electronic domain. The working of the receiver is explained in detail with reference to the accompanying drawings in accordance with an aspect of the present invention.

The high-speed electrical signals obtained from optical-to-electrical converters, which carry the transmitted data information in a coherent transport system, are jointly processed in the analog domain itself (without first conversion of these signals to the digital domain using high speed ADCs), as shown in FIG. 1. Different processing steps, which may include (but may not be limited to) carrier phase & frequency offset recovery and compensation, polarization rotation compensation, equalization to compensate for polarization mode dispersion and/or chromatic dispersion, clock & data recovery and deserialization, may be performed while keeping the information signals in “analog domain” itself. However, the order in which these processing steps are undertaken is not important, as far as this invention is concerned. The “analog domain signals” being referred to here may be defined as the signals that are represented by only one pair of differential voltages or one single ended voltage.

Optical signal from the fibre channel is split into two orthogonal polarizations using a polarization beam splitter (PBS) and each polarization signal is fed to a 90° optical hybrid where it is mixed with the optical signal from local oscillator (LO) laser. The signals on in-phase and quadrature-phase components of the orthogonally polarized received optical fields are delivered to balanced photo detectors (PD), where those are converted to electrical signals. These four signals are fed to an analog signal processing unit, which essentially consists of equalizer, carrier phase recovery & compensation (CPR) and clock & data recovery (CDR) modules.

The description of analog processing modules is explained below in accordance with an aspect of the present invention.

Analog Signal Processing Equalizer:

Since chromatic dispersion (CD) and polarization mode dispersion (PMD) are linear in nature, they can easily be compensated using tapped delay line equalizers (that act as linear transversal filters). Linear equalizer primarily has three unit operations—delay, multiplication and addition. In analog domain, high-speed multiplication can easily be performed using a variable gain amplifier or a Gilbert cell with less than 10 transistors. Similarly, summation of signals can be performed by simply adding currents onto a resistor. In addition, filter coefficients can be represented by charge stored on capacitors.

For dual polarization signals, the multidimensional transversal filter used to equalize the linear dispersive effects can be described by the following equations:

{tilde over (x)}=h _xx ^T x+h _yx ^T y  (1.1)

{tilde over (y)}=h _xy ^T x+h _yy ^T y  (1.2)

Where,

x and y are column vectors of the delayed complex input electrical signals corresponding to two polarization channels {tilde over (x)} and {tilde over (y)} are the complex equalized outputs. h_xx, h_xy, h_yx, h_yy are the column vectors of the complex filter coefficients.

Update of Equalizer Coefficients:

The update of complex filter coefficients can be done using any of the adaptive signal processing algorithms. For example, update equations using least mean square algorithm (LMS) and constant modulus algorithm (CMA) are explained below.

Least Mean Square Algorithm:

The LMS update equations for complex coefficients of the multidimensional equalizer can be represented as:

h _xx,k(t)=β∫^tx(τ−kτ_d)·e_x*(τ)·dτ  (1.3)

h _yx,k(t)=β∫^ty(τ−kτ_d)·e_x*(τ)·dτ  (1.4)

h _xy,k(t)=β∫^tx(τ−kτ_d)·e_y*(τ)·dτ  (1.5)

h _yy,k(t)=β∫^ty(τ−kτ_d)·e_y*(τ)·dτ  (1.6)

where, x(t)=x_I(t)+jx_Q(t) & y(t)=y_I(t)+jy_Q(t) are input signals to the equalizer in X and Y polarizations, e_x(t)=e^I_x(t)+je^Q_x(t) & e_y(t)=e^I_y(t)+je^Q_y (t) are error signals in X and Y polarizations, T_d is the delay provided by a delay stage, and 0≦k≦L. Here L is the number of delay stages per dimension of the equalizer. The real and imaginary parts of h_xx,k(t) in Equation 1.3 are given by:

Re{h _xx,k(t)}=β∫^ι[x_I(τ−kτ_d)·e_x^I(τ)+x_Q(τ−kτ_d)·e_x^Q(τ)]dτ  (1.7)

Im{h _xx,k(t)}=β∫^t[x_Q(τ−kτ_d)·e_x^I(τ)+x_I(τ−kτ_d)·e_x^Q(τ)]dτ  (1.8)

Similarly, expression for other coefficients can also be determined.

