Summary of the invention
The object of the present invention is to provide a kind of improved navigation satellite signal tracking method, the method can accurately judge and follow the tracks of the weak GPS signals with larger noise, reduces the impact of erroneous judgement on tracking results.
Purpose of the present invention can realize by following technical measures:
A kind of improved navigation satellite signal tracking method comprises the following steps:
(1) GPS receiver receiving satellite signal data, carry out navigation message and peel off carrying out signal data after the correlator integration;
(2) data after navigation message is peeled off are carried out fast fourier transform;
(3) extract the signal spectrum peak value and judge that whether this spectrum peak is effective simultaneously, and calculate new carrier frequency value f
l;
(4) new carrier frequency value is carried out to filtering and obtain the tracking signal of revising.
The process that in described step (1), navigation message is peeled off is: at first complete the frame synchronization of navigation data, then read in the symbol of each bit in corresponding frame, finally symbol is multiplied each other with I and Q component respectively.
Judge in described step (2) that the whether effective process of spectrum peak is: a decision threshold is set, and when spectrum peak is greater than this thresholding, this spectrum peak is effective.
Choosing of described decision threshold adopts probability and mathematical statistics method to obtain, or adopts the size that compares peak energy and spectrum mean energy to obtain.
The scope of extracting the signal spectrum peak value in described step (2) is between the 1st~13 in frequency spectrum and between the 116th~128.
Calculate new carrier frequency value f in described step (3)
lprocess be: as the Frequency point n of the maximum by thresholding between 1~13 and spectrum peak effective, frequency variation is n * 0.39; As the Frequency point n of the maximum by thresholding between 116~128 and spectrum peak effective, frequency variation is (n-128) * 0.39; Otherwise, when spectrum peak is invalid, frequency variation is set to 0; New carrier frequency value f
lfor existing carrier frequency adds the value of frequency variation.
In described step (4), to frequency values, adopt Kalman filter Kalman to carry out filtering.
State-transition matrix and the observing matrix of described Kalman wave filter are respectively:
Wherein, Ts means the time interval that frequency is upgraded.
The inventive method has following beneficial effect compared to existing technology:
1, with traditional C ostas phaselocked loop, compare, this method can realize the coherent integration of long period, the problem that the tracking effect while having avoided phase-locked loop method to use long integral time worsens;
1), increase decision threshold 2, with the existing tracking based on FFT, compare, this method mainly combines by three kinds of strategies the impact use reduced after the erroneous judgement of frequency domain peak value supervise:; 2), limit frequency range of choice; 3), the frequency values upgraded is used to filtering.Filtering method wherein particularly, due to the variation tendency of having considered that frequency is overall, so can effectively strengthen the performance of original tracking.
Embodiment
In order to follow the tracks of weak GPS signals, just need to extend the integral time of receiver, and extend the negative effect that PLL is brought for fear of integral time, need to use the frequency domain tracking based on discrete Fourier variation (DFT) or fast fourier transform (FFT) to replace original PLL.As stated in the Background Art, there is the problem of erroneous judgement in the frequency domain tracking when noise is larger, and once erroneous judgement tends to cause lethal effect, so need a kind of method, the general trend of utilizing a period of time intercarrier frequency to change, avoid the impact of minority erroneous judgement on tracking results.Therefore, the present invention is by the improvement of existing frequency domain tracking, realizes under feeble signal more stabilized carrier signal trace.
As shown in Figure 2, the realization of the inventive method mainly comprises the following steps:
(1) GPS receiver receiving satellite signal data, carry out navigation message and peel off carrying out signal data after the correlator integration;
Because modulation has navigation message, can change the sign of integral result, so need to remove the impact of navigation message.Because first three subframe of navigation message is upgraded once in every 2 hours, the every some months of rear two subframe upgrades once, therefore can utilize this repeatability to peel off the navigation message be modulated in carrier wave.The source of navigation message can be by supplementary mode from downloaded, the data that also can utilize receiver to preserve.Adopt the binary coding mode because value is modulated in the navigation message in carrier wave, utilize these data just can know in every frame that each bit is+1 or-1.
Therefore, the process that navigation message is peeled off is: at first complete the frame synchronization of navigation data, then read in the symbol of each bit in corresponding frame, finally symbol is multiplied each other and just completed navigation message and peel off with I and Q component respectively.
