Invention content
According to one embodiment, a kind of method of the unknown errors for estimating carrier frequency, this method packet are provided
Include following steps:The signal of modulated version as the carrier frequency is sampled to form multiple data blocks, each number
According to block include multiple samples and each sample incorporates the phase error caused by the carrier frequency error;For each
A data block:Generate the product of each sample and the data another sample in the block, sample described in another described sample picture
In the data block equally be located at the data block reverse version in same position in, with formed respectively incorporate due to
The combined sample of substantially the same phase error caused by the carrier frequency error;And calculate the combined sample
Average value is to form average sample;And estimate the carrier wave by the multiple average samples generated from the multiple data block
Frequency error.
The method may include by sum the combined sample and divided by data sample in the block quantity
To calculate the average value of the combined sample.
The signal can be the version for the carrier frequency modulated by extension, the method may include
The signal is sampled using the sampling rate equal to the spreading rate for extending the signal.
The signal can be the version for the carrier frequency modulated by extension, the method may include
The data block is formed as to include respectively many samples equal to spreading factor.
The method may include use maximum likelihood estimator module to estimate the carrier frequency error.
The method may include the carrier frequency error is estimated by following steps:To the multiple average sample
Execute Fourier transformation;And select have the frequency of peak as the carrier frequency error in the transformation.
The method may include for mutiple antennas:For each signal received by the multiple antenna, generate
Multiple average samples;To every group of average sample, Fourier transformation is executed;And by quadratic sum is summed combination obtained it is more
A transformation.
Estimated carrier frequency error can incorporate fuzzy factor, the method may include:Fine estimation stages,
It is fine valuation and the rough estimate stage that the carrier frequency error is obtained in the fine estimation stages, rough at this
The fuzzy factor in the fine valuation is parsed in estimation stages.
The fine estimation stages may include according to the method for claim 1 and by using the first sampling frequency
Rate generates fine valuation to generate the data block;And the rough estimate stage may include by according to claim
Method described in 1 parses the mould to generate rough valuation and generate the data block by using the second sample frequency
The factor is pasted, second sample frequency is less than first sample frequency.
The rough estimate stage may include using from the technology used by the fine estimation stages is different is used for
The technology of the carrier frequency error is estimated to generate the rough valuation.
The method may include generate the carrier frequency by following steps for the rough estimate stage to miss
The rough valuation of difference:Make the multiple average sample and two or more frequency compoundings, each frequency in the frequency
It is the multiple of the fuzzy factor;Determine the signal power that each signal in mixed signal includes;And selection life
Rough valuation at the frequency for the mixed signal for including highest signal power as the carrier frequency error.
The method may include ask the rough valuation of estimated carrier frequency error and the carrier frequency error
Together, to form the complete valuation for the carrier frequency error for wherein having parsed the fuzzy factor.
The signal can be the version for the carrier frequency for having used phase-shift keying (PSK) to modulate.
According to second embodiment, a kind of device is provided, which is configured to by realizing such as claim 1 to 12
Any one of described in method carry out the unknown errors of estimating carrier frequency.
According to third, embodiment there is provided a kind of device of the unknown errors for estimating carrier frequency, the device packets
It includes:Sampling unit, the sampling unit are configured to sample with shape the signal of the modulated version as the carrier frequency
At multiple data blocks, each data block includes multiple samples and each sample is incorporated due to the carrier frequency error and led
The phase error of cause;Product unit, the product unit are configured to, for each data block, generate each sample and the data
The product of another sample in the block, another sample sample as described in are located at the data block in the data block
The same position reversed in version in, respectively incorporate caused by the carrier frequency error substantially phase to be formed
The combined sample of same phase error, and the average value of the combined sample is calculated to form average sample;And tone is estimated
Gauge, the pitch estimator are configured to estimate using the multiple average samples generated from the multiple data block described
Carrier frequency error.
Specific implementation mode
The problem of in order to avoid current techniques for estimating carrier frequency error, it is proposed that avoid for judgement, known
Leading or pilot tone needs and be the estimator for being relatively easy to tone (tone) test problems by problem reduction.
In this approach, it is sampled by the signal of the modulated version to being used as carrier frequency to form multiple data
Block carrys out the unknown errors of estimating carrier frequency.Each data block includes multiple samples, and each sample is incorporated due to carrier wave
Phase error caused by frequency error.
Then each number is handled by generating the product of data each sample in the block and same another sample in the block
According to block.Decent of another sample is equally located at the same position in reversing (reversed) version of the data block within the data block
In setting.The result is that respectively incorporate the combined sample of substantially the same phase error caused by carrier frequency error
Vector.Then calculate combined sample average value, and by across multiple data blocks execute these steps obtained it is multiple
Average sample is used for estimating carrier frequency error.
