CN103576170B - A kind of satellite search and rescue signal frequency estimating methods - Google Patents

Abstract

A kind of satellite search and rescue signal frequency estimating methods, adopt the process structure of loop iteration, narrow band filter is utilized to realize Kay fast frequency method of estimation to the requirement of simple signal, performing step is as follows: receive satellite search and rescue signal by Medium-Earth Orbit ground terminal, and coarse frequency estimation is carried out to satellite search and rescue signal, the bandwidth of design bandpass filter BPF1; By signal by the first bandwidth filter BPF1, obtain signal s ₁, by s ₁be divided into two-way; Right, s ₁a road enter ,-row down coversion, reduces sample frequency further by a withdrawal device afterwards, and with the output s of withdrawal device ₂carry out iterative estimate; By s ₂by the second bandwidth filter BPF2, and carry out intermediate frequency estimate obtain ω ₂, with ω ₂as the centre frequency of BPF2 again to s ₂carry out filtering, with this loop iteration, obtain frequencies omega ₃as the centre frequency of wave filter BPF3; By s ₁another road signal by the 3rd bandwidth filter BPF3, obtain signal s ₃; To s ₃carry out fast frequency estimation.

Description

Satellite search and rescue signal frequency estimation method

Technical Field

The invention relates to a high-precision frequency estimation method for satellite search and rescue signals, and belongs to the technical field of satellite navigation.

Background

The global satellite search and rescue system COSPAS-SARSAT was established in 1979 by a united nations of the united states, the former soviet union, france and canada, and is intended to provide effective search and rescue sar (searchandrescue) services to a variety of vessels, airplanes and individual users worldwide. Currently, there are more than 35 member countries in the system, and in 1994, china has also become a member of the organization. In more than 20 years of the operation of the system, 6167 search and rescue actions are successfully carried out as of 2006, more than 22035 persons in distress are successfully rescued, on average, 3 persons are rescued every day, and the system plays a very critical role in the fields of humanitarian rescue and the like. When a user is in danger, the beacon machine sends out a distress signal, the satellite system forwards the received 406MHz beacon signal to a middle earth orbit ground terminal station (MEOLUT), the ground station finishes detection of the beacon signal, extraction and positioning of beacon information, and reports the result to a task control center and a ground rescue center. The positioning of beacons by MEOLUT ground stations is done mainly by means of estimates of the beacon signal frequency estimates (FOA) and (TOA). Therefore, the accuracy of the beacon signal FOA and TOA estimates determines the accuracy of the MEOLUT ground station in locating the beacon position, requiring that the errors in TOA and FOA cannot exceed 7us and 0.1Hz, respectively.

For single frequency estimation, many feasible algorithms exist, for example, Rife follows a method for estimating single frequency complex tone parameters by using a kramer lower limit in a limited number of noise discrete observations, and invents a maximum likelihood estimation algorithm based on discrete fourier transform. However, DFT-based algorithms require a large amount of computation to search for the location of the spectral peak.

The FOA can be accurately measured by using a PLL, but because the signal is in a short format, the loop bandwidth must be increased in order to shorten the acquisition time, but the increase of the loop bandwidth can reduce the noise suppression capability, so that the tracking accuracy is reduced, and the improved method adopts a frequency auxiliary method to reduce the initial frequency difference to be within a quick acquisition band, so that the accurate synchronization can be ensured to be realized in a pilot frequency band, but an additional accurate frequency measurement link outside the PLL loop is required, so that the complexity of the system is increased.

By using a fast frequency estimation algorithm (Fastfrequency estimator) proposed by Kay, frequency estimation can be rapidly, accurately and efficiently carried out under the condition of high signal-to-noise ratio (SNR). However, at the MEOLUT ground station, before the FOA measurement, the signal passes through the signal detection unit, where the frequency of the signal is estimated within a frequency error range of several Hz, and through the coarse frequency estimation, the requirement of the fast frequency estimation algorithm for high signal-to-noise ratio can be achieved by passing the signal through a narrow-band pass filter, which takes the carrier coarse frequency estimation result obtained by the signal detection unit as the center frequency.

Disclosure of Invention

The invention solves the problems: the method for estimating the frequency of the satellite search and rescue signal overcomes the defects of the prior art, can finish accurate frequency estimation in the short time of burst of the satellite search and rescue signal, and has higher estimation accuracy; and the computational complexity is low.