Due to the existence of navigation message, make the cumulative time of coherent integration can not surpass 20ms, after peeling off navigation message, just can carry out integral operation for a long time arbitrarily.The data of peeling off after navigation message can further be carried out the cumulative operation of integration, until deadline length is long coherent integration Ts second.That is: first carry out the coherent integration in 20ms, then peel off navigation message, can further surpass afterwards the coherent integration of 20ms.
The value obtained for top integration, can think to be exactly the sampled value in the Ts time interval to signal medium and low frequency part.When frequency variation is determined in back, the inverse of this Ts value is exactly the frequency size of each Frequency point time interval after fast fourier transform.
(2) data after navigation message is peeled off are carried out fast fourier transform; Integral time in tracing process and each bit synchronous, 20ms namely, being equivalent to like this sampling period is 20ms, the respective frequencies scope is 50Hz.Suppose to adopt the FFT of 128, can obtain the frequency resolution of about 0.39Hz, wherein, the counting of Fourier transform can not also be selected or etc. at 64 at 256, selects at present to obtain result preferably at 128.Fig. 3 is exactly the spectrogram after Fourier transform.
(3) extract the signal spectrum peak value and judge that whether this spectrum peak is effective simultaneously, and calculate new carrier frequency value f
l; The scope of extracting the signal spectrum peak value is between the 1st~13 in frequency spectrum and between the 116th~128, and this range size is determined by the experience of practical operation.Judge that the whether effective process of spectrum peak is: a decision threshold is set, and when spectrum peak is greater than this thresholding, this spectrum peak is effective.
Choosing of decision threshold adopts probability and mathematical statistics method to obtain, or adopts the size that compares peak energy and spectrum mean energy to obtain.
Calculate new carrier frequency value f
lprocess be: as the Frequency point n of the maximum by thresholding between 1~13 and spectrum peak effective, frequency variation is n * 0.39; As the Frequency point n of the maximum by thresholding between 116~128 and spectrum peak effective, frequency variation is (n-128) * 0.39; Otherwise, when spectrum peak is invalid, frequency variation is set to 0; New carrier frequency value f
lfor existing carrier frequency adds the value of frequency variation.
The frequency that peak value in frequency spectrum is corresponding is exactly frequency-splitting f
l-f
c, but as mentioned before, if when the noise in signal is larger, correctly extract comparatively difficulty of spectrum peak, so a decision threshold can be set, when spectrum peak is greater than this thresholding, just think effectively.Choosing of thresholding can, by the method for probability and mathematical statistics, also can simply be compared the size of peak energy and spectrum mean energy.
When receiver is static, the doppler changing rate maximum that satellite motion causes is no more than 1Hz/s, and average doppler changing rate is 0.54Hz/s, and works as receiver in the low-speed motion state, and doppler changing rate is also not too large.Here the main utilization of considering indoor situation, therefore can think that receiver is in dynamically low.Suppose that in integral time, frequency change is no more than 5Hz, the scope that frequency is selected so is exactly the 1st~13 and the 116th~128 point in frequency spectrum.By such restriction, can avoid choosing that noise causes, be in the peak value of frequency spectrum center section, reduce the possibility of erroneous judgement.
(4) to frequency values f
lcarry out Kalman filtering and obtain the tracking signal of revising.
Suppose that Current GPS receiver receives Mou road satellite-signal and is:
r(t)=AD(t)c(t)cosθ+n(t)
In formula, A means the gps signal amplitude; N (t) means the additive white Gaussian noise that power is σ 2; D (t) means the satellite message data; C (t) means the C/A code.The input signal of receiver base band is digital medium-frequency signal, and r (t) discretize is expressed as r (kTs), and Ts means sampling interval.As shown in Figure 2, the P road signal before integration can be expressed as I (kTs) and Q (kTs), when P road local code aligns with the GPS C/A code received, can obtain:
F
cmean intermediate frequency carrier,
mean doppler changing rate (Hz/s), f
lmean local reproduction carrier wave.Owing to being equivalent to have carried out low-pass filtering longer integral time, so only comprised low frequency signal in above formula.Obviously only needing just can be by frequency difference information f by fast fourier transform (FFT)
l-f
cextract.