Constant Modulus Algorithm:

The CMA update equations for complex coefficients of the multidimensional equalizer can be represented as:

h _xx,k(t)=β∫^tx(τ−kτ_d)·{tilde over (x)}*(τ)·(1−|{tilde over (x)}|²)·dτ  (1.9)

h _yx,k(t)=β∫^ty(τ−kτ_d)·{tilde over (x)}*(τ)·(1−|{tilde over (x)}|²)·dτ  (1.10)

h _xy,k(t)=β∫^tx(τ−kτ_d)·{tilde over (y)}*(τ)·(1−|{tilde over (y)}|²)·dτ  (1.11)

h _yy,k(t)=β∫^ty(τ−kτ_d)·{tilde over (y)}*(τ)·(1−|{tilde over (y)}|²)·dτ  (1.12)

where, x(t)=x_I(t)+jx_Q(t) & y(t)=y=(t)+jy(t) are input signals to the equalizer in X and Y polarizations, x^˜ & ˜y are equalized signals in X and Y polarizations, T_d is the delay provided by a delay stage, and 0≦k≦L. Here L is the number of delay stages per dimension of the equalizer.

Equalizer Architecture:

The equalizer can be implemented as a feed forward equalizer (FFE) or as a decision feedback equalizer (DFE). The description of FFE and DFE are explained below in detail.

Feed Forward Equalizer:

In the proposed invention joint equalization of the high-speed signals can be performed using a feed forward equalizer (FFE), which involves addition of the four high-speed analog signals (obtained from the photo-detectors) and their delayed versions (with different delays) with proper weight coefficients, in terms of voltages or currents. As a result, four new analog signals are obtained and are termed as equalizer outputs. Threshold detectors can be used on these signals to obtain the estimates of transmitted data symbols. The number of delay cells (and taps) can be varied depending on the requirement, and the figure just illustrates some examples. FIG. 2 shows the block diagram of a feed forward equalizer having 3 taps. Here the delay line is having two delay stages with four delay cells (one for each real signal) in each stage. Each delay stage delays the signal by T_d seconds. Signals tapped from the delay line are multiplied with the filter coefficients in butterfly multipliers as shown in the block diagram. Outputs from butterfly multipliers are added together to generate equalized signals, ˜x_I(t); ˜x_Q(t); ˜y_I(t) and ˜y_Q(t). Weight coefficient adjustment is done using adaptive algorithms.

Decision Feedback Equalizer:

In another embodiment of the present invention the joint equalization can also be performed using a decision feedback equalizer (DFE) with the analog processing method, as shown in FIG. 3. In the DFE, the delayed copies of the “analog signals” obtained after threshold detection from the FFE (with different delays) are fed back and added, with proper weights, to the signals at the inputs of the threshold detectors. The number of delay cells (and taps) can be varied depending on the requirement.

Decision feedback equalizer has both feed forward and feedback taps. FIG. 3 shows the block diagram of a DFE having 3 feed forward taps and 1 feedback tap. Here, decisions are made on the equalized signals ˜x_I (t); ˜x_Q(t); ˜y_I (t) and ˜y_Q(t). Delayed and weighted versions of the decision are fed back to the summation block of the equalizer. Adaptive algorithms as described in equalizer coefficients section are used for weight coefficient adjustment.

Weight coefficients required for adding the signals have to be obtained adaptively by another block in the equalizer (i.e. the Weight Coefficient Adjustment block) as shown in FIG. 2. Different algorithms can be used for adjustment of the weight coefficients, such as CMA (constant modulus algorithm), LMS (least mean squared) algorithms etc. Since these weight coefficients themselves have to vary at a much slower rate (according to the dynamics of the channel) than the high-speed data carrying signals, these coefficients can themselves be easily updated and stored in the digital domain. In another embodiment, it is possible to update these signals in the analog domain itself. In this case, the algorithms mentioned above for weight coefficient adjustment can be implement in continuous time mode using different operations (such as addition, subtraction, multiplication etc.) in analog domain, and the storage of weight coefficients on capacitive elements.