In one example, using phase-shift keying (PSK), specifically binary phase shift keying (BPSK) is modulated data.This
Technology described in text may be in the communication system using a kind of modulation that wherein data are encoded by means of phase-modulation
Particularly advantageous;It is to rely on the phase of received signal due to carrier frequency error and by particular characteristic when changing
The modulation technique of the phase of deterioration.However, this is an example, and it should be understood that other modulator approaches can be used.
Techniques described herein is also possible to be particularly advantageous in the communication system using extension.Due to carrier frequency
It may be especially severe in the system using extension that the performance deterioration that phase place generates is harmful to caused by error, because
Signal is no longer coherently summed.This shortcoming can be solved by techniques described herein, because they to be harmful to
Phase place is removed after having had estimated carrier frequency error.Therefore, although being retouched to estimation technique with reference to extension signal
It states, it should be appreciated that, the advantages of this will only illustrate these technologies.
The example of the method for estimating carrier frequency error is shown in FIG. 1.This method is by will be received
Signal separator starts (step 101) at data block.Each data block includes multiple samples.Data block can be same size or
Different sizes.In one example, data block includes respectively being equal to spreading factor NsMany samples.It can also use other
The data block (such as apparent from the coarse frequency estimator being described below) of size.
Signal has been extended in this illustration, and so received chip is multiplied by corresponding expanding value sm(step
102).Then calculate by data block indicate vector with itself reversing version by element product (step 103).As a result
Each element of obtained vector incorporates substantially the same harmful phase place caused by carrier frequency error.Cause
This, calculates the average value (step 104) of product, these elements are mainly coherently summed when being averaged.In L data symbol
For this sequence of each data block in measurement period, judge whether to have handled each data block (step 105).It can be serially
Execute step so that each data block is handled in turn, or concurrently can efficiently be handled.L is not necessarily data sequence
Length, and the tradeoff that may be chosen between accuracy and error changeability.As a result the vector of the length L obtained is frequency
Rate 2feAliasing tone and noise sum.Final step is to estimate the pitch frequency being present in the vector to obtain carrier frequency
Valuation (the step 106) of rate error.
Due to aliasing, estimated carrier frequency error, which has, to be equal toMultiple fuzzy factor (ambiguity).
Coarse frequency estimation stages can be parsed by executing process similar with process shown in FIG. 1 after fine frequency correction
This fuzzy factor.This is described in more detail below.
It is more fully described shown in Fig. 1 now with reference to binary phase shift keying (BPSK) and the signal of extension modulation is used
Method.This purpose being given for example only.As described above, the present invention is limited to the modulation or extension of particular form.
Binary data bmBy certain binary sequence SkIt extends and is then mapped to BPSK constellation points dk∈{-1,+1}。
Data can also distinctively be encoded if necessary.Then by certain pulse (for example, root raised cosine) p (t) to transmitted
Signal is filtered.Therefore transmitted signal is
Wherein TcIt is chip delay (i.e. three dB bandwidth BwInverse) and NsIt is spreading factor.
Received signal is undergone caused by the frequency error of crystal error, channel effect and transmitted signal
Frequency error.This frequency error is written as feAnd it can change with the time.Explanation assumes that frequency error exists below
For constant on the burst (burst) of evaluated error.However, the principle of the present invention can also be applied to meet with the letter of variation
The system of road condition, wherein feIt can change and (this is described in more detail below).
Received signal can be written as:
It is spreading rate that signal, which is matched and filters and take 1/10th,.Discrete signal can be written as:
This assume receiver due in the transmission of burst or the timing from some pretreatment stage alignment and
Knowledge with timing.
Formula 3, which is ignored, to be used for considering the item that Nyquist (Nyquist) pulse structure is destroyed by frequency error.This will cause to accord with
It is interfered between number, but only under very high frequency(VHF) error.Therefore it can ignore it in analysis.
Processing of the description for signaling channel symbol k first.The received vector of this channel symbol is written as:
Wherein φkIt is the initial phase of k-th of symbol.
It extends through and is multiplied by escape character smAnd it is removed.Because these can be ± 1, result can be obtained
Sequence is received to be written as:
Then version is reversed to this sequence and timeIt executes and presses element multiplication:
BPSK is modulated, for all kThis is provided:
As shown in Equation 7, by each sample, the corresponding quantity in the reversing version of data block is multiplied with it, and generation is whole to be had
There is the combined sample of same phase.(in fact, these phase components may not be definitely identical;It may across data block
In the presence of the relatively small variation caused by vibration frequency error and channel).Then the average value of combined sample is calculated.It can
Any type of average, such as average, intermediate value, pattern can be used.Preferred option will calculate average:
Because combined sample has basically the same phase, when calculating average, they are mainly coherently closed
Meter.
