CN101471727B - Digitalization automatic frequency detection method based on true signal - Google Patents

Abstract

The invention discloses a digital automatic frequency detection method based on real signals, which relates to the technical field of frequency synchronization of a transiting side and a receiving side in communication. The method comprises the following steps: sampling a channel of real received signals; calculating the detection threshold and the measurement of the frequency reference signal; and determining whether the frequency reference signal exits in the received signals according to the calculated detection threshold and the measurement, wherein the frequency reference signal is utilized to maintain the frequency synchronization between the transiting side and the receiving side during the frequency correction process. Compared with the prior frequency detection method based on complex signals, the digital automatic frequency detection method doesn't need to acquire the mutually-orthogonal complex received signals of the channel I and the channel Q; overcomes the problem of imbalance compensation of the channel I and the channel Q; and greatly reduces the complexity of system implementation on the premise of ensuring the frequency detection performance.

Description

Digitlization automatic frequency detection method based on real signal

[technical field]

The present invention relates to the Frequency Synchronization system in the field of communication transmission, relate in particular to a kind of digitlization automatic frequency detection method based on real signal.

[background technology]

In telecommunication transmission system, the frequency source signal frequency of receiving terminal need to be consistent with the frequency source signal frequency of transmitting terminal usually.In order to accomplish this point, a kind of method that can adopt is to adopt high-precision crystal oscillator at transmitting-receiving two-end.But high-precision crystal oscillator is usually relatively more expensive, and meeting occurrence frequency drift after long-time the use, need to regularly carry out manual frequency and proofread and correct.

The another kind of method that can adopt is to send a frequency reference signal (being called again calibrating frequency signal or pilot signal) by transmitting terminal to receiving terminal, receiving terminal carries out automatic calibration according to the frequency reference signal that receives to the frequency of local frequency source signal, so that the frequency of local frequency source signal is followed the tracks of the frequency in frequency reference signal in the scope that precision allows, thereby reach the purpose of receiving-transmitting sides Frequency Synchronization.This method is called automatic frequency tracking or automatic frequency correction.

Sometimes frequency reference signal is not to exist always, but intermittently send.Must detect first to receive in the signal whether have frequency reference signal this moment before implementing automatic frequency correction, only under the prerequisite that frequency reference signal exists, just can implement automatic frequency correction, and this method that whether the detection frequency reference signal exists in receiving signal is called the automatic frequency detection.

The exemplary that automatic frequency detects is the frequency correction between base station and travelling carriage or the repeater in the gsm mobile communication system.The base station is in the downlink broadcast channel, send the calibrating frequency burst packets (FCB) that a duration is about 577s every about 46ms or 51ms, be frequency reference signal, these FCB are detected in travelling carriage or repeater in receiving signal, and proofread and correct the frequency source frequency of oneself according to these FCB.

Automatic frequency detection method is divided into analog and digital two kinds.Analog automatic frequency detection method obtains thresholding and tolerance for frequency detecting by the processing to analog signal, digital automatic frequency detection method is then first with analog signal figure, and then by digital signal being processed obtain thresholding and the tolerance for frequency detecting.

Common digital automatic frequency detection method is to implement according to the plural form (being I, the mutually orthogonal real signal of Q two-way) that receives signal, for example United States Patent (USP) bulletin US 6226336B1 and US2005/0129149A1.This handled sampled data of digital automatic frequency detection method based on the complex frequency reference signal is plural number, and its real part and imaginary part are respectively the real number sample values that I, the mutually orthogonal real signal sampling of Q two-way is obtained at synchronization.

Yet the digital automatic frequency detection method based on complex signal has higher complexity.The I, the Q two paths of signals that obtain plural form need to carry out the quadrature frequency conversion to be realized by road signal to real number form, and this is more more complex than the road signal that obtains real number form; In addition, it is unbalance to introduce again some in the process that obtains I, Q two paths of signals, comprise amplitude imbalance (I, Q two-way amplitude different), phase imbalance (I, Q two-way phase place non-orthogonal) etc., these are unbalance can to affect the performance of automatic frequency detection method.Need to compensate unbalance in order addressing this problem, but to do like this and can make the complexity increase that system realizes.

