CN103780519B - Channel equalization and frequency deviation estimation joint parallel method based on LMS - Google Patents
Channel equalization and frequency deviation estimation joint parallel method based on LMS Download PDFInfo
- Publication number
- CN103780519B CN103780519B CN201410007137.XA CN201410007137A CN103780519B CN 103780519 B CN103780519 B CN 103780519B CN 201410007137 A CN201410007137 A CN 201410007137A CN 103780519 B CN103780519 B CN 103780519B
- Authority
- CN
- China
- Prior art keywords
- signal
- parallel
- branch road
- data
- offset estimation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/03—Shaping networks in transmitter or receiver, e.g. adaptive shaping networks
- H04L25/03006—Arrangements for removing intersymbol interference
- H04L25/03178—Arrangements involving sequence estimation techniques
- H04L25/03248—Arrangements for operating in conjunction with other apparatus
- H04L25/03273—Arrangements for operating in conjunction with other apparatus with carrier recovery circuitry
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/03—Shaping networks in transmitter or receiver, e.g. adaptive shaping networks
- H04L25/03006—Arrangements for removing intersymbol interference
- H04L25/03012—Arrangements for removing intersymbol interference operating in the time domain
- H04L25/03114—Arrangements for removing intersymbol interference operating in the time domain non-adaptive, i.e. not adjustable, manually adjustable, or adjustable only during the reception of special signals
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/03—Shaping networks in transmitter or receiver, e.g. adaptive shaping networks
- H04L25/03006—Arrangements for removing intersymbol interference
- H04L2025/03592—Adaptation methods
- H04L2025/03598—Algorithms
- H04L2025/03611—Iterative algorithms
- H04L2025/03636—Algorithms using least mean square [LMS]
Abstract
The invention discloses a channel equalization and frequency deviation estimation joint parallel method based on LMS. At the initialization phase of an optical receiver, train sequence signals are converted into parallel signals and are sent into parallel signal processing branches for balance and frequency deviation estimation. The frequency deviation estimation average value of all branches and an error signal of each branch are computed. A unified equalizer tap coefficient is used for each group of parallel signals. When the equalizer tap coefficient is updated, the average value of the branch error signals is used. At a data transmission phase, a transmitting end inserts train symbols into data symbols. The optical receiver converts data signals into parallel signals. The train signals use balance signals and known train symbols to carry out frequency deviation estimation. The data signals use accumulation phase error corresponding to the train signals and the acquired frequency deviation estimation average value to carry out compensation and decision, and uses the balance signals and decision signals to carry out frequency deviation estimation. According to the invention, signal parallel processing is used, which reduces the limitation influence of data signal processing hardware on the system performance.
Description
Technical field
The invention belongs to optical burst receiver technical field, more specifically, it is related to one kind and is based on lms (least
Mean square, least mean square algorithm) channel equalization and offset estimation joint parallel method.
Background technology
In current high speed coherent optical communication system, pdm-qpsk (palarization multiplexing-four phase absolute phase shift keying) is concerned with
Optical transmission system is one of most potential technical scheme.It is concerned with optical transmission system in pdm-qpsk, transmission signal is mainly subject to
Chromatic dispersion (chromatic dispersion, cd) and polarization mode dispersion (polarization mode to optical fiber
Dispersion, pmd) linear damage and frequency shift (FS) produced by the laser instrument of sending and receiving end impact, this two problems are tight
Affect the service behaviour of photoreceiver again.And adaptive balancing technique can substantially eliminate the intersymbol string being brought by dispersion
Disturb, frequency offset estimation technique can be used to solve the impact that frequency deviation is brought.Due to can influence each other between balanced device and frequency offset estimator, because
This can be using the unified algorithm of time domain equalization and offset estimation.
For light burst transmission system, the feature of light burst requires balanced device in optical burst receiver will be capable of quickly
Convergence.Fig. 1 is the system block diagram of optical burst receiver.As shown in figure 1, input signal r (t) of receiver is two-way polarization side
Signal to the fibre channel transmission through polarization coupled and through certain distance for the orthogonal smooth pdm-qpsk signal.?
During fibre channel transmission, optical signal can be affected by factors such as dispersion, polarization mode dispersion, noise of optical amplifier, leads to
The Quality Down of transmission signal.Pdm-qpsk signal r (t) is entered 90 degree together with ftlo (fast tunable laser instrument) light wave and mixes
Ripple device carries out coherent demodulation.Coherent demodulation Hou tetra- road signal carries out ad sampling and quantization.Sampled and quantify after output 4 tunnels
Signal ix, qx, iy, qy represent homophase and the orthogonal demodulation signal of two polarization states x, y respectively, and this 4 road signal enters numeral letter
Number processing module carries out channel equalization (channel equalization) and offset estimation (frequency offset
Estimation, foe) and compensate, last phase place judgement recovers sent data.
