CN1446014A - Realizing Method for automatic frequency tracking of mobile terminal system - Google Patents

Abstract

A tracking method in the present invention carries on derotating treatment for digit orthogonal signal at receiving end at first; and then calculates the phase of baseband signal after derotating treatment; and spreads it out to become a function similar to linear one as well as utilizes the curve-fitting to derivate a burst frequency deviation value to carry on frequency deviation peak value error detection for the frequency deviation value, and finally carries on the smoothing treatment for it and uses the treated frequency deviation value as the regulation signal for the voltage-controlled oscillator.

Description

A kind of method that realizes automatic frequency tracking of mobile terminal system

Technical field

The present invention relates to the mobile terminal system in the field of mobile communication, specifically, relate to automatic frequency tracking and control (AFC) method in GSM or the GPRS mobile terminal system.

Background technology

In mobile communication, the channel fading that is caused by frequency change has two kinds: a kind of is the caused decline of inherent frequency error between the local oscillating frequency of base station and travelling carriage; Another kind is the channel fading that Doppler frequency shift caused that travelling carriage moves introducing.The former frequency difference is floated frequently by the long-term or short-term of physical device and is caused, belongs to slow fading; Comparatively speaking, the latter belongs to rapid fading.Therefore in portable terminal, must adopt automatic frequency control technology (AFC) that two kinds of declines are compensated, to guarantee correct reception.

GSM or gprs system all adopt Gauss's minimum phase shift keying (GMSK) modulation.GMSK is a kind of permanent envelope Continuous Phase Modulation technology, and for each code element, beginning to the difference of the carrier phase of code element termination from code element is pi/2, and carrier phase is pressed linear rule and changed in a code-element period.So the portable terminal receiver is before doing symbol judgement, the maximum phase of its symbol be offset must not and be surpassed pi/2, mixes repeatedly otherwise phase place will occur, cause to adjudicate and make mistakes, to such an extent as to data can not correctly recover, promptly

2 π Δ fT _s≤ pi/2 wherein, according to the regulation of mobile communication system to the time, Ts=3.69us, the maximum frequency deviation Δ f that therefore can bear is 271KHz/4=67.7KHz, promptly mobile terminal system requires to follow the tracks of the frequency shift (FS) of maximum 67.7 KHz.

At present common automatic frequency tracking algorithm comprises two kinds of time-delay product algorithm and discrete Fourier transform (DFT) (DFT) algorithms.

Time-delay product algorithm mainly is to grip product altogether by received signal r (t) that asks the T0 sampling instant and its inhibit signal r (t-T), and then calculating phase place, therefore require hypothesis in a symbolic range, frequency departure between the fading channel factor, phase deviation, portable terminal local frequency and the received signal carrier frequency and the I that receives, Q symbol do not change, and obtain phase place and frequency shift (FS) by arctan function then.Because have the hypothesis prerequisite, the accuracy of this method is subjected to very big influence, especially to estimate under the situation that frequency deviation is bigger to be forbidden, error is bigger.

Discrete Fourier transform (DFT) (DFT) algorithm is by calculating the energy of each sampled point in the DFT transformation range $P_k=R_k^2+M_k^2$ , try to achieve a peak value of frequency spectrum in the DFT transformation range, the value of this peak value correspondence is exactly a frequency deviation.Specific to gprs system, owing to the DFT algorithm is realized under character rate, so the frequency shift (FS) of gained is with respect to character rate.The character rate of gprs system is 270KHz, and a precision problem is also arranged here.Such as the DFT that calculates 256 points, the frequency of gained ceiling capacity is located at the 38th, and then the actual frequency of this point is:

(38/256)*270kHz＝40kHz。

Clearly, 256 DFT precision can only reach about 1KHz, even 2048 DFT, precision also just can reach 132Hz.On the other hand, when DFT counts when increasing, can increase operation time, and 1 burst comprises 156 bits in the gprs system, and 2048 DFT need tens bursts at least so, and adopting the words of this method to postpone will be too big.

In sum, the AFC track algorithm of two kinds of classics all can't satisfy the needs of GSM or gprs system more than.