[summary of the invention]

Purpose of the present invention is exactly for overcoming the above-mentioned higher shortcoming of digital automatic frequency detection method complexity based on complex signal, a kind of digitlization automatic frequency detection method based on real signal being provided.

For this reason, the present invention takes following technical scheme to realize:

The present invention's the digitlization automatic frequency detection method based on real signal, the road simulation real signal that transmitting terminal transmission is come by receiving terminal carries out digitlization and detects, being confirmed whether to exist frequency reference signal, comprising:

With described one road intermediate frequency or baseband analog real signal sample, analog-to-digital conversion is the step of digital signal;

Calculate the step of decision threshold benchmark with the sampled data of this real signal;

Calculate step for detection of the detection metric of frequency reference signal with the sampled data of this real signal;

With described detection metric be multiply by the step that formed decision threshold compares behind the default coefficient by described decision threshold benchmark, if surpass this decision threshold, then produce the marking signal that the described frequency reference signal of sign exists.

In described sampling, the analog-to-digital step, sampling clock is obtained through frequency division/frequency multiplication by local frequency source signal, according to transmitting-receiving frequency difference allowed bands different in the practical application, sample frequency is got different preset values from the ratio of local frequency source signal frequency, and sample frequency changes along with the adjustment of local frequency source.

In the calculation procedure of described decision threshold benchmark, the decision threshold benchmark is that the quadratic sum by two neighbouring samples real sampled data constantly obtains after smoothly.

In the calculation procedure of described decision threshold benchmark, more specifically, described two neighbouring samples real sampled data constantly, each is through the multiplier involution first, and the result carries out smoothing processing again through the adder addition, obtains described decision threshold benchmark.

In the calculation procedure of described detection metric, detection metric is obtained after level and smooth by 2 times of the difference of the product of two groups of sampled datas, wherein every group all comprises two real sampled datas, the sampling instant sum of two real sampled datas equates with the sampling instant sum of two real sampled datas in another group in one group, and the difference of the sampling instant of two sampled datas is the even-multiple in sampling time in every group.

In the calculation procedure of described detection metric, more specifically, after two real sampled datas of every group multiplied each other with multiplier, the product of two groups of real sampled datas all exported subtracter to and calculates both poor, then it is carried out smoothing processing, obtains described detection metric.

In the step that described detection metric and decision threshold compare, described decision threshold benchmark obtains decision threshold after amplifying with default multiple, compare by threshold compataror with described detection metric, then produce the marking signal that the described frequency reference signal of sign exists if detection metric surpasses described decision threshold.

Compared with prior art, the invention has the advantages that:

At first, only need the real sampled data of the reception signal (i.e. one road real signal) of real number form is directly processed, just can automatically detect frequency reference signal and whether exist, and the reception signal of plural form is sampled and process simply too much;

Secondly, owing to need not to obtain I, the mutually orthogonal signal of Q two-way of plural form, also do not have the imbalance compensation problem, so the present invention has reduced the complexity that system realizes greatly under the prerequisite of guaranteeing the frequency detecting performance;

Moreover the present invention is convenient to adopt dsp chip, fpga chip or asic chip etc. to realize.

[description of drawings]

Fig. 1 be take the present invention based on the digitlization automatic frequency detection method of the real signal theory diagram as the system on basis;

Fig. 2 is parameter A in the embodiment of the invention ²[n], C[n], S[n] the algorithm structure schematic diagram;

Fig. 3 is parameter A in the embodiment of the invention ² _L[n], C _M[n], S _NThe algorithm structure schematic diagram of [n];

Fig. 4 calculates the algorithm structure schematic diagram that local frequency source is adjusted signal in the embodiment of the invention;

Fig. 5 is in the embodiment of the invention, when frequency reference signal existence and noiseless, and detection metric C _M[n] and frequency error function S _NThe relation curve of [n] and sampling time τ.