Channel equalization based on lms and the unified algorithm of offset estimation, are a kind of having of raising coherent optical heterodyne communicatio performance
Efficacious prescriptions method.But in the design of optical burst receiver, using fpga, (field-programmable gate array shows
Field programmable gate array) or when special IC realizes digital signal processing algorithm, calculating speed and chip area are two
The subject matter mutually restricting.Therefore, it is necessary in performance and realize making a choice between complexity.Height due to fiber optic communication
Speed feature, taking the pdm-qpsk fibre-optic transmission system (FOTS) of 112gb/s as a example, the symbol on each road of coherent demodulation Hou tetra- road signal
Speed is 28g/s, and 4 branch road electric signals are sampled firstly the need of the ad carrying out dual-rate and quantified, each tributary signal speed
Rate is up to 56g/s, so the symbol entering balanced device is the discrete signal that character rate is 56g/s.At follow-up data signal
Reason unit (dspu) cannot realize the process to this speed on hardware, so must be by the way of parallel processing, according to defeated
Enter the speed of data and the processing speed of chip, parallel branch number is possibly used larger numerical value, this requires in real time should
Used time, algorithm must is fulfilled for the requirement of parallel processing.The renewal of equalizer tap coefficient simultaneously and the frequency deviation based on pre- judgement
Algorithm for estimating is required for the feedback of signal, and the time delay being caused by feedback in Parallel Implementation is also very big to the performance impact of system.
Therefore, implementing in design it is necessary to consider shadow parallel and that delay of feedback is to receiver performance in optical burst receiver
Ring.
Content of the invention
It is an object of the invention to overcoming the deficiencies in the prior art, a kind of channel equalization based on lms and frequency deviation is provided to estimate
Meter joint parallel method, reduces data-signal and processes the limitations affect to systematic function for the hardware.
For achieving the above object, the channel equalization based on lms for the present invention and offset estimation joint parallel method, including
Following steps:
S1: initialized using training sequence, including step:
S1.1: send training sequence to optical burst receiver, through the training sequence signal of coherent demodulation and sample quantization
Carry out serial to parallel conversion and obtain n road parallel signal;The corresponding equalizer tap coefficient of the n-th=1 group of parallel signal of setting
S1.2: n-th group parallel signal enters n Parallel signal processing branch road, and each Parallel signal processing branch road includes all
Weighing apparatus and frequency deviation estimating modules, the balanced device of i-th branch road is equalized signalWherein k=(n-1) × n+i, 1
≤i≤n;
S1.3: frequency deviation estimating modules are according to known training symbolTo equalizing signalCarry out offset estimation, accumulated
Phase errorWith offset estimation value
S1.4: by the offset estimation value of n branch roadThe offset estimation averagely being obtained n-th group parallel signal is put down
Average
S1.5:n branch road calculates its error signal respectivelyn,i:
S1.6: the equalizer tap coefficient that (n+1)th group of parallel signal of renewal uses:
Wherein,Represent (n+1)th group, the equalizer tap coefficient that used of n-th group parallel signal respectively;D table
Show the delay of error signal;λ is the iteration step length of setting, is positive number;ncRepresent the participation tap coefficient selecting from n branch road
The error signal quantity calculating, 1≤nc≤ n, 1≤ic≤nc;RepresentCorresponding observation vector,RepresentConjugation;
S1.7: whether training of judgement sequence is disposed, if untreated finish, return to step s1.2 continues with next
Group parallel signal, if be disposed, enters step s2;
S2: the entrance data is activation stage is processed to data, including step:
S2.1: data sending terminal inserts training symbol in data symbol, and its insertion method is: sending symbol with n is
One group, then n transmission symbol is divided into r group, comprise a training symbol, r training in n/r transmission symbol of every group
Sequence number in parallel symbol for the symbol is designated as ir,1≤r≤r;
S2.2: send data-signal to optical burst receiver, the data-signal through coherent demodulation and sample quantization is entered
Row serial to parallel conversion obtains n road parallel signal, and n-th group parallel signal enters n Parallel signal processing branch road, the data is activation stage
Parallel signal processing branch road includes balanced device, frequency deviation estimating modules and judging module, and the equalizer processes of i-th branch road obtain all
Weighing apparatus signal
S2.3: n bar branch road is carried out respectively with offset estimation:
When branch road is for training signal, directly according to known training symbolTo equalizing signalCarry out offset estimation,
Obtain accumulated phase errorWith offset estimation value
When branch road is for data-signal, first to equalizing signalCarry out phase compensation, the signal after phase compensationFor:
Wherein, d represents the delay of offset estimation mean value,Expression rounds up;Judging module is to signalSentenced
Certainly obtain decision signalAccording to decision signalTo equalizing signalCarry out offset estimation, obtain offset estimation valueAnd accumulated phase error
S2.4: by the offset estimation value of n branch roadThe offset estimation averagely being obtained n-th group parallel signal is put down
Average
S2.5:n branch road calculates its error signal respectivelyn,i:
When branch road is training signal, i.e. i=irWhen, error signaln,iFor:
When branch road is for data-signal, error signaln,iFor:
S2.6: the equalizer tap coefficient that (n+1)th group of parallel symbol of renewal uses:
Wherein, λ*It is the iteration step length of data is activation stage setting,Represent that the data is activation stage selects from n branch road
Participate in tap coefficient calculate error signal quantity,
S2.7: judge whether data is disposed, if untreated finish, return to step s2.2 continues with, if place
Reason finishes, and terminates.