The technical solution of the invention is as follows: a satellite search and rescue signal frequency estimation method adopts a processing structure of loop iteration, utilizes a narrow-band filter to realize the requirement of a Kay rapid frequency estimation method on a single-frequency signal, and comprises the following steps:

the method comprises the steps of receiving a satellite search and rescue signal through a medium earth orbit ground terminal station (MEOLUT), and performing coarse frequency estimation on the satellite search and rescue signal to obtain a frequency omega₁；

Second, the bandwidth of the BPF1 is designed to cover the error range of the coarse frequency estimation of the signal and is set to ω₁As the initial center frequency of BPF 1;

thirdly, the signal is passed through a first bandwidth filter BPF1 to obtain a signal s₁A 1 is to₁Is divided into two paths;

the fourth step, pair, s₁After down-conversion, the sampling frequency is further reduced by a decimator and is provided at the output s of the decimator₂Carrying out iterative estimation; the down-conversion and decimator in the fourth step can effectively reduce the operation amount and improve the operation efficiency;

the fifth step of mixing s₂Passing through a second bandwidth filter BPF2 and performing medium frequency estimation to obtain omega₂At ω₂As the center frequency of the BPF2 again to s₂Filtering is carried out, and the frequency omega is obtained by the iteration of the loop₃As the center frequency of filter BPF 3;

a sixth step of mixing s₁Passes through a third bandwidth filter BPF3 to obtain a signal s₃；

Seventh step of pairing s₃A fast frequency estimation is performed.

The fast frequency estimation in the seventh step adopts a modified fast frequency estimation algorithm as follows:

pilot frequency s of ground receiving signal₃Can be expressed as:

$r\left(n\right)=A{e}^{j({\mathrm{\ω}}_0\mathrm{nT}+\frac{1}{2}{\mathrm{\ω}}^{\′}{n}^2{T}^2+\mathrm{\θ})}+z\left(n\right)$

where z (n) is noise, A is signal amplitude, ω₀Is the carrier frequency, ω' is the current frequency modulation frequency, T is the signal period, θ is the initial phase, and Z (n) is the noise.

Carrier frequency omega₀The estimated values of (c) are:

wherein, $w\left(n\right)=\frac{\frac{3}{2}N}{N^2-1}\{1-{\left[\frac{n-(\frac{N}{2}-1)}{\frac{N}{2}}\right]}^2\},$ $C=\frac{{\mathrm{\σ}}_z^2}{2A}\left(\begin{array}{cccccc}2& -1& 0& 0& \·\·\·& 0\\ -1& 2& -1& 0& \·\·\·& 0\\ \·\·\·& \·\·\·& \·\·\·& \·\·\·& \·\·\·& 0\\ 0& 0& \·\·\·& 0& -1& 2\end{array}\right),$ n is the upper limit of N;for adjacent samplingPhase differential signal between points

Wherein v is_Q(n) is equivalent phase noise ${\mathrm{varv}}_Q\left(n\right)=\frac{A^2}{2{\mathrm{\σ}}_z^2\mathrm{}}=\frac{1}{2\mathrm{SN}{R}_r},$ A is the amplitude of the signal and,is the variance of the noise;for the phase of the signal

When the signal-to-noise ratio is high, the estimation result is equivalent to:

${\widehat{\mathrm{\ω}}}_0=\arctan \left\{\underset{t=0}{\overset{N-2}{\mathrm{\Σ}}}w\left(n\right)r^{*}\left(n\right)r(n+1)\right\}$

the estimated value of the chirp ω' obtained by the modified fast frequency estimation algorithm is:

$\left(\begin{array}{c}{\mathrm{\ω}}_0\\ {\mathrm{\ω}}^{\′}\end{array}\right)={\left[H^{\′}{C}^{-1}H\right]}^{-1}{H}^{\′}{C}^{-1}\mathrm{\Δ},$

Δ is the frequency estimation variance

$H={\left(\begin{array}{cccc}T& T& \·\·\·& T\\ T^2/2& {3T}^2/2& \·\·\·& (2N-1)T^2/2\end{array}\right)}^T,$ H' is the transposed matrix of H.