[embodiment]

Below in conjunction with accompanying drawing, specific embodiments of the invention and principle are described in detail:

As shown in Figure 1; the present invention based on the digitlization automatic frequency detection method of real signal in another application, protected simultaneously with the applicant and in this application the open digitalized automatic frequency correction based on true frequency reference signal realizes therein, comprising A/D converter (also being A-D converter) 102, automatic frequency detection and correcting algorithm module 104, D/A converter (also being digital-to-analog converter) 106, low pass filter 108, voltage controlled oscillator 110 and frequency division/frequency multiplier 112.

In Fig. 1, the simulation real signal r (t) of reception comes from intermediate frequency or the base band output of transmitting terminal.R (t) is through sampling and the analog to digital conversion of A/D converter 102, the digital real signal r[n that obtains dispersing], send into automatic frequency and detect and correcting algorithm module 104.

The sampling clock of A/D converter 102 is provided by the local frequency source that voltage controlled oscillator 110 and frequency division/frequency multiplier 112 consist of.If the frequency of the frequency reference signal of the real signal that receives is f, the frequency of local frequency source signal is f+ Δ f, then the frequency error between local frequency source signal and the described frequency reference signal is Δ f, and the pass of sampling time τ and local frequency source signal frequency f+ Δ f is

$\mathrm{\τ}=\frac{4k+1}{4(f+\mathrm{\Δf})},$

Or

$\mathrm{\τ}=\frac{4k+3}{4(f+\mathrm{\Δf})}.$

Wherein k is design parameter, and k ∈ 0,1,2 ... }, can choose according to the needs of practical application.

Because there is frequency error Δ f in local frequency source, the sampling time τ that therefore chooses according to above-mentioned relation is uncertain, and in the present invention, its scope that allows to change is

$\frac{k}{f}<\mathrm{\&tau;}<\frac{2k+1}{2f},$

Or

$\frac{2k+1}{2f}<\mathrm{\&tau;}<\frac{k+1}{f}.$

The permission excursion of corresponding frequency error is with it

$-\frac{f}{4k+2}<\mathrm{\&Delta;f}<\frac{f}{4k},$

Or

$-\frac{f}{4k+4}<\mathrm{\&Delta;f}<\frac{f}{4k+2}.$

That is to say, in case selected sampling time τ according to above-mentioned relation, if frequency error Δ f is within the above-mentioned permission excursion, then can realize that with the present invention automatic frequency detects and proofreaies and correct; , can't realize frequency error Δ f that automatic frequency detects and proofreaies and correct otherwise will exceeding correcting range of the present invention.

Algorithm structure such as Fig. 2, Fig. 3, shown in Figure 4 of automatic frequency detection and correcting algorithm module 104 inside.This algorithm at first in Fig. 2 according to following formula

A ²[n]＝r ²[n]+r ²[n-1]，

C[n]＝2(r[n]r[n-4]-r[n-1]r[n-3])，

S[n]＝2(r[n]r[n-3]-r[n-1]r[n-2])；

Finish A respectively ²[n], C[n], S[n] calculating; Then in Fig. 3 according to following formula

$A_L^2\left[n\right]=\frac{1}{L}\underset{m=0}{\overset{L-1}{\mathrm{\&Sigma;}}}{A}^2[n-m],$

$C_M\left[n\right]=\frac{1}{M}\underset{m=0}{\overset{M-1}{\mathrm{\&Sigma;}}}C[n-m],$

$S_N\left[n\right]=\frac{1}{N}\underset{m=0}{\overset{N-1}{\mathrm{\&Sigma;}}}S[n-m].$

To A ²[n], C[n], S[n] carry out respectively smoothly, obtain A ² _L[n], C _M[n], S _N[n].Wherein L, M, N are design parameter, and L, M, N ∈ 1,2,3 ... }, can choose according to the needs of application scenario.