Further, the concrete grammar of offset estimation comprises the following steps:
S3.1: calculate equalizing signalAccumulated phase error:
When branch road is for training symbol, accumulated phase errorWherein θkThe phase of training symbol known to expression
Position, φkRepresent equalizing signalPhase place;
When branch road is for data-signal, accumulated phase errorWhereinRepresent data decision signal
Phase place;
S3.2: calculate the offset estimation value of this branch roadWhereinRepresent -1 signal of kth
Accumulated phase error.
The channel equalization based on lms for the present invention and offset estimation joint parallel method, in the initial phase of photoreceiver,
The sampled signal of training sequence signal is converted into after parallel signal by serial to parallel conversion and sends into Parallel signal processing branch road, every
Branch road is equalized and offset estimation respectively, and it is average that the offset estimation value that all branch roads are obtained averagely is obtained offset estimation
Value, every branch road calculates its error signal respectively according to offset estimation mean value, and every group of parallel signal adopts unified balanced device
Tap coefficient, equalizer tap coefficient is updated using the average of branch road error signal when updating;In the data is activation stage, send out
Sending end inserts training symbol in data symbol, and the data-signal through coherent demodulation and sample quantization is obtained by serial to parallel conversion
To parallel signal, training signal carries out offset estimation using equalizing signal and known training symbol, and data-signal is using corresponding instruction
Practice the accumulated phase error of signal and the offset estimation mean value having obtained compensate after reconvict, then using equalizing signal with
Decision signal carries out offset estimation, then using the renewal carrying out equalizer tap coefficient with initial phase same procedure.
The method have the advantages that
(1) present invention passes through parallelization, reduces signal rate, thus reducing data-signal to process hardware to systematicness
The limitations affect of energy;
(2) in initial phase, using offset estimation mean value calculation error signal, balanced device can be improved initialized
Reliability;
(3) in parallel symbol, in the data is activation stage, insert the training symbol of some, judgement, frequently can be improved
Partially estimate the accuracy with error signal feedback;
(4) show, the present invention has good tolerance to the delay of error signal through emulation.
Brief description
Fig. 1 is the system block diagram of optical burst receiver;
Fig. 2 is initial phase parallel signal branch road algorithm schematic diagram;
Fig. 3 is data is activation stage pipeline signal branch algorithm schematic diagram;
Fig. 4 is the equalizer convergence velocity contrast figure of parallel method of the present invention and serial approach;
Fig. 5 is to the convergence rate comparison diagram having computing relay and no computing relay in parallel method of the present invention;
Fig. 6 is the bit error rate comparison diagram of parallel method under serial approach and different delays of the present invention.
Specific embodiment
Below in conjunction with the accompanying drawings the specific embodiment of the present invention is described, so that those skilled in the art is preferably
Understand the present invention.Requiring particular attention is that, in the following description, when known function and design detailed description perhaps
Can desalinate the present invention main contents when, these descriptions will be ignored here.
Embodiment
The present embodiment, still, the signal of 56g/s speed is passed through taking the pdm-qpsk fibre-optic transmission system (FOTS) of 112gb/s as a example
Serioparallel exchange is multi-path parallel signal, herein for 256 tunnels, so that the speed on each road can be effectively reduced, such that it is able to drop
Its physics realization difficulty low.Because for two samplings, judgement is drawn 128 symbols it is seen that needing by therefore each group of 256 road signals
Parallel processing branches quantity n=128 to be configured.
The work of optical burst receiver is divided into two stages: initial phase data transmission phase.It is being detected with light
After burst reaches, optical burst receiver initially enters initial phase, using training sequence, equalizer tap coefficient is changed
In generation, updates until convergence, completes the initialization of balanced device and optical burst receiver.After the completion of optical burst receiver initialization, then
Enter the data is activation stage.Below the algorithm in two stages in the present invention is described in detail.
First, initial phase
Fig. 2 is initial phase parallel signal branch road algorithm schematic diagram.As shown in Fig. 2 compared with the algorithm of serial, this
The difference of bright parallel algorithm is, one group of 128 parallel symbol has corresponded to 128 balanced devices (eq) based on lms algorithm,
Each balanced device adopts identical tap coefficient.The renewal of tap coefficient is the initialized key of balanced device, and tap coefficient is more
Newly need using error signal it is therefore desirable to first obtain the error signal of every branch road.The training sequence that initial phase is used
Row should long enough, to ensure that initializing the equalizer tap coefficient obtaining can restrain.Initial phase is included in detail below
Step:
S101: send training sequence to optical burst receiver, through the training sequence signal of coherent demodulation and sample quantization
Carry out serial to parallel conversion and obtain n road parallel signal;The corresponding equalizer tap coefficient of the n-th=1 group of parallel signal of settingThis enforcement
Example Zhong Wei 128 tunnel.