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

(1) the invention can complete accurate frequency estimation in the short time of satellite search and rescue signal burst, compared with the phase-locked loop method, the invention does not need to track for a long time and directly carries out filtering processing on the signal.

(2) By filtering out the AM signal and most of the sidebands of BPSK, the equivalent increased carrier time can be extended from 160ms to 440ms, as shown in fig. 2.

(3) Has lower computational complexity and further reduces the operation amount through a down-sampling and decimator. The phase-locked loop method requires additional high-precision estimation to ensure that the initial frequency difference of the PLL is small enough, but increases the complexity;

(4) by the cooperation of iteration and a narrow-band filter, the signal frequency range is gradually reduced, noise is filtered, and the requirements of the fast frequency estimation method provided by Kay on high signal-to-noise ratio and single frequency are met.

(5) The fast frequency estimation algorithm is modified, the beacon signals received by the ground station are retransmitted by the Galileo satellite, and Doppler frequency drift is generated due to non-uniform motion of the satellite relative to the ground station, and the drift can be approximately equivalent to linear frequency modulation of carrier frequency.

(6) The invention uses an iterative method, continuously reduces the frequency range, can ensure that the signal does not fall, and meets the requirements of a quick frequency estimation method on signal single frequency and high signal-to-noise ratio through narrow-band filtering.

Drawings

FIG. 1 is a general flow diagram of the present invention;

FIG. 2 is a COSPAS-SARSAT search and rescue signal model 1-short format;

FIG. 3 is a COSPAS-SARSAT search and rescue signal model 2, long format.

Detailed Description

As shown in fig. 1, the present invention is embodied as follows.

In a first step, a MEUT ground station receives a satellite signal.

The Galileo satellite works in a navigation constellation of a middle orbit, is provided with a search and rescue signal transponder, and when a ground user is in danger, the signal in danger is transparently forwarded to a ground station to complete the receiving and positioning of the signal and rescue the user. The tracking demodulation of the beacon signal is required to be completed to solve the frame information contained in the signal, and the signal has the characteristics of burst (random transmission time), short time (long format 520ms, short format 440ms) and unknown transmission frequency.

The beacon signal is transmitted to the ground station after being forwarded by a satellite, a MEUT ground station receives a signal which consists of a search and rescue signal and noise, wherein z (t) = y (t) + s (t), s (t) is a Gaussian white noise signal, the search and rescue signal y (t) is divided into a long format and a short format 2 (as shown in the attached figures 2 and 3), the duration of the short format beacon signal is about 440ms, wherein a pure carrier wave is about 160ms, data is about 280ms, a 24-bit preamble synchronization sequence is arranged at the beginning, and then 88-bit beacon information bits are arranged; the long format beacon signal is approximately 520ms in duration, with a pure carrier approximately 160ms and data approximately 360ms, and consists of a 24-bit preamble synchronization sequence and 12-bit beacon information bits. The data rate of the information is about 400 bps. The signal is manchester encoded and then BPSK modulated with plus or minus 1.1 rad.

The beacon signal formulation:

where m (n) = ± 1, n =1,2 … 224 (or 228), represents the nth manchester-encoded signal, a₀As amplitude, omega, of the beacon signal_cIs the carrier frequency and is,is the initial phase of the carrier.

The beacon burst signal passes through a galileo mid-orbit satellite to a MEOLUT ground station. The signals received by the ground station are:

the invention rewrites r (t) as the sum of two signals, one being the standard AM (amplitude modulated) signal r₁(t), the other being BPSK signal r₂(t), as follows:

r(t)＝r₁(t)+r₂(t)

$m_a\left(t\right)=\left\{\begin{array}{cc}\frac{1-\cos 1.1}{2}& 0\≤t\≤160\mathrm{ms}\\ \frac{-1\cos 1.1}{2}& 160\mathrm{ms}\≤t\≤440\mathrm{ms}\left(\mathrm{or}520\mathrm{ms}\right)\end{array}\right.$

performing coarse frequency estimation on the received signal, wherein a fast frequency estimation algorithm of Kay is applied to obtain a frequency omega₁。