A ² _LThe physical meaning of [n] is for receiving the estimated value of real signal power (or its 2 times), the decision threshold benchmark whether the real signal medium frequency reference signal that receives as detection exists herein; C _MThe detection metric whether physical meaning of [n] exists for detect the real signal medium frequency reference signal that receives in certain frequency error range; S _NThe physical meaning of [n] is the frequency error function of local frequency source signal and frequency reference signal.

Only comprise 4 kinds of rudimentary algorithm unit among Fig. 2 and Fig. 3, they are delayer 202, multiplier 204, adder 206 and amplifier 208.

In Fig. 4, A ² _LThe decision threshold benchmark whether [n] exists as the detection frequency reference signal is through obtaining decision threshold α A after the amplifier 208 ² _L[n], wherein α is design parameter, and α ∈ (0,1), can choose according to the needs of practical application.

C _M[n] as detect detection metric that whether frequency reference signal exist in threshold compataror 402 with decision threshold α A ² _L[n] compares, if

C _M[n]＞αA ² _L[n]，

Judge that then frequency reference signal exists, by producing a prearranged signal, trigger and open gate 404, allow S _N[n] passes through; Otherwise judge that frequency reference signal does not exist, produce another signal at stop gate 404, forbid S _N[n] passes through.

It is pointed out that above-mentioned to A ²[n], C[n], A ² _L[n], C _MThe calculating of [n], and utilize A ² _L[n], C _MWhether [n] exists the detection of carrying out to frequency reference signal is unwanted under the occasion that the given frequency reference signal exists always.

S _N[n] if frequency reference signal exists, then is added to converter 406 enterprising line translations by gate 404 as the frequency error function, obtains the adjustment increment of local frequency source.This adjustment increment is adjusted signal through the numeral that accumulator (being made of delayer 202 and adder 206) obtains local frequency source afterwards, is sent to the D/A converter 106 among Fig. 1.

If the selected sampling time is

$\mathrm{\&tau;}=\frac{4k+1}{4(f+\mathrm{\&Delta;f})},$

Then work as

S _N[n]＞0

The time, sampling time τ is bigger than normal in expression, and namely the frequency of local frequency source signal is less than the frequency of frequency reference signal, and the local frequency source signal of capable of regulating this moment increases its frequency.When

S _N[n]＜0

The time, sampling time τ is less than normal in expression, and namely the frequency of local frequency source signal is greater than the frequency of frequency reference signal, and the local frequency source signal of capable of regulating this moment reduces its frequency.

If the selected sampling time is

$\mathrm{\&tau;}=\frac{4k+3}{4(f+\mathrm{\&Delta;f})},$

Then work as

S _N[n]＞0

The time, sampling time τ is less than normal in expression, and namely the frequency of local frequency source signal is greater than the frequency of frequency reference signal, and the local frequency source signal of capable of regulating this moment reduces its frequency.When

S _N[n]＜0

The time, sampling time τ is bigger than normal in expression, and namely the frequency of local frequency source signal is less than the frequency reference signal frequency, and the local frequency source signal of capable of regulating this moment increases its frequency.

For being adjusted increment to S _NThe conversion that [n] implements can be ± ρ sign{S _N[n] } or ± ρ S _NForms such as [n], wherein

$\mathrm{sign}\left\{x\right\}=\left\{\begin{array}{cc}-1& x<0\\ 0& x=0,\\ 1& x>0\end{array}\right.\mathrm{\&rho;}>0;$

The choosing of positive and negative number and ρ depended on symbol and the size with the voltage-controlled slope of curve of VCO chosen in sampling time.