S102: n-th group parallel signal enters n Parallel signal processing branch road, and each Parallel signal processing branch road includes all
Weighing apparatus and frequency deviation estimating modules, the balanced device of i-th branch road is equalized signalWherein k=(n-1) × n+i, 1
≤i≤n.
S103: frequency deviation estimating modules are according to known training symbolTo equalizing signalCarry out offset estimation, tired out
Long-pending phase errorWith offset estimation value
The frequency deviation estimating method adopting in present embodiment is the phase estimation method based on pre- judgement, and algorithm thinking is:
Equalizing signalPhasekIt is represented by:
Wherein, θkIt is the phase information that symbol carries,It is accumulated phase error.Phase errorIt is represented by:
Wherein, φ0,kCaused by laser phase noise, be slowly varying for high speed signal, so to one
It is regarded as constant, φ for group parallel symbolnIt is the phase fluctuation being caused by ase (spontaneous radiation) noise, k δ ωn,iT is then
Caused by frequency deviation.It can be seen that, the phase place of every road signal is removed symbol phase θkAfter be left withAccumulation to adjacent-symbol
Phase error carries out calculus of differences, obtains the offset estimation value of each branch roadCarry out average again
Computing obtains offset estimation mean valueφ just can be suppressed to a certain extentnImpact.Frequency deviation is for the letter of two-forty
It is slowly varying for number, so in the present invention, it is believed that its frequency deviation size is identical for one group of parallel symbol
's.The concrete steps of offset estimation include:
S3.1: calculate equalizing signalAccumulated phase errorWherein, φkRepresent equalizing signal
Phase place.
S3.2: calculate the offset estimation value of this branch roadWhereinThe kth of expression training sequence-
The accumulated phase error of 1 symbol.It is apparent that when in initial phase, the 1st branch road calculates offset estimation value, accumulated phase is by mistake
DifferenceInitial value
In present embodiment, as shown in Fig. 2 step s3.1 and step s3.2 are to adopt equalizing signalWith training sequence symbol
NumberConjugation (conj ()) be multiplied after obtainWithConjugation carry out multiplication and obtainRightAngle (arg ()) is taken to can get offset estimation value
S104: by the frequency offset estimation result of n branch roadAveragely obtained the offset estimation of n-th group parallel signal
Mean valueI.e.
S105:n branch road calculates its error signal respectivelyn,i:
Due to using known training sequence in initial phase, therefore do not need in error signal to use
Decision signal, but directly adopt known training symbol.
S106: the equalizer tap coefficient that (n+1)th group of parallel symbol of renewal uses:
Wherein,Represent (n+1)th group, the equalizer tap coefficient that used of n-th group parallel signal respectively.D table
Show the delay of error signal, that is, parallel symbol is input to the time that error signal feeds back to balanced device.Postpone due to existing, therefore
As 1≤n≤d, tap coefficient cannot be updated, tap coefficient is always using initial valueλ is the iteration of setting
Step-length, is positive number, its selection need sufficiently small to guarantee that iterative process can restrain.
The balanced device adopting in the present invention is the balanced device based on lms algorithm, when carrying out the renewal of tap coefficient is not
The error signal of single branch road, but the average using branch road error signal, that is,ncTable
Show that select from n branch road participates in the error signal quantity that tap coefficient calculates, 1≤nc≤ n, 1≤ic≤nc.Work as circuitry number
When measuring larger, all calculate a grouping error signal and can produce larger delay, therefore under conditions of disclosure satisfy that convergence, permissible
Reduce the error signal number participating in average computation, i.e. nc< n.RepresentCorresponding observation vector, that is, defeated
Enter the signal of balanced device,RepresentConjugation.
Whether reception processing finishes s107: training of judgement sequence signal, if untreated finish, return to step s102 continues
Processing next and organize parallel signal, if be disposed, entering the data is activation stage.
2nd, the data is activation stage
With the Main Differences of initial phase, the data is activation stage is that the phase place of data symbol is unknown, need by judgement
Recovered.Simultaneously because the phase place judging process of data symbol may malfunction it is therefore desirable to insert in the data symbol sending
The training symbol entering some obtains the reference phase that phase compensation needs.Fig. 3 is data is activation stage pipeline signal branch
Algorithm schematic diagram.As shown in figure 3, the data is activation stage comprises the following steps:
S201: in data sending terminal, insert training symbol in data symbol, its insertion method is: with n transmission symbol
For one group, then n transmission symbol is divided into r group, in n/r transmission symbol of every group, comprises a training symbol, r instruction
Practice sequence number in parallel symbol for the symbol and be designated as ir,1≤r≤r.