Summary of the invention

Therefore technical problem to be solved by this invention provides a kind of automatic frequency tracking method that is used for portable terminal, not high to overcome in the prior art precision, can't satisfy the shortcoming that mobile communication system requires, produce reliable frequency adjusted signal, improve the performance of automatic frequency tracking.

Automatic frequency tracking method of the present invention may further comprise the steps:

One, the digital quadrature signal that receiving terminal is received goes rotation processing;

Two, calculate the phase place go the baseband signal after the rotation processing;

Three, the result of step 2 is launched into the function of a similar linearity;

Four, utilize the method for curve fit to try to achieve the exemplary frequency deviation values of burst;

Five, the frequency deviation value to the step 4 gained carries out the detection of frequency deviation peak error;

Six, frequency deviation value is carried out smoothing processing, and with the adjustment signal of treated frequency deviation value as voltage controlled oscillator.

The method of the invention adopts curve-fitting method estimated frequency error, and detects and the frequency shift (FS) smoothing processing by peak error, makes frequency offset computations precision height, and can reduce The noise, satisfies the requirement of mobile communication system for frequency-tracking.Simultaneously, the present invention can adopt devices commonly used such as DSP or FPGA to realize, and is convenient and easy.

Description of drawings

Fig. 1 is the structured flowchart that adopts GPRS terminal system receiving unit of the present invention.

Fig. 2 realizes automatic frequency tracking method schematic diagram of the present invention.

Embodiment

Below in conjunction with accompanying drawing, the method for the invention is described in detail.

The method of automatic frequency tracking of the present invention is at the frequency calibration pulse FB of mobile communcations system, that is to say, before carrying out automatic frequency tracking, at first will determine the original position of calibration pulse FB, and finish symbol level synchronously.

In the block diagram of GPRS terminal receiver shown in Figure 1, receiver comprises antenna, radio-frequency front-end 12, low-converter 14, local oscillator 18, analog to digital converter 16, digital signal processor 20, digital to analog converter 22 and voltage controlled oscillator 24.After antenna receives signal, be coupled to radio-frequency front-end 12, the input of low-converter 14 is coupled in the output of radio-frequency front-end 12, the input of analog to digital converter 16 is coupled in the output of low-converter 14, the input of digital signal processor 20 is coupled in the output of analog to digital converter 16, digital signal processor 20 output frequency offset estimating values, and by digital to analog converter 22 and voltage controlled oscillator 24 control local oscillators 18 adjusting local oscillation signal frequencies.

Gprs system adopts the GMSK modulation technique.Receive the GMSK modulation signal of base station transmission when antenna after, amplify through radio-frequency front-end 12 earlier, remove carrier wave by low-converter 14 then, such baseband signal is converted to digital signal Ir (k), Qr (k) by analog to digital converter 16.And digital signal processor 20 is used to follow the tracks of the drift of carrier frequency, the deviation of estimating carrier frequency, this frequency deviation value is converted to analog signal through digital to analog converter 22 then, adjust the frequency of local oscillator 18 by voltage controlled oscillator 24, thereby reduce because the decline of the GPRS receiver performance that frequency shift (FS) causes.The method of the invention is mainly reflected in the digital signal processor 20.

Fig. 2 is the schematic block diagram of automatic frequency tracking method of the present invention.Digital signal processor 20 can be divided into rotary module 24, phase calculation and expansion module 26, frequency deviation estimating modules 28, the level and smooth module 32 of frequency deviation, phase place peak error detection module 30, memory module 34.The quadrature Ir (k) of analog to digital converter 16 outputs, Qr (k) signal are as the input of removing rotary module 24, because the burst of the frequency calibration among GPRS FB is complete ' 0 ' sequence, the natural frequency skew of 67.7KHz after modulating through GMSK, signal is arranged, therefore at first signal Ir (k), Qr (k) to be converted into baseband signal I (k), Q (k) fully, as phase calculation and the input that launches module 26.Because a burst has 148 sampled points, so the value of k is from 0 to 147.Phase calculation with launch the phase place of module 26 according to I (k), 148 sampled points of Q (k) calculated signals, its scope be (π ,+π), and exist and suddenly change.The phase unwrapping of 148 sampled points that will calculate then becomes the function # (k) of a similar linearity, k=0,147, output to frequency deviation estimating modules 28, at this moment there is not sudden change between all phase values, because disturb and The noise, each θ (k) is located substantially on the both sides of straight line.Frequency deviation estimating modules 28 adopts curve-fitting methods, the frequency shift (FS) f ' that can accurately estimate this burst (k), this also is one of characteristics of the present invention, can guarantee precision and the reliability calculated simultaneously.The output of frequency deviation estimating modules 28 is as the input of phase place peak error detection module 30.Because the influence of decline, bigger peak or paddy can appear in phase place, and for fear of the instability of the automatic frequency tracking system that causes thus, the present invention also detects the phase deviation peak value that calculates, and deletes off-limits phase value.At last the frequency deviation of current burst and the frequency deviation of a last burst are carried out smoothly, get frequency deviation value f (k) to the end, export it to D/A converter module 22 and be converted to analog signal, the out-put supply of control voltage controlled oscillator 24, thereby further control local frequency, also f (k) is kept in the memory 34 simultaneously, carries out frequency deviation for next burst and smoothly use.