In Fig. 1, D/A converter 106 is adjusted signal with the numeral that automatic frequency detects and correcting algorithm module 104 is calculated and is transformed into simulation adjustment signal, carry out forming voltage controlled oscillator 110 control voltages after the loop filtering through low pass filter 108 again, the local frequency source that control is made of voltage controlled oscillator 110 and frequency division/frequency multiplier 112 is so that the frequency-tracking of local frequency source signal is in the frequency of receive frequency reference signal.

Described above is the specific embodiment of the present invention, and the below is further described principle of the present invention.

Receiving real signal can be expressed as

Wherein Asin (2 π ft+ φ) is frequency reference signal, and A, f, φ are respectively amplitude, frequency and the initial phase of frequency reference signal, and η (t) is noise, and s (t) is other signal, and it is a random process.

In order to make things convenient for this to the explanation of inventive principle, VCO control voltage remains unchanged before can making adjustment, be that sampling time τ is constant, and ignore for the time being the impact of noise η (t), suppose simultaneously the sample sequence s[n of s (t)] be ergodic independent same distribution discrete random sequence, average is 0, and variance is D ²/ 2.

When frequency reference signal does not exist, have

A ²[n]＝s ²[n]+s ²[n-1]，

C[n]＝2(s[n]s[n-4]-s[n-1]s[n-3])，

S[n]＝2(s[n]s[n-3]-s[n-1]s[n-2])；

$A_L^2\left[n\right]=\frac{1}{L}\underset{m=0}{\overset{L-1}{\mathrm{\&Sigma;}}}\left(s^2\right[n-m]+s^2[n-m-1\left]\right),$

$C_M\left[n\right]=\frac{1}{M}\underset{m=0}{\overset{M-1}{\mathrm{\&Sigma;}}}2\left(s\right[n-m\left]s\right[n-m-4]-s[n-m-1\left]s\right[n-m-3\left]\right),$

$S_N\left[n\right]=\frac{1}{N}\underset{m=0}{\overset{N-1}{\mathrm{\&Sigma;}}}2\left(s\right[n-m\left]s\right[n-m-3]-s[n-m-1\left]s\right[n-m-2\left]\right).$

When L, M, N → ∞, have

A ² _L[n]→D ²，

C _M[n]→0，

S _N[n]→0。

Adjudicate relational expression for certain given α ∈ (0,1) this moment

C _M[n]＞αA ² _L[n]

Be false, therefore when L, M, N choose enough greatly, C _M[n] can be used as the non-existent detection metric of judgement frequency reference signal.

When frequency reference signal exists, have

A ²[n]＝A ²-A ²cos(2πfτ))cos(2πf(2n-1)τ+2φ)，

C[n]＝A ²(cos(8πfτ)-cos(4πfτ))，

S[n]＝A ²(cos(6πfτ)-cos(2πfτ))；

$A_L^2\left[n\right]=A^2-{\mathrm{<figure-callout\; id="2"\; label="A"\; filenames="H2008101849778E00012.png,H2008101849778E00021.png"\; state="\{\{state\}\}">A</figure-callout>}}^2\cos \left(2\mathrm{\&pi;f\&tau;}\right)\frac{1}{L}\underset{m=0}{\overset{L-1}{\mathrm{\&Sigma;}}}\cos (2\mathrm{\&pi;f}(2n-2m-1)\mathrm{\&tau;}+2\mathrm{\&phi;}),$

C _M[n]＝A ²(cos(8πfτ)-cos(4πfτ))，

S _N[n]＝A ²(cos(6πfτ)-cos(2πfτ))。

When L, M, N → ∞, have

A ² _L[n]→A ²，

C _M[n]≡A ²(cos(8πfτ)-cos(4πfτ))，

S _N[n]≡A ²(cos(6πfτ)-cos(2πfτ))。

As seen C _M[n] and S _N[n] be the function of sampling time τ just, and irrelevant with sampling instant n, and the relation curve of they and τ as shown in Figure 5.