In the data is activation stage, because every group of n symbol after parallel needs to carry out pre- judgement to n bar branch road simultaneously, sentence
If error when certainly, the error signal of feedback is just very possible to malfunction, and can produce ill effect to the performance of system.So being
Obtain a relatively accurate phase compensation when making decisions, the present invention is in transmitting terminal sending signal in sending symbol
Insert a number of training symbol.The computational methods phase of the corresponding branch road of training symbol and every branch road of training sequence stage
With training signal branch road first calculates accurate accumulated phase and uses during several data-signal branch decisions as before and after this branch road
Reference phase to carry out accurate offset estimation.
Noise in order to be exaggerated when making offset estimation value in judgement reaches minimum, insertion symbol should be inserted in this little
The centre position of group, so that one group 128 send symbol as a example, the position of the training symbol at this moment inserting is: 128/ (r × 2)+x
× 128/r, x=0,1 ..., r-1.For example during 4 symbols of every group of insertion, 32 symbols of Ze Mei group are based on an identical and tire out
Long-pending phase error makes decisions, and therefore 4 training symbols is respectively inserted into i-th1=16, i2=48, i3=80, i4Article=112,
On branch road, so can reduce the error of phase compensation when judgement, maximum uses training signal branch road offset estimation value
16 times.
S202: send data-signal to optical burst receiver, the data-signal through coherent demodulation and sample quantization is entered
Row serial to parallel conversion obtains n road parallel signal, and n-th group parallel signal enters n Parallel signal processing branch road, the data is activation stage
Parallel signal processing branch road includes balanced device, frequency deviation estimating modules and judging module, and the equalizer processes of i-th branch road obtain all
Weighing apparatus signal
Parallel symbol group sequence number n in data is activation stage is to continue from the sequence number of last parallel symbol group of initial phase
Continuous arrangement, the equalizer tap coefficient that first group of parallel symbol of data is activation stage is carried out during equalizer processes is and initializes
The equalizer tap coefficient that stage finally obtains.
S203: offset estimation is carried out respectively to n bar branch road.The handling process of training signal data signal is otherwise varied.
When branch road is for training signal, x as shown in Figure 3n,16Road signal, obtains using with initial phase identical algorithm
To error signal it may be assumed that directly according to known training symbolTo equalizing signalCarry out offset estimation, obtain offset estimation valueAnd accumulated phase errorI.e.With
When branch road is for data-signal, x as shown in Figure 3n,15And xn,17Road signal, first to equalizing signalEnter line phase
Compensate, the signal after phase compensationFor:
Wherein, d represents the delay of offset estimation mean value,Expression rounds up.It can be seen that, data is activation stage n-th group
In parallel signal, the phase compensation of data-signal uses the n-th-d and organizes offset estimation mean value and its institute that parallel signal obtains
The accumulated phase error that in genus group, training signal obtains.
With xn,15As a example, due toSignal therefore after its phase compensation
For:
Judging module is to signalMake decisions and obtain decision signalAccording to decision signalTo equalizing signal
Carry out offset estimation, obtain offset estimation valueAnd accumulated phase errorIn as shown in Figure 3
With
S204: by the frequency offset estimation result of n branch roadAveragely obtained the offset estimation of n-th group parallel signal
Mean valueI.e.
S205:n branch road calculates its error signal respectivelyn,i.Likewise, the processing method of training signal data signal
Otherwise varied.
When branch road is training signal, i.e. i=irWhen, error signaln,iFor:
When branch road is for data-signal, error signaln,iFor:
S206: the equalizer tap coefficient that (n+1)th group of parallel symbol of renewal uses:
Wherein, λ*It is the iteration step length of data is activation stage setting,Represent that the data is activation stage selects from n branch road
Participate in tap coefficient calculate error signal quantity,If initial phase data
Error signal quantity n of average computation is participated in transmission phasecWithDiffer, then iteration step length λ and λ that they use*?
Need corresponding adjustment.
In the data is activation stage, because the run time of system alreadys exceed error signal delay d, therefore every time permissible
Realize the renewal of equalizer tap coefficient.
S207: judge whether data is disposed, if untreated finish, return to step s202 continues with next group simultaneously
Row signal, if be disposed, terminates.
Below simulating, verifying is carried out to the channel equalization based on lms for the present invention and offset estimation joint parallel method.Emulation
Middle parallel branch quantity is 256, and using standard single-mode fiber, Optical Fiber Transmission distance is about 50km, and balanced device uses 11 taps.