Carrier Phase Recovery and Compensation:

In the present invention, the operation of carrier phase recovery and compensation is also performed in the analog domain. The four signals obtained from the equalizer, i.e. two each corresponding to the two orthogonal polarizations, are sent to two carrier phase recovery and correction blocks, each corresponding to one polarization. Architectures based on different algorithms, such as m^th power method, reverse modulation technique, or Costas Loop can be employed for these blocks.

A Costas Loop based technique for Quadrature Phase Shift Keying (QPSK) modulation is shown in FIG. 4. In this method, for each pair of incoming signals from the equalizer, i.e. (x_I, x_Q) and (y_I, y_Q), the architecture shown in FIG. 4 can be used. The phases of the incoming signals contain the time varying carrier phase (and frequency) offset represented by the phase Φ_in(t) and the message phase represented by Φ_m(t). Single sideband mixing of these signals with the QVCO (Quadrature phase Voltage Controlled Oscillator) outputs results in subtraction of Φ_fb(t) from the phases of x_I and x_Q. The whole system behaves like a Phase Locked Loop. There is a small residual phase error Φerr(t) in the outputs of the SSB mixers. Applying threshold detectors (or limiters) to the signals obtained from the SSB mixers results in the estimates of the transmitted data symbols. Further, multiplication of these signals with the signals from the SSB mixers, followed by subtraction, as shown in FIG. 4, results in a signal which is roughly proportional to Φ_err(t), when Φ_err(t) is small. The signal, goes to a loop filter, followed by QVCO that generates cos(Φ_fb(t)) and sin(Φ_fb(t)) for the single sideband mixers.

The implementation of different blocks in of the carrier phase recovery module may be done completely in the analog domain. However, some components, such as the QVCO and the loop filter, can be implemented in the digital domain as well, although, the high speed data carrying signals are not digitized and remain in the analog domain only to avoid high speed ADCs and DSP. The threshold detection (or limiter) operation is not being considered as an ADC operation here (since the number of bits at the output of the threshold detector) is not more than the number of bits represented by the analog signal.

In the proposed receiver decision assisted version of Costas phase-locked loop (PLL) has been used for carrier phase recovery. The block diagram of carrier phase recovery is shown in FIG. 4. Normalized in-phase and quadrature-phase components of the input are given by:

x _I=cos(φ_off(t)+φ_m(t))  (1.13)

x _Q=sin(φ_off(t)+φ_m(t))  (1.14)

where, Φ_off (t)=(w_s−w_lo)t+Φ₁−Φ₂ is the resultant phase offset between the incoming carrier and the LO signal. The output of balanced detector is mixed with the local oscillator signal using a single side band (SSB) mixer. The output I and Q components of the SSB mixer are given as:

I _SSB=cos(φ_m(t)+φ_err(t))  (1.15)

Q _SSB=sin(φ_m(t)+φ_err(t))  (1.16)

where, Φ_err(t)=Φ_off (t)−Φ_fb(t), which will be negligible once locking is achieved. Here, Φ_fb(t) is the phase of the quadrature voltage controlled oscillator (QVCO) outputs. Decisions are made on I_SSB and Q_SSB and passed onto multipliers which will act as correlators. Difference between the product signals, v_e=sin(Φ_err(t)) is used to control frequency and phase of the QVCO after passing through a loop filter. If the error is very small, it can be approximated as Φ_err(t). When QVCO is locked to the beat signal of carrier and local oscillator, the error voltage becomes zero, and the output of SSB mixer gives baseband signal without phase offsets.

In the present invention, it is not necessary that the equalizer block, the carrier phase recovery & compensation block, and the clock & data recovery block are placed sequentially in the flow the data signals. For example, as shown in FIG. 5, a practical implementation may involve a feed-forward equalizer block, followed by a carrier phase recovery and correction block, which is followed by decision feedback block which feeds signals back to the outputs of the feed-forward block. The final signals coming out of these blocks (D_Ix, D_Qx, D_Iy and D_Qy), which represent the estimates of the transmitted signals, are sent to clock and data recovery circuits, which may be implemented using conventional techniques, to finally get the desired data signals.