Phase presses 2 π f between continuous symbole(Ns- 1) increase.Therefore, φk+1=φk+2πfe(Ns-1).This is implied:
Substitute into θ0It is provided as intialization phase:
Therefore average sample forms following sequence, wherein carrier frequency error draws from a sample to next sample
Play phase change.From formula 10 it is clear that therefore the vectorial z of data is pitch frequency 2fe。
Present be pitch frequency estimation problem by problem reduction.In the presence of the multiple sides for the frequency that can be used for estimating the tone
Method, and any tone algorithm for estimating can be applied to vectorial z.Preferably, maximum likelihood estimator module is used for estimating tone.Realize this
A practical way can be via Fourier transformation.For example, L points FFT can be executed to vectorial z.It identifies with maximum absolute
The piecemeal (bin) of value.Frequency corresponding with the piecemeal is confirmed as pitch frequency 2fe.This close to maximum likelihood estimator module,
In sole difference between this maximum likelihood estimator module and true maximum likelihood estimator module be piecemeal discrete nature.
It in the case of multiple reception antennas, can will be for the result quadratic sum of the FFT of each in reception antenna
Add up to combine these results.This example that how may be combined only from the result of multiple reception antennas, and
It should be understood that the present invention covers the use of any suitable method.
It also may be assumed that the carrier frequency error of variation.Frequency error is preferably constant in the length of escape character
's.Maximum authorized frequency error deviation with average isAverage can be rangeInterior any value.This paper institutes
The method of description spreads the length estimate average of escape character burst.After being corrected to this, any deviation frequency
Rate error should be it is small and can by known tracking scheme (such as by using Costas loop Phase Tracking) come with
Track.
Frequency acquisition range depends on sample frequency.In example described above, it means that frequency acquisition range
Dependent on spreading factor Ns.This is introduced and Nyquist (or folding) frequency alias for sample frequency half.By being retouched above
The carrier frequency error valuation for the output of process stated is in rangeIt is interior.Two extraneous factor on denominator is because in vector z
Tone have frequency 2fe.Especially for larger spreading factor, frequency acquisition range may be restrictive.
Limited capture range means that estimated error may include fuzzy factor.This can generate and be made to go
Except the correction of harmful phase place in the reception data of mistake.Therefore coarse frequency estimation stages can be applied.This is shown
In fig. 2, coarse stage (step 202) is applied after elaboration stage (step 201) in fig. 2.By the rough estimate stage
The frequency ambiguity factor of output is added to fine valuation to generate the complete valuation (step 203) of carrier frequency error.
Fuzzy factor can only take particular value.The range of these values is for l=-1,0 ,+1 etc.Effective l values
Quantity is the design alternative of the desired extent based on frequency error to be processed.For illustration purposes only, below in reference to l
=-1,0, method of+1 description for parsing this fuzzy factor.
The basic principle of coarse stage is to create new tone using the same technology summarized for fine frequency estimation
Signal.This is exemplified in figure 3.Data application fine frequency error is compensated to remove the fine valuation (step of error first
301).Coarse frequency estimates the data block most likely with the different length estimated from fine frequency to obtain larger frequency
Capture range.Therefore step 302 and 303 is substantially the repetition of step 101 and 102 but is directed to different sampling rates.?
In one example, the N of each channel symbolsA chip is divided into a pair of sample.Rough estimate then proceed to it is identical in Fig. 1
Processing step (step 304 to 306) but for different sample frequencys to obtain new tone signalIt is accorded in each channel
Number be divided into the example of a pair of sample, this substantially generate repeats the above processing arranged in formula 4 to formula 10 but wherein
Ns=2.Tone signalMore have a noise than z, but we had it is anticipated that tone where (that is, for l=-1,0 ,+1
?Near) knowledge.If there are small errors in estimated fine frequency error, it is less likely definitely
Locate in any of those frequencies.
Coarse frequency estimation stages deviate fine frequency estimation stages when it starts to estimate tone.Tone signalVery may be used
It can be noisy, but only need identifying which of effective l values are to identify tone in degree necessary to right value
Frequency.Therefore, different estimation techniques is likely to be applied to the rough estimate stage from fine estimation stages.In the example of Fig. 3
In, pass through first for selected l valuesIt is mixed with(step 307).As a result then the signal obtained passes through low pass filtered
Wave device (step 308).This filter can be set based on the expection residual frequency error of channel and Doppler (Doppler) extension
The bandwidth of wave device.For example, simple moving average filter can be used.Calculate the power for including in the signal that result obtains
To obtain performance number PI(step 309).Step 306 is repeated to step 308 (step 310) for the l values of each selection.Finally,
Maximum power value P is provided by selectingIValue l values estimate coarse frequency(step 311).Other selections can be used
Method.However, this method both realized it is simple and close to maximum likelihood estimator module.
If there is multiple reception antennas, then it can add up to corresponding P from each reception antenna by being directed to given l valuesI
Value generates the corresponding P of each l valuesIValue.