First to error signal quantity n in the renewal of initial phase equalizer tap coefficientcSize to systematic function
Impact is emulated.The computation delay Qu Liao10Ge clock-unit of equalizer error signal, when the delay that foe calculates also takes 10
Clock unit, 20 timers altogether.In initial phase, employ the training data of 12 frames, every 1024 symbols of frame, continue
Time is about 440ns.4 training symbols are inserted in data is activation stage, each group of parallel data.Other used in emulation
Parameter includes the frequency deviation of 1g, and OSNR (osnr) is fixed as 13db.Table 1 is different error letters during equalizer tap coefficient updates
The impact to systematic function for the number amount.
nc | The bit error rate | λ/nc |
32 | 1.7516×10-2 | 0.2/32 |
64 | 4.6241×10-4 | 0.2/64 |
128 | 4.2286×10-4 | 0.2/128 |
Table 1
As it can be seen from table 1 when OSNR is for 13db it is only necessary to 64 error signals are just obtained in that good system
System performance.
Then the impact to systematic function for training symbol quantity r every group of transmission symbol of data transmission phase inserted is carried out
Emulation, is emulated, the simulation parameter that other specification is used with table 1 in the case of osnr respectively 12db and 13db twice
Identical.Table 2 is every group of data is activation stage to send the impact to systematic function for the training symbol quantity inserted in symbol.
r | The bit error rate (osnr 12db) | The bit error rate (osnr 13db) |
2 | 3.2787×10-2 | 5.5445×10-4 |
4 | 3.1583×10-2 | 4.0015×10-4 |
8 | 3.1293×10-2 | 4.7803×10-4 |
Table 2
From Table 2, it can be seen that in the case of osnr respectively 12db and 13db, the number of insertion training symbol is 4
When can obtain performance good enough.In specific design burst mode optical receivers, can be according to different system design needs
Select different insertion training symbol numbers.
Fig. 4 is the equalizer convergence velocity contrast figure of parallel method of the present invention and serial approach.Used in this emulation
Osnr is 13db, and mse represents mean square error (mean squared error).As shown in figure 4, the parallelization to serial algorithm,
Although the convergence rate of balanced device can be reduced to a certain extent, the performance after restraining can't be affected.
Fig. 5 is to the convergence rate comparison diagram having computing relay and no computing relay in present invention joint parallel method.Should
Used in emulation, total delay size is 20 clock-units.As shown in figure 5, when there being computing relay, the convergence of balanced device is only
Be as iterate to calculate delay and phase delay, have no effect on the performance after iteration convergence.And if use in emulation
Other different delay sizes, equally can obtain similar effect it can be seen that the present invention has for the size of computing relay
Tolerance well.
Fig. 6 is the bit error rate comparison diagram of parallel method under serial approach and different delays of the present invention.As long as shown in fig. 6,
Optical burst receiver is correct initialize after, computing relay size in the present invention bit error rate (ber) to optical burst receiver
Performance does not affect.Meanwhile, compared with the performance of preferable serial approach, the performance loss that brought due to parallelization in the present invention
Also only has general 0.2db.
Although to the present invention, illustrative specific embodiment is described above, in order to the technology of the art
Personnel understand the present invention, the common skill it should be apparent that the invention is not restricted to the scope of specific embodiment, to the art
For art personnel, as long as various change is in the spirit and scope of the present invention of appended claim restriction and determination, these
Change is it will be apparent that all utilize the innovation and creation of present inventive concept all in the row of protection.
Claims (3)
1. a kind of channel equalization based on lms and offset estimation joint parallel method are it is characterised in that include lower step:
S1: initialized using training sequence, including step:
S1.1: send training sequence to optical burst receiver, carry out through the training sequence signal of coherent demodulation and sample quantization
Serial to parallel conversion obtains n road parallel signal;The corresponding equalizer tap coefficient of the n-th=1 group of parallel signal of setting
S1.2: n-th group parallel signal enters n Parallel signal processing branch road, and each Parallel signal processing branch road includes balanced device
And frequency deviation estimating modules, the balanced device of i-th branch road is equalized signalWherein k=(n-1) × n+i, 1≤i≤
n;
S1.3: frequency deviation estimating modules are according to known training symbolTo equalizing signalCarry out offset estimation, obtain accumulated phase
ErrorWith offset estimation value
S1.4: by the offset estimation value of n branch roadAveragely obtained the offset estimation mean value of n-th group parallel signal
S1.5:n branch road calculates its error signal respectivelyn,i:
S1.6: the equalizer tap coefficient that (n+1)th group of parallel signal of renewal uses:
Wherein,Represent (n+1)th group, the equalizer tap coefficient that used of n-th group parallel signal respectively;D represents by mistake
