CN107733569B

CN107733569B - Satellite load multi-beam sampling data compression method

Info

Publication number: CN107733569B
Application number: CN201710888627.9A
Authority: CN
Inventors: 田野; 王力男
Original assignee: CETC 54 Research Institute
Current assignee: CETC 54 Research Institute
Priority date: 2017-09-27
Filing date: 2017-09-27
Publication date: 2020-07-07
Anticipated expiration: 2037-09-27
Also published as: CN107733569A

Abstract

The invention provides a compression method of satellite load multi-beam sampling data, and particularly relates to a bit number quantization method for compressing satellite multipath large-dynamic-range signals. The method has the following characteristics: (1) on the satellite, firstly, carrying out 12-bit quantization on the multi-beam aliasing signals and extracting the maximum value of the amplitude of each path of signals; secondly, constructing a linear transformation matrix by using the obtained maximum value of the quantization amplitude to realize the compression of the dynamic range of the amplitude; and carrying out 6-bit secondary quantization on the compressed multi-path signals again and transmitting the signals through a Ka band. (2) At a ground station, firstly, acquiring a dictionary matrix under ideal signals by using a K-SVD method; and then, recovering the initial signal of the user by utilizing the received satellite data and a sparse reconstruction algorithm. The invention has the advantages that: the on-satellite forwarding cost is low, the compression effect is good, and the applicability is strong.

Description

Satellite load multi-beam sampling data compression method

Technical Field

The invention provides a compression method of satellite load multi-beam sampling data, and particularly relates to a bit number quantization method for compressing satellite multipath large-dynamic-range signals.

Background

In recent years, with the rapid development of high-performance satellite communication systems, the market demand of a quantization bit number compression method for loading multiple paths of large dynamic range signals on a satellite is more and more strong. The first 2 satellites of the maj MUOS system completed the launch task in 2014 and 2015 and were already put into operation. The forward direction of the load on the satellite is traditional transparent transmission, and the reverse direction adopts a method for reducing the quantization bit number of the multipath related signals, which is proposed by DavidK and Randall in 2006. The method firstly carries out 12-bit A/D conversion on multi-path parallel signals, then carries out decorrelation processing by using a complex scrambling code sequence, and carries out Hadamard linear transformation on the signals after the decorrelation, so that the multi-path signals after the transformation have equal average power. And selecting proper quantization bit numbers to carry out secondary quantization on the multipath signals with the same average power according to the performance requirement of the system. The method has the main defects that the information is damaged, and the on-satellite forwarding complexity is high.

In fact, Crowther et al proposed a method to reduce the number of quantization bits by applying Hadamard transform as early as 1967. The method utilizes the strong correlation characteristic among the multi-path input signals and the linear transformation characteristic of the Hadamard matrix to realize the compression of the dynamic range of the A/D input signals. The method is simple and convenient to implement, but has the main problems that the compression ratios of all paths of input signals are inconsistent, the number of quantization bits needs to be dynamically allocated, and the actual cost of the system is increased.

Based on research results of Crowther et al, Frangoulis and Turner further proposed Hadamard-Haar quantization bit compression method in 1977. After the Hadamard and Haar linear transformation are carried out in sequence, not only the dynamic range of the multi-path A/D input signal is compressed, but also the signal power is concentrated on a small part of Haar coefficients, which is a meaningful exploration for reducing the quantization bit number and the transmission rate. However, the information impairment of this method is further increased due to the Haar transform and the application of partial main coefficients.

Babarada et al proposed an Adaptive Gain (AGC) control method in 2011. The method is to add a self-adaptive control circuit before A/D sampling, and can compress the dynamic range of the A/D input signal without depending on the correlation and decorrelation operation among multiple paths of signals, thereby reducing the quantization bit number. The method has the main defects that the clipping phenomenon is serious in the adjusting process, the linear relation of signals is damaged, and the complexity is high, so that the on-satellite forwarding is not facilitated.

In addition, the commonly used methods for reducing the peak-to-average power ratio in the multi-carrier OFDM system, such as the direct amplitude limiting method, the amplitude limiting method with a peak window, the selective mapping method, the partial sequence transmission method, and the like, can also effectively compress the dynamic range of the input signal, but because the inverse process is mostly difficult to implement and usually requires additional sideband information to recover the original data, the methods are not suitable for transparent forwarding and the frequency band utilization rate is not high.

Disclosure of Invention

The invention aims to solve the technical problems that a dynamic range compression method suitable for spread spectrum signals and non-spread spectrum signals is researched aiming at the sampling data of a multi-beam satellite reverse link, and the series of difficult problems that the existing method has large information damage, low compression ratio, and the decorrelation operation is needed are overcome.