Further specify the method for the invention below.

Owing to adopt the GMSK modulation technique, certainly exist a fixing phase deviation, for burst FB, its information code element is ' 0 ' entirely, then intrinsic phase deviation is expressed as: $\mathrm{\θd}=\frac{\mathrm{\π}}{2}k,k=\mathrm{0,1},...147;$ That is to say that what obtain is not baseband signal after receiving terminal unloads ripple, but centre frequency is the signal of 67.7kHz.Therefore before frequency offset estimating, at first need rotation, be about to received signal and be transformed into base band.Particularly, signal Ir (the k)+iQr (k) that will unload exactly behind the ripple multiply by twiddle factor

Go postrotational baseband signal still to divide I road and Q road signal, be expressed as respectively:

I(k)＝αsin(θ ₁(k))

Q (k)=α cos (θ ₁(k)) wherein α represents signal amplitude, θ ₁(k) expression signal phase, k=0,1 ... 147.

By following formula as can be known, phase theta ₁(k)=and atan2 (Q (k), I (k)) ,-π≤θ ₁(k)＜π.Observe θ ₁(k) curve will find that there is sudden change in phase place, phase phasic difference ± 2 π before and after the sudden change.The reason that SPA sudden phase anomalies occurs is that atan2 () function folds phase place in order to guarantee the codomain of function.Therefore when running into this sudden change, it should be launched a similar linear function, as the input of frequency offset estimating curve fit.

Be to decide method of deploying by the phase difference of judging neighbouring sample point during expansion, θ (k) is the phase place after the expansion, as the input of frequency offset calculation:

1) as | θ ₁(k)-θ ₁(k-1) | when being not more than π, then need not launch, i.e. θ (k)=θ ₁(k);

2) as | θ ₁(k)-θ ₁(k-1) | greater than π and θ ₁(k)-θ ₁(k-1) greater than 0, θ (k)=θ then ₁(k)-2 π;

3) as | θ ₁(k)-θ ₁(k-1) | greater than π and θ ₁(k)-θ ₁(k-1) be not more than 0, then θ (k)=θ ₁(k)+2 π.

Phase theta (k) can be expressed as

θ (k)=ω ₀KT _s+ θ ₀+ η (k), k=0,1 ... 147 ω wherein ₀The angular frequency of expression received signal, θ ₀The initial phase of expression received signal, the phase error that η (k) expression is caused by noise.Curve-fitting method to seek a curve exactly and the phase sample value the most approaching, just ask the estimated value ω ' and the θ ' of frequency and initial phase, make error $\mathrm{\ϵ}=\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{[\mathrm{\θ}\left(k\right)-({\mathrm{\ω}}^{\′}k{T}_s+{\mathrm{\θ}}^{\′})]}^2$ Minimum.

The method of employing curve fit can be in the hope of the estimated value of angular frequency ${\mathrm{\ω}}^{\′}=\underset{k=1}{\overset{N}{\mathrm{\Σ}}}(\mathrm{Ak}+B)\mathrm{\θ}\left(k\right)$ Wherein coefficient A, B are respectively $A=\frac{12}{T_s(N-1)N(N+1)}$ $B=-\frac{6}{T_s(N-1)N}$ T wherein _sThe expression symbol sampler time, be 3.69us; N represents the significant bit number that a burst comprises, and promptly value is 147.