The delay of difference signal;λ is the iteration step length of setting, is positive number;ncRepresent that the participation tap coefficient selecting from n branch road calculates
Error signal quantity, 1≤nc≤ n, 1≤ic≤nc;RepresentCorresponding observation vector,
RepresentConjugation;
S1.7: whether training of judgement sequence is disposed, if untreated finish, return to step s1.2 continues with next group simultaneously
Row signal, if be disposed, enters step s2;
S2: the entrance data is activation stage is processed to data, including step:
S2.1: data sending terminal inserts training symbol in data symbol, and its insertion method is: send symbol for one group with n,
Again n transmission symbol is divided into r group, in n/r transmission symbol of every group, comprises a training symbol, r training symbol
Sequence number in parallel symbol is designated as ir,1≤r≤r;
S2.2: send data-signal to optical burst receiver, the data-signal through coherent demodulation and sample quantization is gone here and there
And convert and obtain n road parallel signal, n-th group parallel signal enters n Parallel signal processing branch road, the data is activation stage parallel
Signal transacting branch road includes balanced device, frequency deviation estimating modules and judging module, and the equalizer processes of i-th branch road are equalized letter
Number
S2.3: n bar branch road is carried out respectively with offset estimation:
When branch road is for training signal, directly according to known training symbolTo equalizing signalCarry out offset estimation, tired out
Long-pending phase errorWith offset estimation value
When branch road is for data-signal, first to equalizing signalCarry out phase compensation, the signal after phase compensationFor:
Wherein, d represents the delay of offset estimation mean value,Expression rounds up;Judging module is to signalMake decisions
To decision signalAccording to decision signalTo equalizing signalCarry out offset estimation, obtain offset estimation value
And accumulated phase error
S2.4: by the offset estimation value of n branch roadAveragely obtained the offset estimation mean value of n-th group parallel signal
S2.5:n branch road calculates its error signal respectivelyn,i:
When branch road is training signal, i.e. i=irWhen, error signaln,iFor:
When branch road is for data-signal, error signaln,iFor:
S2.6: the equalizer tap coefficient that (n+1)th group of parallel symbol of renewal uses:
Wherein, λ*It is the iteration step length of data is activation stage setting,The ginseng that the expression data is activation stage selects from n branch road
The error signal quantity calculating with tap coefficient,
S2.7: judge whether data-signal is disposed, if untreated finish, return to step s2.2 continues with, if place
Reason finishes, and terminates.
2. according to claim 1 joint parallel method it is characterised in that the concrete grammar of described offset estimation include with
Lower step:
S3.1: calculate equalizing signalAccumulated phase error:
When branch road is for training symbol, accumulated phase errorWherein θkThe phase place of training symbol known to expression,
φkRepresent equalizing signalPhase place;
When branch road is for data-signal, accumulated phase errorWhereinRepresent data decision signalPhase
Position;
S3.2: calculate the offset estimation value of this branch roadWhereinRepresent the accumulation phase of -1 signal of kth
Position error.
3. parallel method according to claim 1 it is characterised in that in described step s2.1 training symbol insertion position
For n/ (r × 2)+x × n/r, x=0,1 ..., r-1.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410007137.XA CN103780519B (en) | 2014-01-07 | 2014-01-07 | Channel equalization and frequency deviation estimation joint parallel method based on LMS |
PCT/CN2014/072120 WO2015103804A1 (en) | 2014-01-07 | 2014-02-16 | Channel equalization and frequency offset estimation joint parallel method based on lms |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410007137.XA CN103780519B (en) | 2014-01-07 | 2014-01-07 | Channel equalization and frequency deviation estimation joint parallel method based on LMS |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103780519A CN103780519A (en) | 2014-05-07 |
CN103780519B true CN103780519B (en) | 2017-01-18 |
Family
ID=50572363
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410007137.XA Expired - Fee Related CN103780519B (en) | 2014-01-07 | 2014-01-07 | Channel equalization and frequency deviation estimation joint parallel method based on LMS |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN103780519B (en) |
WO (1) | WO2015103804A1 (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11848802B2 (en) * | 2021-03-04 | 2023-12-19 | Realtek Semiconductor Corporation | Received data equalization apparatus and method |
Families Citing this family (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107925483B (en) | 2015-08-27 | 2019-12-17 | 华为技术有限公司 | optical signal processing method and coherent receiver |
CN105610507B (en) * | 2015-12-24 | 2018-02-16 | 武汉邮电科学研究院 | A kind of parallel constant modulus algorithm suitable for coherent optical communication system simplifies implementation method |
CN105653806A (en) * | 2015-12-30 | 2016-06-08 | 芜湖乐锐思信息咨询有限公司 | Complex product simulation system based on internet |