The technical scheme adopted by the invention is as follows:

the invention provides a compression method of satellite load multi-beam sampling data, which comprises the following steps:

in the on-satellite forwarding process:

(1) quantizing the satellite multi-beam aliasing signals to obtain quantized signals;

(2) extracting the maximum value of each row of elements of the quantized signal, and constructing a linear transformation matrix A after performing inversion and diagonalization in sequence;

(3) the amplitude dynamic range compression is realized in a matrix multiplication mode; multiplying the quantized signal by a linear transformation matrix to obtain a compression matrix;

(4) DA conversion is carried out on the compression matrix, and then the compression matrix is quantized and transmitted through a Ka waveband;

in the ground station receiving process:

(5) acquiring a dictionary matrix phi by using a dictionary learning method under an ideal noise-free signal;

(6) and recovering the received on-satellite forwarding data into the user initial signal by using a sparse reconstruction algorithm under the dictionary matrix phi.

Wherein, the step (5) comprises the following steps:

[501] partitioning the ideal noise-free signal;

[502] carrying out sparse coding on the partitioned ideal noiseless signals to construct an initial dictionary;

[503] and iteratively updating the initial dictionary by a dictionary learning method to obtain a dictionary matrix.

Wherein, step [503]When the iteration precision is smaller than a set threshold or the iteration times meet that C is larger than or equal to WQ, the iteration is stopped, wherein Q is the number of blocks divided by an ideal noiseless signal, and W is larger than 1 and smaller than 1

Is an integer of (1).

Wherein, the step (6) is specifically as follows:

[601] recovering the received on-satellite forwarding data into multi-path block data by utilizing a sparse reconstruction algorithm under a dictionary matrix phi;

[602] and recovering the multi-path block data into the user initial signal in a blocking way.

Wherein the step (1) wherein the multi-beam aliased signal is in the form of a spread spectrum signal or a non-spread spectrum signal in appearance; the modulation mode is a phase modulation signal, a frequency modulation signal or an amplitude modulation signal.

Wherein, the ideal noise-free signal in the step (4) is the same as the multi-beam aliasing signal in modulation mode.

And (3) in the step (5), the dictionary learning method is a K-SVD (K-singular value decomposition) method, an optimal dictionary learning method or a Fisher discriminant dictionary learning method.

Wherein, the sparse reconstruction algorithm in the step (6) is a matching pursuit algorithm, a convex optimization algorithm or an iterative non-convex algorithm with a fast convergence characteristic.

The invention has the following advantages:

(1) on-satellite forwarding does not need scrambling sequence decorrelation and Hadamard transformation steps, and complexity and forwarding cost are low;

(2) the ground station recovers the user signals by adopting a sparse reconstruction method, and has strong anti-interference capability and good compression effect;

(3) the method is suitable for spread spectrum signals and non-spread spectrum signals, and has good overall applicability.

Drawings

Fig. 1 is a general flow diagram illustrating an on-board load multi-beam sampled data compression method according to the present invention;

FIG. 2 is a diagram illustrating a process of constructing a linear transformation matrix A according to the present invention;

FIG. 3 is a graph showing a comparison of dynamic ranges before and after compression conversion of an on-board multipath spread spectrum received signal in accordance with the present invention;

FIG. 4 is a graph showing a comparison of dynamic ranges before and after a compression transform of a multi-path non-spread-spectrum received signal on a satellite according to the present invention;

FIG. 5 is a diagram illustrating a process of on-satellite quadratic quantization according to the present invention;

FIG. 6 is a flow chart illustrating obtaining a dictionary matrix Φ according to the K-SVD method employed by the present invention;

FIG. 7 is a flow chart illustrating recovery of a user's original signal using received data and a sparse reconstruction algorithm under a dictionary matrix Φ;

FIG. 8 is a graph showing results of a simulation experiment according to the present invention;

fig. 9 is a graph showing a result of a simulation experiment according to the present invention.

Detailed Description

A satellite load multi-beam sampling data compression method comprises the following steps:

fig. 1 is a general flow chart of a method for compression of satellite borne multi-beam sampled data according to the present invention. The method specifically comprises the following steps:

in the on-satellite forwarding process:

[101] and AD sampling is carried out on the M paths of beam aliasing signals X, and 12-bit quantization is completed to obtain quantized signals Xq. In the embodiment, the dimension M is more than or equal to 2, and the adopted quantization can be uniform quantization or non-uniform quantization, and is determined according to the system performance requirement;

[102] extracting the maximum value of each line of elements of Xq, and constructing a linear transformation matrix A after carrying out negation and diagonalization treatment in sequence;

[103] amplitude compression is realized by a matrix multiplication mode, namely Y is A × Xq;

[104] DA conversion is carried out on Y, and 6-bit secondary quantization processing is carried out;

[105] transmitting to a ground station through a Ka waveband link;

in the ground station receiving process:

[106] obtaining a dictionary matrix phi by using an ideal noise-free signal and a K-SVD method at a ground station;

[107] and recovering the initial signal of the user by utilizing the received data and a sparse reconstruction algorithm under the dictionary matrix phi.