The present invention do not relate to the estimated values theta of initial phase ', therefore no longer provide its computing formula.

According to the GMSK modulation theory, be no more than pi/2 between the start-phase of each sampled point and the termination phase.Calculate poor between phase estimation value and each phase sample value so, if maximum difference thinks then that greater than pi/2 the peak error of phase place has surpassed allowed band, the frequency offset estimating value of this secondary burst is no longer output just.

Frequency deviation value level and smooth actual is the filtering to frequency regulating amount, can avoid the influence that big rise and big fall brought of frequency deviation value.The smoothing method of frequency deviation value is:

To first burst, get f _Ave(0)=f ' (0);

To second burst, then get f _Ave(1)=(f ' (1)+f ' (0))/2;

For other bursts, then get $f_{\mathrm{ave}}\left(k\right)=\frac{7}{8}{f}_{\mathrm{ave}}(k-1)+\frac{1}{8}{f}^{\′}\left(k\right).$

The frequency deviation value f that will obtain through aforementioned calculation _Ave(k) be converted to analog signal, the out-put supply of control voltage controlled oscillator 24 is further controlled local frequency; Simultaneously with frequency deviation value f _Ave(k) also to be kept in the memory 34, be used for the calculating of next burst.

In sum, biggest advantage of the present invention is to guarantee the precision of frequency offset estimating, satisfy the requirement of GPRS terminal system, and smoothly can significantly reduce the shake of frequency offset estimating value by detection of phase place peak error and frequency, guarantee the steady of AFC system and local oscillation signal, thereby the performance of whole receiver is provided.

Claims (5)

1, a kind of method that realizes automatic frequency tracking of mobile terminal system is characterized in that, may further comprise the steps:

One, the digital quadrature signal that receiving terminal is received goes rotation processing;

Two, calculate the phase place go the baseband signal after the rotation processing;

Three, the result of step 2 is launched into the function of a similar linearity;

Four, utilize the method for curve fit to try to achieve the exemplary frequency deviation values of burst;

Five, the frequency deviation value to the step 4 gained carries out the detection of frequency deviation peak error;

Six, frequency deviation value is carried out smoothing processing, and with the adjustment signal of treated frequency deviation value as voltage controlled oscillator.

2, the method for automatic frequency tracking as claimed in claim 1 is characterized in that, going rotation processing in the described step 1 is to unload signal times behind the ripple with twiddle factor

3, the method for automatic frequency tracking as claimed in claim 1 is characterized in that, the method for deploying of described step 3 is to realize by the phase difference of judging neighbouring sample point, specifically is

1) as | θ ₁(k)-θ ₁(k-1) | when being not more than π, then need not launch, i.e. θ (k)=θ ₁(k);

2) as | θ ₁(k)-θ ₁(k-1) | greater than π and θ ₁(k)-θ ₁(k-1) greater than 0, θ (k)=θ then ₁(k)-2 π;

3) as | θ ₁(k)-θ ₁(k-1) | greater than π and θ ₁(k)-θ ₁(k-1) be not more than 0, then θ (k)=θ ₁(k)+2 π; Wherein θ (k) is the phase place after launching, as the input of frequency offset calculation.

4, the method for automatic frequency tracking as claimed in claim 1, it is characterized in that, described step 5 frequency deviation peak error detects and comprises and calculate poor between phase estimation value and each phase sample value, if maximum difference is greater than pi/2, think that then the peak error of phase place has surpassed allowed band, the frequency offset estimating value of this secondary burst is no longer output just.

5, the method for automatic frequency tracking as claimed in claim 1 is characterized in that, the smoothing method of described step 6 frequency deviation value is: to first burst, get f _Ave(0)=f ' (0); To second burst, then get f _Ave(1)=(f ' (1)+f ' (0))/2; For other bursts, then get $f_{\mathrm{ave}}\left(k\right)=\frac{7}{8}{f}_{\mathrm{ave}}(k-1)+\frac{1}{8}{f}^{\′}\left(k\right).$

Images