In the second step, the bandwidth of the band pass filter BPF1 is set to 100Hz, so that it can cover the error range of the coarse frequency estimation of the signal and is measured by ω₁As the initial center frequency of the BPF 1. After the signal r (t) passes through the narrow band-pass filter BPF1, the AM signal and most of the sidebands of BPSK are filtered out, resulting in a signal close to the carrier:

ω_r(t)＝ω_c+ω_d(t)

wherein ω is_d(t) is the Doppler shift introduced by the radial motion of the orbiting satellite in Galileo, which is time-varying and non-linear due to the uncertainty of the radial motion. Since the duration of the signal burst is short, ω can be ignored_dThe higher order components of (t), and therefore can be well estimated from the following ramp frequency signal:

ω_r(t)＝ω_c+ω_d0(t)+ω_at

wherein ω is_d0(t) is constant in the burst, ω_aIs the signal slope. The frequency estimate of the beacon is an instantaneous frequency measurement of the edge portion of the 24-bit synchronization sequence portion. Due to omega_aUsually less than 1Hz/s, the burst duration is only 440ms or 550ms, the frequency offset accumulated in the burst duration is less than 1Hz, therefore, the carrier wave can be approximately regarded as a single-frequency signal, and the classical high-precision signalA single frequency estimation algorithm can be used to measure this frequency.

Thirdly, the signal s obtained by the filter BPF1 is filtered₁Is divided into two paths.

The fourth step, pair, s₁In the intermediate-stage estimation branch, the signal is firstly down-converted to about 300KHz, then the sampling frequency is further reduced to 2KHz by a decimator, and the output s of the decimator is used₂Iterative estimation is performed, and a fast frequency estimation algorithm is used in each iteration.

The fifth step of mixing s₂Passes through a filter BPF2 and performs medium frequency estimation to obtain omega₂At ω₂As the center frequency of the BPF2 again to s₂Filtering is performed, and iteration is performed in a loop. To obtain the frequency omega₃As the center frequency of the filter BPF3, the bandwidth of the BPF3 was set to 4 KHz.

A sixth step of mixing s₁The other signal passes through BPF3 to obtain signal s₃。

Seventh step of pairing s₃Since the fast frequency estimation algorithm of Kay is further adapted to single frequency signal estimation, the beacon signal received by the ground station is forwarded by the galileo satellite, while the doppler frequency drift, which may be approximately equivalent to a chirp of the carrier frequency, i.e. the pilot s of the ground received signal, is generated due to the non-uniform motion of the satellite with respect to the ground station₃Can be expressed as:

$r\left(n\right)=A{e}^{j({\mathrm{\ω}}_0\mathrm{nT}+\frac{1}{2}{\mathrm{\ω}}^{\′}{n}^2{T}^2+\mathrm{\θ})}+z\left(n\right)$

where z (n) is noise, A is signal amplitude, ω₀Is the carrier frequency, ω' is the current frequency modulation frequency, T is the signal period, θ is the initial phase, and Z (n) is the noise.

Signal to noise ratioThe higher time received signal may be expressed as:

$r\left(n\right)\≈{\mathrm{Ae}}^{j[{\mathrm{\ω}}_0\mathrm{nT}+\mathrm{\θ}+v_q\left(n\right)]}={\mathrm{Ae}}^{\mathrm{j\φ}\left(n\right)}$

the signal phase at this time is:

$\mathrm{\φ}\left(n\right)={\mathrm{\ω}}_0\mathrm{nT}+\frac{1}{2}{\mathrm{\ω}}^{\′}{n}^2{T}^2+\mathrm{\θ}$

the invention only needs to estimate the frequency, does not need to estimate the initial phase, and can avoid the problem by a difference method. And the phase difference signal between adjacent sampling points is:

$\mathrm{\Δ}\left(n\right)=\mathrm{\φ}\left(n\right)-\mathrm{\φ}(n-1)={\mathrm{\ω}}_{0}T+\frac{1}{2}{\mathrm{\ω}}^{\′}{n}^2{T}^2+v_Q\left(n\right)-v_Q(n-1)$

wherein v is_Q(n) is equivalent phase noise ${\mathrm{varv}}_Q\left(n\right)=\frac{A^2}{2{\mathrm{\σ}}_z^2\mathrm{}}=\frac{1}{2\mathrm{SN}{R}_r},$ A is the amplitude of the signal and,is the variance of the noise.