Depending on the amount of dispersion that needs to be compensated, the proposed analog receiver coherent receiver can have different forms as explained below.

Analog Coherent Receiver for Short Distance Links:

Architecture of the proposed analog coherent receiver is shown in FIG. 6. Since dispersion is very low in short distance links, a polarization rotator (PR) block can be used instead of equalizer. In-phase and quadrature-phase components of EX (E_X,I and E_X,Q) and E_Y (E_Y,I and E_Y,Q) are jointly processed in analog domain using PR and CPR modules to retrieve the transmitted data. PR module rotates the polarization by an angle α, where −α is the angle of polarization rotation introduced in the channel and receiver PBS. The angle α obtained from the CPR module is used for generating cos α and sin α. The outputs of the PR module are given by:

$\left(\begin{array}{c}{X}_{\alpha }\\ Y_{\alpha }\end{array}\right)=\left(\begin{array}{cc}\cos \alpha & -\sin \alpha \\ \sin \alpha & \cos \alpha \end{array}\right)·\left(\begin{array}{c}{E}_{X,I}+jE_{X,Q}\\ E_{Y,I}+jE_{Y,Q}\end{array}\right)$

Where, X_α=X_α,I++j X_α,Q and Y_α=Y_α,I+j Y_α,Q. the signals X_α and Y_α are then fed to the CPR module.

The CPR module has two independent Costas loops—one for each polarization. Each of the incoming signals to the CPR module is passed through a single side band mixer (SSBM), where it is getting mixed with the quadrature signals (cos Φ and sin Φ) corresponding to the phase error Φ due to carrier phase & frequency offsets. Input signals to the SSBMs for X and Y polarizations can be represented as:

X _α=cos [φ_m,X(t)+φ_X(t)]+j sin [φ_m,X(t)+φ_X(t)]

Y _α=cos [φ_m,Y(t)+φ_Y(t)]+j sin [φ_m,Y(t)+φ_Y(t)]

where, Φ_m,X(t) & Φ_m,Y(t) are the phases corresponding to the transmitted messages and Φ_X(t) & Φ_y (t) are the phase errors in X and Y polarizations respectively. The in-phase and quadrature-phase signals at SSBM outputs are given as:

X _I,SSBM=cos [φ_m,X(t)+φ_res,X(t)]

X _Q,SSBM=sin [φ_m,X(t)+φ_res,X(t)]

Y _I,SSBM=cos [φ_m,Y(t)+φ_res,Y(t)]

Y _Q,SSBM=sin [φ_m,Y(t)+φ_res,Y(t)]

where, Φ_res,X and Φ_res,Y are the negligible residual phase errors in X and Y polarizations respectively.

Decision is made on the in-phase and quadrature-phase components of the SSBM outputs and fed to multipliers which act as cross-correlators. Phase error is calculated by subtracting the multiplier outputs and is passed through a loop filter to generate control voltage for quadrature voltage controlled oscillator (QVCO). A similar cross-correlator structure is used for finding the polarization rotation angle α.

The PR and CPR modules are essentially phase-locked loops (PLLs), the parameters of which can be optimized based on the dynamics of the system, such as laser line widths and speed of polarization rotation. The estimated signals X′ and Y′ corresponding to X and Y polarizations are passed on to a clock and data recovery (CDR) module which re-times (and deserializes) the transmitted data.

Analog Coherent Receiver with Equalizer for Longer Distance Links:

For longer distance optical links, where channel dispersion is significant, an equalizer need to be incorporated in the receiver. Two versions of such a receiver are described in detail with reference to the drawings in accordance with an aspect of the present invention.

Receiver with LMS Equalizer:

FIG. 7 shows the block diagram of an analog coherent receiver using an LMS based equalizer. In this architecture, decisions made in the CPR modules are fed back to the equalizer module after rotating by an angle −Φ, where Φ is the input angle to the QVCO. The fed back signals can be used as training signals or as decision signals for the DFE implementation of LMS equalizer.