In the situation that frequency reference signal exists, if sampling time τ is chosen for

$\mathrm{\&tau;}=\frac{1}{4(f+\mathrm{\&Delta;f})},$

Then the permission excursion of sampling time τ is

$0<\mathrm{\&tau;}<\frac{1}{2f},$

The permission excursion that is frequency error Δ f is

$-\frac{\mathrm{<figure-callout\; id="2"\; label="f"\; filenames="H2008101849778E00012.png,H2008101849778E00021.png"\; state="\{\{state\}\}">f</figure-callout>}}{2}<\mathrm{\&Delta;f}<\&infin;.$

As can be seen from Figure 5, interval (0,1/2f) in, C _M[n] is even function with respect to τ=1/4f.As Δ f=0, namely during τ=1/4f, C _M[n]=A ², reach maximum.F departs from 0 when Δ, when namely τ departs from 1/4f, and C _M[n] diminishes gradually.Always exist the neighborhood δ (α) of τ=1/4f for certain given α ∈ (0,1) this moment, as τ ∈ δ (α), and when namely frequency error Δ f is less than certain definite value, the judgement relational expression

C _M[n]＞αA ² _L[n]

Set up, therefore when L, M, N choose enough greatly, C _M[n] can be used as the detection metric that the judgement frequency reference signal exists in certain frequency error range.

It can also be seen that from Fig. 5, interval (0,1/2f) in, S _N[n] is odd function with respect to τ=1/4f.As Δ f=0, namely during τ=1/4f, S _N[n]=0.When Δ f＜0, namely during τ＞1/4f, S _N[n]＞0; Otherwise, when Δ f＞0, namely during τ＜1/4f, S _N[n]＜0.Therefore known or detect in the situation that frequency reference signal exists, when L, M, N choose enough greatly, S _N[n] can be used as the frequency error function of local frequency source signal and frequency reference signal.

By to S _N[n] carries out suitable conversion, can obtain the adjustment increment of local frequency source.The voltage-controlled curve of supposing VCO has positive slope, to S _NThe conversion of [n] can be ρ sign{S _N[n] }.If τ is near the scope of the relative narrower of 1/4f, then S _N[n] is approximately linear relationship with the relation of τ, and this moment is to S _NThe conversion of [n] also can be ρ S _N[n].Certainly, to S _NThe conversion of [n] can also have other form.

The adjustment increment of local frequency source is sent to the D/A converter and removes to produce VCO control voltage through after cumulative.If Δ f＜0, i.e. τ＞1/4f, then adjust increment on the occasion of, VCO control voltage is increased after cumulative, the frequency of local frequency source signal increases thereupon, thereby causes Δ f to increase; Otherwise, if Δ f＞0, i.e. τ＜1/4f, then adjusting increment is negative value, VCO control voltage is reduced after cumulative, the frequency of local frequency source signal reduces thereupon, thereby causes Δ f to reduce.Therefore, by above-mentioned corrective action, Δ f will fluctuate up and down round 0, thereby so that the frequency-tracking of local frequency source signal in the frequency of receive frequency reference signal, is realized the function of automatic frequency correction.

In the situation that frequency reference signal exists, if sampling time t is chosen for

$\mathrm{\&tau;}=\frac{3}{4(f+\mathrm{\&Delta;f})},$

Then the permission excursion of sampling time τ is

$\frac{1}{2f}<\mathrm{\&tau;}<\frac{1}{f},$

The permission excursion that is frequency error Δ f is

$-\frac{\mathrm{<figure-callout\; id="4"\; label="f"\; filenames="H2008101849778E00012.png"\; state="\{\{state\}\}">f</figure-callout>}}{4}<\mathrm{\&Delta;f}<\frac{\mathrm{<figure-callout\; id="2"\; label="f"\; filenames="H2008101849778E00012.png,H2008101849778E00021.png"\; state="\{\{state\}\}">f</figure-callout>}}{2}.$