A part for the receiver of wireless communication system is likely to be implemented as realizing the device of the invention.It is such
Receiver may generally include for realizing the RF circuits well known to wireless communication, including such as antenna system, RF transceivers, one
A or more amplifier, tuner, one or more oscillators, digital signal processor, memory etc..
The example of device is shown in FIG. 4.Device usually shown in 401 is configured to receive signal r (402).
Although " receiving " signal can be directly from the signal of antenna, it is also possible to before being input into device
It is pre-processed by certain.
Device includes the first sampling unit 402 for being used to form data block.Device further include be used to form average value to
Measure the product unit 403 of z.These are drawn as with the step I in the step 101 in Fig. 1 to the corresponding Fig. 4 of step 105 simultaneously
With.Product unit 403 is followed by pitch detector 404, is configured to calculate fine frequency errorThis is as in Fig. 4
Step II is cited and corresponding to the step 106 in Fig. 1.Fine frequency error is output to summation unit 405.
Device can also be configured to execute coarse frequency estimation, but if frequency acquisition range be considered as it is enough from
And coarse frequency be estimated as it is unnecessary, then can disable this functionality.
Device includes being configured to make received signal and estimated fine frequency errorMixed mixer
406.Device further includes the second sampling unit and the second product unit (407 and 408), be configured to by average value it is another to
Amount is output to one group of parallel processing path to estimate the fuzzy factor in fine valuation.Each path includes fuzzy hybrid unit
409.These are connected to low-pass filter 410 and power calculation unit 411.Final unit includes maximum value recognition unit, this is most
Big value recognition unit be configured to receive the performance number exported by power calculation unit and by with the maximum in those performance numbers
It is coarse frequency valuation to be worth associated frequency identification.Maximum value recognition unit is configured to coarse frequency valuation being output to and ask
With unit 405.
It similarly may gradually realize parallel processing path illustrated by Fig. 4.Similarly, it should be appreciated that other components can
Some that can be performed in parallel in their all processing by product unit 403, the second product unit 408 (for example, crossed over more
The product that a data block executes calculates).
Structure shown in Fig. 4 is intended to correspond to many functional blocks.This is only used for illustrative purpose.Fig. 4 is not intended to be limiting core
The stringent division between distinct program, process or function between the different piece of the hardware of on piece or in software.In some realities
It applies in mode, some or all of algorithm described herein can be executed entirely or partly with hardware.In many
In embodiment, at least part of communication device can be realized by the processor acted under software control.It is any such
Software is stored preferably in non-transitory computer-readable medium, memory (RAM, cache, hard disk etc.) or its
Its storage device (USB sticks, CD, disk etc.).
Approach described herein even also permits in very low noise environment especially for the BPSK systems using extension
Perhaps accurate Carrier frequency offset estimation.These methods are not only suitable for the system of experience flat fading and are equally applicable to multipath letter
Road.Frequency acquisition range also increases on current algorithm for estimating, to close to chip frequency rather than symbol frequency.These methods
It can also be applied to mutiple antennas system.
Method described herein can be applied to the communication network for Internet of Things (IoT) communications fabric.Example will
Including being configured to according to WeightlessTMNetwork that agreement is operated (but approach described herein can be by constructing
Network to be operated according to different agreements (such as LTE, bluetooth, WiFi, VoIP) is easily realized).In general, network will
Include many communication equipments (such as base station) for being respectively configured to be communicated with a large amount of terminals being geographically separated from each other.This
Communication device described in text can be realized only by such communication equipment or terminal.Network can be cellular network,
In each communication equipment be responsible for the air communication with the terminal in the respective cell.Communication equipment is via wireline interface or wirelessly
Interface is suitably communicated with core net, and can be acted under the control of core net at least partly.
In one example, communication as described herein device may be constructed such that according to WeightlessTMIoT specifications
It is operated.WeightlessTMUsing honeycomb WAN frameworks, wherein agreement is directed to requirement (low terminal cost, the low end of IoT systems
Hold work period and therefore low-power consumption, and the scalability to very low data rate) it is optimised.It is originally designed into
It is operated in TV blank frequency spectrums from 470MHz to 790MHz, but PHY is extended in the licensed band for changing bandwidth, shares and permitted
It can access frequency band and exempt and operated in licensed band.
The applicant therefore disclose in isolation each independent feature described herein and two or more as
Any combinations of feature, such feature or combination can be generally in view of known in those skilled in the art in this sense
Knowledge is performed based on this specification, and whether the combination but regardless of such feature or these features solves disclosed herein
What problem, and it is not limited to the scope of the claims.Applicant indicate that the aspect of the present invention may include any such single
The combination of uniqueness sign or these features.In view of the description of front, will be apparent that those skilled in the art, Ke Yi
Various modifications are made in the scope of the present invention.