Receiver with CMA Equalizer:

CMA equalizer doesn't need decisions made on the signals for convergence. Hence, CMA equalizer can always be implemented as an FFE structure. An analog coherent receiver employing CMA equalizer is shown in FIG. 8. In this architecture, signals from the photo detector are equalized in a CMA based FFE and the outputs are fed to two independent CPRs (one for each polarization). Rest of the signal processing is similar to the previous architectures.

Claims

1. A receiver for coherent optical transport systems based on analog signal processing, comprising: a. an analog signal processing module.wherein, said analog signal processing module further comprising: i. an equalizer module;ii. a carrier recovery and phase compensation module; andiii. a clock and data recovery module.

2. The receiver for coherent optical transport systems based on analog signal processing as claimed in claim 1, wherein chromatic dispersion (CD) and polarization mode dispersion (PMD) compensation are carried out in said equalizer module.

3. The receiver for coherent optical transport systems based on analog signal processing as claimed in claim 1, wherein update of equalizer coefficients are done using any one of adaptive signal processing algorithms.

4. The receiver for coherent optical transport systems based on analog signal processing as claimed in claim 3, wherein said signal processsing algorithms can be preferably least mean square algorithm (LMS) or constant modulus algorithm (CMA).

5. The receiver for coherent optical transport systems based on analog signal processing as claimed in claim 1, wherein said equalizer can be implemented as a feed forward equalizer (FFE) or as a decision feedback equalizer (DFE).

6. The feed forward equalizer (FFE) as claimed in claim 5, further comprising: a. at least one tap; andb. a delay line for more than one said tap comprising an delay stage with an delay cells in each of said delay stage.

7. The decision feedback equalizer (DFE) as claimed in claim 5, comprising both feed forward and feedback taps.

8. The receiver for coherent optical transport systems based on analog signal processing as claimed in claim 1, wherein a decision assisted version of costas phase-locked loop (PLL) is used in said carrier phase recovery module.

9. The carrier phase recovery module as claimed in claim 8, comprising: a. at least one quadrature voltage controlled oscillator;b. at least one single side band (SSB) mixer; andc. at least one loop filter.

10. A method of recovering transmitted data by processing signals in analog domain, comprising the steps of: a. performing joint equalization of signals using said equalizer;b. performing carrier phase recovery and compensation to said signals obtained from said equalizer; andc. recovering said transmitted data using said clock and data recovery circuits.

11. The method of recovering transmitted data by processing signals as claimed in claim 10, wherein said joint equalization is performed using said feed forward equalizer (FFE) or said decision feedback equalizer (DFE).

12. The joint equalization using feed forward equalizer (FFE) as claimed in claim 11, wherein said analog signals obtained from photo detector and their delayed signals obtained from said delay cells multiplied with weight coefficients are added to obtain equalizer outputs.

13. The joint equalization using feed forward equalizer (FFE) as claimed in claim 12, wherein estimates of transmitted data symbols from said equalizer outputs are obtained using threshold detectors.

14. The joint equalization using decision feedback equalizer (DFE) as claimed in claim 11, wherein said equalized signals obtained from said feed forward equalizer (FFE) after threshold detection by said threshold detectors along with proper weights are fedback and added to said signals at inputs of said threshold detectors.

15. The method of recovering transmitted data by processing signals as claimed in claim 11, wherein said weight coefficients are obtained using said adaptive signal processing algorithms.

16. The method of recovering transmitted data by processing signals as claimed in claim 10, wherein said signals from said equalizer are sent to said costas loop carrier phase recovery and correction module.

17. The method of recovering transmitted data by processing signals as claimed in claim 16, wherein single sideband mixing of phases of said signals containing a time varying carrier phase offset and a message phase using said SSB mixers with said quadrature phase voltage controlled oscillator (QVCO) outputs.

18. The method of recovering transmitted data by processing signals as claimed in claim 17, wherein said threshold detectors are applied to said output signals from said SSB mixers produces estimates of said transmitted data signals.

19. The method of recovering transmitted data by processing signals as claimed in claim 18, wherein said estimates of said transmitted data signals are sent to said clock and data recovery module which produces said transmitted data.