As can be seen from Figure 5, the C in interval (1/2f, 1/f) _M[n] and S _N[n] respectively with interval (0, the C in 1/2f) _M[n] and S _N[n] shape is identical, but S _NThe opposite in sign of [n] is namely at τ=S of 1/4f place _N[n] passes zero point with positive slope, and at τ=S of 3/4f place _N[n] passes zero point with negative slope.Therefore, identical with above-mentioned principle, when L, M, N choose enough greatly, C _M[n] still can be used as the detection metric that the judgement frequency reference signal exists in certain frequency error range, S _N[n] still can be used as the frequency error function of local frequency source signal and frequency reference signal, but when calculating local frequency source and adjust increment to S _NIn the conversion that [n] does, S _N[n] will use-S _N[n] replaces.

Above-mentioned conclusion is promoted.In the situation that frequency reference signal exists, if sampling time τ is chosen for

$\mathrm{\&tau;}=\frac{4k+1}{4(f+\mathrm{\&Delta;f})},$

Then the permission excursion of sampling time τ is

$\frac{k}{f}<\mathrm{\&tau;}<\frac{2k+1}{2f},$

The permission excursion that is Δ f is

$-\frac{f}{4k+2}<\mathrm{\&Delta;f}<\frac{f}{4k}.$

If the sampling time, τ was chosen for

$\mathrm{\&tau;}=\frac{4k+3}{4(f+\mathrm{\&Delta;f})},$

Then the permission excursion of sampling time τ is

$\frac{2k+1}{2f}<\mathrm{\&tau;}<\frac{k+1}{f},$

The permission excursion that is Δ f is

$-\frac{f}{4k+4}<\mathrm{\&Delta;f}<\frac{f}{4k+2}.$

Wherein, k ∈ 0,1,2 ... }.For above two kinds of situations, when L, M, N choose enough greatly, C _M[n] can be as the detection metric of judgement frequency reference signal existence in certain frequency error range, S _N[n] can be as the frequency error function of local frequency source signal and frequency reference signal.But with regard to the calculating that local frequency source is adjusted increment, the adjustment increment that calculates under the previous case can be directly used in the cumulative of frequency adjustment amount, and the adjustment increment that then calculates in a kind of situation must could be used for the cumulative of frequency adjustment amount after the symbol negate.

Above explanation of the principles of the present invention is to provide ignoring in the situation of noise.If consider the impact of noise η (t), the performance of above method will worsen to some extent.The sample sequence η [n] that supposes η (t) is 0 for average, and variance is σ ²/ 2 white noise sequence, and ergodic does not then exist and L, M, N → during ∞, has when frequency reference signal

A ² _L[n]→D ²+σ ²，

C _M[n]→0，

S _N[n]→0。

When the existence of receive frequency reference signal and L, M, N → ∞, have

A ² _L[n]→A ²+σ ²，

C _M[n]→A ²(cos(8πfτ)-cos(4πfτ))，

S _N[n]→A ²(cos(6πfτ)-cos(2πfτ))。

Therefore, enough in the large situation, said method is still set up at L, M, N, and the deterioration of performance can be controlled within the range of permission.

Need to prove, in said method, after the detection threshold factor-alpha was given, sampling time τ must be near the scope of the relative narrower of (2k+1)/4f, be that frequency error must be limited in the less scope, frequency reference signal just can be detected.In order to enlarge detection range, namely enlarge the allowed band of frequency error, can adopt some measures.For example, can use Replace C _M[n] also can add a sawtooch sweep voltage at the voltage-controlled end of the past VCO of detection-phase and carry out at a slow speed frequency scanning as detection metric.In addition, in order to enlarge the range of linearity of frequency error function, can use arctg (S _N[n]/C _M[n]) replacement S _N[n] is as the frequency error function.But adopting above measure is that to increase the complexity of realization be cost.