CN105635008B (en) * | 2016-04-01 | 2018-11-20 | 电子科技大学 | A kind of adaptive equilibrium method, device and adaptive equalizer |
CN106209100B (en) * | 2016-08-04 | 2019-06-07 | 成都博思微科技有限公司 | A kind of LMS error correcting system and method applied to high-speed ADC |
CN106789793B (en) * | 2016-12-28 | 2019-07-16 | 武汉邮电科学研究院 | The LMS method without frequency difference difference feedback updated for adaptive equalizer coefficients |
CN109450563B (en) * | 2018-10-16 | 2021-04-30 | 电子科技大学 | WiFi chip phase offset correction method |
CN109510668B (en) * | 2018-12-11 | 2020-06-30 | 武汉邮电科学研究院有限公司 | Adaptive equalizer based on quasi feedforward control in coherent optical communication and method |
CN109547369B (en) * | 2018-12-26 | 2021-08-03 | 西安烽火电子科技有限责任公司 | Self-adaptive channel parameter estimation method |
CN111245499B (en) * | 2020-01-08 | 2021-07-27 | 西安电子科技大学 | Pre-shaping-based time domain parallel fractional interval equalizer and equalization method |
CN114039623B (en) * | 2021-10-22 | 2023-05-19 | 中国电子科技集团公司第五十四研究所 | Short burst spread spectrum signal tracking method with low carrier-to-noise ratio |
CN116132236A (en) * | 2022-12-15 | 2023-05-16 | 西安电子科技大学 | Single-symbol self-adaptive frequency offset estimation and compensation method applied to 5G NR system |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103095614A (en) * | 2013-01-24 | 2013-05-08 | 电子科技大学 | Joint equalization and frequency offset estimation device in proruption coherent optical fiber communications |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7016405B2 (en) * | 2002-02-14 | 2006-03-21 | Mitsubishi Electric Research Laboratories, Inc. | Signal circulation for adaptive signal equalization in digital communication systems |
CN102422571B (en) * | 2009-05-18 | 2016-06-15 | 日本电信电话株式会社 | Signal generating circuit, optical signal transmitter, signal receiving circuit, optical signal synchronization establishment method and optical signal synchronization system |
-
2014
- 2014-01-07 CN CN201410007137.XA patent/CN103780519B/en not_active Expired - Fee Related
- 2014-02-16 WO PCT/CN2014/072120 patent/WO2015103804A1/en active Application Filing
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103095614A (en) * | 2013-01-24 | 2013-05-08 | 电子科技大学 | Joint equalization and frequency offset estimation device in proruption coherent optical fiber communications |
Non-Patent Citations (1)
Title |
---|
一种适用于突发相干光通信的均衡和频偏估计的联合方法;果宏艳等;《光电子激光》;20130930;第24卷(第9期);第1714-1719页 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11848802B2 (en) * | 2021-03-04 | 2023-12-19 | Realtek Semiconductor Corporation | Received data equalization apparatus and method |
Also Published As
Publication number | Publication date |
---|---|
CN103780519A (en) | 2014-05-07 |
WO2015103804A1 (en) | 2015-07-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103780519B (en) | Channel equalization and frequency deviation estimation joint parallel method based on LMS | |
CN101965698B (en) | Apparatus and method for equalizing chromatic dispersion and digital coherent optical receiver | |
EP2997676B1 (en) | Method, device and communication system for reducing optical transmission impairments | |
CN110971295B (en) | Optical fiber nonlinear noise monitoring device and optical receiver | |
CN114793305A (en) | Method, apparatus, device and medium for optical communication | |
CN103812806A (en) | Channel equalization method and system based on time domain channel estimation | |
CN108123908A (en) | A kind of improvement SVM equalization methods and system for NG-PON | |
Choutagunta et al. | Efficient quantification and simulation of modal dynamics in multimode fiber links | |
CN106209222A (en) | A kind of method and device of optical signal-to-noise ratio monitoring | |
CN106656881A (en) | Adaptive blind equalization method based on deviation compensation | |
WO2011017852A1 (en) | Simulation device and simulation method | |
US20220131619A1 (en) | Optical receiving apparatus and coefficient optimization method | |
Fan et al. | A high performance nonlinear compensation algorithm with reduced complexity based on XPM model | |
CN105591690A (en) | Method and device for monitoring polarization-mode dispersion | |
Bosco et al. | A novel update algorithm in stokes space for adaptive equalization in coherent receivers | |
Grellier et al. | Are multilevel pseudorandom sequences really needed to emulate highly dispersive optical transmission systems? | |
CN110380789A (en) | A kind of signal processing method and device | |
CN107005307B (en) | A kind of method and balancer that balancer is set | |
Do et al. | Chromatic dispersion estimation based on complementary Golay sequences for 80 Gb/s QPSK single-carrier system with frequency domain equalization | |
CN114866145B (en) | Method, apparatus, device and computer readable medium for optical communication | |
CN104144139A (en) | Improved time domain self-adaptation blind equalization method | |
Guo et al. | Optical Signal-to-Noise Ratio Monitoring Based On Back-Propagation Network With an Autoencoder | |
CN103647632B (en) | Method and device for compensating for frequency differences of data signals | |
He et al. | Low complexity single-stage adaptive frequency domain equalizer for SDM systems using few mode fibers | |
CN115378507B (en) | Noise equalization method and device suitable for multi-eigenvalue nonlinear frequency division multiplexing system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170118 Termination date: 20200107 |