Carrier frequency omega₀The maximum likelihood estimation of (c) can be equivalent to the minimum variance unbiased estimation of the above linear model, i.e.:

${\widehat{\mathrm{\ω}}}_0=\frac{1}{T}\frac{I^T{C}^{-1}\mathrm{\Δ}}{I^T{C}^{-1}I},$ wherein $C=\frac{{\mathrm{\σ}}_z^2}{2A}\left(\begin{array}{cccccc}2& -1& 0& 0& \·\·\·& 0\\ -1& 2& -1& 0& \·\·\·& 0\\ \·\·\·& \·\·\·& \·\·\·& \·\·\·& \·\·\·& 0\\ 0& 0& \·\·\·& 0& -1& 2\end{array}\right)$

From this it can be deduced that: omega₀Estimated value of (a):

$\begin{array}{c}{\widehat{\mathrm{\ω}}}_0=\frac{1}{T}\frac{I^T{C}^{-1}\mathrm{\Δ}}{I^T{C}^{-1}I}\\ =\underset{t=0}{\overset{N-2}{\mathrm{\Σ}}}w\left(n\right)\mathrm{\Δ\φ}\left(n\right)\\ =\underset{t=0}{\overset{N-2}{\mathrm{\Σ}}}w\left(n\right)\arctan \left[r^{*}\left(n\right)r(n+1)\right]\end{array}$

wherein, $w\left(n\right)=\frac{\frac{3}{2}N}{N^2-1}\{1-{\left[\frac{n-(\frac{N}{2}-1)}{\frac{N}{2}}\right]}^2\},$ n is the upper limit of N.

When the signal-to-noise ratio is high, the estimation result can be equivalent to:

${\widehat{\mathrm{\ω}}}_0=\arctan \left\{\underset{t=0}{\overset{N-2}{\mathrm{\Σ}}}w\left(n\right)r^{*}\left(n\right)r(n+1)\right\}$

meanwhile, by using a modified fast frequency estimation algorithm, the value of the chirp ω' can also be estimated, and at this time, the frequency estimation variance:

$\mathrm{\Δ}=H\left(\begin{array}{c}{\mathrm{\ω}}_0\\ {\mathrm{\ω}}^{\′}\end{array}\right)-V$

$H={\left(\begin{array}{cccc}T& T& \·\·\·& T\\ T^2/2& {3T}^2/2& \·\·\·& (2N-1)T^2/2\end{array}\right)}^T$

frequency estimation:

$\left(\begin{array}{c}{\mathrm{\ω}}_0\\ {\mathrm{\ω}}^{\′}\end{array}\right)={\left[H^{\′}{C}^{-1}H\right]}^{-1}{H}^{\′}{C}^{-1}\mathrm{\Δ},$ h' is the transposed matrix of H.

Claims (1)

1. A satellite search and rescue signal frequency estimation method is characterized in that a processing structure of loop iteration is adopted, a narrow-band filter is utilized to realize the requirement of a Kay rapid frequency estimation method on a single-frequency signal, and the method comprises the following steps:

the method comprises the steps of receiving a satellite search and rescue signal through a medium earth orbit ground terminal station (MEOLUT), and performing coarse frequency estimation on the satellite search and rescue signal to obtain a frequency omega₁；

Second, the bandwidth of the first band-pass filter BPF1 is designed to cover the error range of the coarse frequency estimation of the signal, and thenω₁As the initial center frequency of BPF 1;

thirdly, the signal is passed through a first band-pass filter BPF1 to obtain a signal s₁A 1 is to₁Is divided into two paths;

the fourth step is to s₁Down-converting the signal, further reducing the sampling frequency by a decimator, and outputting s by the decimator₂Carrying out iterative estimation;

the fifth step of mixing s₂Passing through a second band-pass filter BPF2 and performing intermediate frequency estimation to obtain omega₂At ω₂As the center frequency of the BPF2 again to s₂Filtering is carried out, and the frequency omega is obtained by the iteration of the loop₃As the center frequency of the third band-pass filter BPF 3;

a sixth step of mixing s₁Passes through a third band-pass filter BPF3 to obtain a signal s₃；

Seventh step of pairing s₃A fast frequency estimation is performed.