The present invention is specially adapted to those only need to be in the system that real number field is processed to received signal.The frequency-shift repeater in the gsm mobile communication system for example, it only need to forward one road radio frequency receiving signal of real number form at the real number field shift frequency, after amplifying, and does not need a road of real number form is received processing such as carrying out again demodulates information after I that signal is transformed into plural form, the Q two-way orthogonal signalling in complex field.

In the gsm system frequency-shift repeater, adopt above-mentioned digital automatic frequency based on true frequency reference signal to detect and the descending calibrating frequency signal (being frequency reference signal) of bearing calibration tracking base stations, only need a common VCXO (VCXO), add simple algorithm realization circuit, just can for frequency-shift repeater provides high-precision local frequency source, make the relative error of the frequency of frequency-shift repeater and base station remain on desired ± 5 * 10 of GSM standard (ETSI EN 300609-4, YD/T 1337-2005) ^-8In.Provide the method for local frequency source to compare with adopting expensive constant-temperature crystal oscillator, both reduced cost, do not need again regularly to carry out artificial calibrating frequency.Compare with bearing calibration based on the digital automatic frequency detection of complex frequency reference signal with employing, owing to need not to obtain I, the mutually orthogonal frequency reference signal of Q two-way of plural form, there is not the imbalance compensation problem yet, therefore under the prerequisite of guaranteeing the frequency correction precision, greatly reduced the complexity that system realizes.

Claims (5)

1. the digitlization automatic frequency detection method based on real signal carries out digitlization for receiving terminal road simulation real signal that transmission comes to transmitting terminal and automatically detects, and to be confirmed whether to exist frequency reference signal, it is characterized in that comprising:

One road intermediate frequency or the baseband analog real signal of in the future spontaneous emission end sampled, analog-to-digital conversion is the step of digital signal;

Calculate the step of decision threshold benchmark with the sampled data of this real signal, the decision threshold benchmark is that the quadratic sum by two neighbouring samples real sampled data constantly obtains after smoothly;

Calculate step for detection of the detection metric of frequency reference signal with the sampled data of this real signal, detection metric is obtained after level and smooth by 2 times of the difference of the product of two groups of sampled datas, wherein every group all comprises two real sampled datas, the sampling instant sum of two real sampled datas equates with the sampling instant sum of two real sampled datas in another group in one group, and the difference of the sampling instant of two sampled datas is the even-multiple in sampling time in every group;

With described detection metric be multiply by the step that formed decision threshold compares behind the default coefficient by described decision threshold benchmark, if surpass this decision threshold, then produce the marking signal that the described frequency reference signal of sign exists.

2. the digitlization automatic frequency detection method based on real signal according to claim 1 is characterized in that:

In described sampling, the analog-to-digital step, sampling clock is obtained through frequency division/frequency multiplication by local frequency source signal, according to transmitting-receiving frequency difference allowed bands different in the practical application, sample frequency is got different preset values from the ratio of local frequency source signal frequency, and sample frequency changes along with the adjustment of local frequency source.

3. the digitlization automatic frequency detection method based on real signal according to claim 1 is characterized in that:

In the calculation procedure of described decision threshold benchmark, described two neighbouring samples real sampled data constantly, each is through the multiplier involution first, and the result carries out smoothing processing again through the adder addition, obtains described decision threshold benchmark.

4. the digitlization automatic frequency detection method based on real signal according to claim 1 is characterized in that:

In the calculation procedure of described detection metric, after two real sampled datas of every group multiplied each other with multiplier, the product of two groups of real sampled datas all exported subtracter to and calculates both poor, then it is carried out smoothing processing, obtains described detection metric.

5. the described digitlization automatic frequency detection method based on real signal of any one in 4 according to claim 1 is characterized in that:

In the step that described detection metric and decision threshold compare, described decision threshold benchmark obtains decision threshold after amplifying with default multiple, compare by threshold compataror with described detection metric, then produce the marking signal that the described frequency reference signal of sign exists if detection metric surpasses described decision threshold.

Images