By the operation of front, by the analysis to frequency domain, obtained the variable quantity of a carrier frequency, but still can not guarantee that in the situation that weak signal, the frequency obtained is correct.The fact should be noted that is: the relative motion of receiver and satellite is continuous, and the variation of frequency must be also continuous, the error that certainly exists noise to cause if large sudden change appears in result of calculation.So just can carry out filtering to the frequency of choosing previously, make the result of variations of frequency more level and smooth.
Kalman filter (Kalman Filter) is a kind of more satisfactory selection.The rate of change of frequency is taken into account, just can obtain the dynamic model of frequency change.Due to the existence of noise, may in tracing process, select wrong spectrum peak, obtain wrong frequency change.For the larger variation of overall variation trend difference with frequency, wave filter can be revised it, the noise wherein comprised that has been equivalent to filtering.
The beneficial effect of the inventive method below is described according to the data of actual acquisition.At first the configuration effort environment, be positioned over outdoor spacious place by sampling instrument, and the intermediate frequency of collection signal is 4.092MHz, and sample frequency is 16.368MHz, and 2bit quantizes.The instrument that algorithm is realized is Matlab 7.0 softwares.It is as follows that concrete signal is processed tracing process: the signal collected is added to noise, phaselocked loop can't correctly be followed the tracks of, attempt following the tracks of by this method more subsequently.Specific operation process is as follows:
1, the carrier signal of input signal and local reproduction multiplies each other, then carry out synchronous related operation with the C/A code of this locality reproduction, the 20ms of take after bit synchronous is the cycle, obtains 2 integral results, be respectively and input signal homophase and anti-phase component, be designated as Ip and Qp.Here Ip and Qp can regard common expression integral result as, because they are two values after the integral result Orthogonal Decomposition.
2, utilize the local almanac data of preserving to carry out symbol to Ip and Qp and peel off, obtain a complex values M
1=Ip+jQp.At first complete frame synchronization, then by demodulation, navigation message (almanac data is wherein arranged) is out compared with the local almanac data of preserving, afterwards just can peeling off for navigation message on correspondence.Stripping process is multiplied by text+and 1 or-1.
3,, through 128 20ms, can obtain the sequence M of 128 complex values
1m
128, this sequence is carried out to 128 FFT computings, obtain the frequency domain distribution of 128.Here in other words, carry out continuous 128 coherent integrations of long 20ms (total 2.56s consuming time) respectively, obtain 128 integral results, the sequence that these 128 results are formed is carried out the FFT computing and is obtained a frequency domain distribution.
4, use the method for the size that compares peak energy and spectrum mean energy to be adjudicated, find an amount of amplitude maximum in the 1st~13 and the 116th~128 in frequency spectrum, mean value by this amount divided by other spectrum components, threshold value is chosen as 3, if the ratio obtained is greater than 3, think that this component represents effective frequency change, otherwise think that frequency does not change (being that variable quantity is 0).
5, frequency variation is added to primary frequency brings the Kalman wave filter into as new frequency observed reading, calculate new frequency after filtering.State-transition matrix and the observing matrix of Kalman wave filter are respectively:
Ts means the time interval that frequency is upgraded, namely coherent integration time.
Systematic error and observational error matrix are respectively:
This is the matrix that must use in the Kalman filtering operation, after determining these matrixes and input quantity, according to fixing calculation step, just can complete filtering.Q
xand R
xthe system noise equation and the observation noise variance that mean respectively frequency, Q
xand R
xthe system noise equation and the observation noise variance that mean respectively frequency change rate.
The state vector of system is chosen as
wherein x means frequency,
mean frequency change rate.Wherein, state vector (corresponding also have observation vector) is the same with the matrix of front, is the element of fixing requirement in Kalman filtering, and description will be carried out some quantity of state in the system of filtering.
Fig. 4 has shown the tracking results of employing traditional C ostas phaselocked loop and the tracking results contrast signal of the inventive method.Can find out, if there is no filtering, when 23s, the occurrence frequency judgement makes mistakes, and can cause the complete mistake of tracking results of back.Visible by after filtering, the stability of following the tracks of being had very great help.
Embodiments of the present invention are not limited to this; under the above-mentioned basic fundamental thought of the present invention prerequisite; modification, replacement or change according to the ordinary skill knowledge of this area and customary means to other various ways that content of the present invention is made, within all dropping on rights protection scope of the present invention.