The multi-beam aliasing signal X mentioned in the present invention may be a spread spectrum signal or a non-spread spectrum signal in the form of representation. The modulation method may be a phase modulation signal, a frequency modulation signal, or an amplitude modulation signal. Since the above signals are well known to those skilled in the art, they will not be described in too much detail.

The specific embodiment of the invention is as follows:

(1) quantizing the satellite multi-beam aliasing signal X to obtain a quantized signal Xq;

(2) extracting the maximum value of each line of elements of Xq, and constructing a linear transformation matrix A after carrying out negation and diagonalization treatment in sequence;

fig. 2 is a diagram showing a composition process of a linear transformation matrix a according to the present invention, including the steps of:

[201] extracting the maximum value of each line element of Xq to form a vector Max ═ Max1, Max2, … maxM ];

[202]and carrying out inversion operation. In this embodiment according to

The operation of taking the inverse is realized,

representing an upward rounding operation;

[203] the inverted data vector P is diagonalized, i.e., a ═ diag (P).

(3) Amplitude compression is realized by a matrix multiplication mode, namely Y is A × Xq;

fig. 3 and 4 are graphs showing dynamic range comparisons before and after compression conversion of on-board multi-spread and non-spread received signals, respectively, according to the present invention. In the specific embodiment provided by the present invention, the beam dimension is set to M-32, and there is a case where an even beam is aliased to an odd beam. In fig. 3, the difference between the maximum normalized average power value 301 and the small normalized average power value 302 is only 0.238. In fig. 4, the difference between the maximum normalized average power value 303 and the small normalized average power value 304 is only 0.162. Therefore, the satellite load multi-beam sampling data compression method provided by the invention can effectively compress the amplitude ranges of satellite multipath spread spectrum and non-spread spectrum receiving signals. In addition, as can be clearly seen from the comparison curve in the figure, the data compression performance of the method provided by the invention is obviously superior to that of the direct Hadamard transform compression method.

(4) And D/A conversion is carried out on Y, and then the Y is transmitted through a Ka waveband after 6-bit quantization.

Fig. 5 is a diagram illustrating a secondary quantization process on a satellite according to the present invention. The method comprises the following steps:

[401] DA conversion is effected for Y at a reference voltage V. In the present embodiment, the DA-converted reference voltage V is the same as the AD-converted voltage in step (1).

[402] And performing 6-bit secondary quantization on the DA-converted Y to obtain Y1. In this embodiment, the 6-bit secondary quantization may be uniform quantization or non-uniform quantization, which is determined according to the system performance requirement.

[403] And modulating the Y1 to a Ka wave band and then transmitting the Ka wave band to the ground station.

In the ground station receiving process:

(5) obtaining a dictionary matrix phi by using a K-SVD method under an ideal noise-free signal;

fig. 6 is a flow chart illustrating the acquisition of a dictionary matrix Φ according to the K-SVD method employed in the present invention, comprising the steps of:

[501] and partitioning the ideal noise-free signal. In a specific embodiment, an ideal noise-free signal of dimension N is divided evenly into Q, and 1< Q < N;

[502]in the embodiment, the dimension of the dictionary is Q × WQ, and W is more than 1 and less than 1

The specific value of W is determined according to the actual situation; in a QPSK signal dictionary construction of the present invention, W ═ 2 is a better choice.

[503] Updating the dictionary through iteration of a KSVD (K singular value decomposition) method;

[504]and outputting the dictionary phi. In this embodiment, when the accuracy is less than the threshold τ ≦ 10^-3Or when the iteration times C is more than or equal to WQ, terminating the iteration and outputting the dictionary phi.

It should be further noted that the dictionary matrix Φ mentioned in the present invention is not limited to be obtained by the K-SVD method, and may also be obtained by other dictionary learning methods such as the optimal dictionary learning method, the Fisher discriminant dictionary learning method, and the like. Since the dictionary learning method is well known to those skilled in the art, it is not described in detail.

(6) And recovering the initial signal of the user by utilizing the received data and a sparse reconstruction algorithm under the dictionary matrix phi.

FIG. 7 is a flow chart illustrating recovery of a user's original signal using received data and a sparse reconstruction algorithm under a dictionary matrix Φ; the specific implementation process comprises the following steps:

[601] and recovering the multi-path block data by utilizing the received data and a sparse reconstruction algorithm under the dictionary matrix phi. In specific embodiments, a matching pursuit algorithm (such as MP, OMP), a convex optimization algorithm (such as BP, BPDN) or an iterative non-convex algorithm with a fast convergence property may be used according to the complexity and accuracy requirements of the system.

[602] And recovering the reconstructed multi-path data in a block mode. In a specific embodiment, the reconstructed multi-path data is recovered in blocks according to a process reverse to the step [501], and an initial signal which is the same as or similar to that of the user sending end is finally output.

The effect of the method for compressing satellite load multi-beam sampling data is explained from the results of simulation experiments.

First, the effect of compressing the satellite multipath spread spectrum signals is explained: in the specific embodiment of the present invention, the experimental conditions are set as follows: the dimension of the wave beam is M-32, QPSK signals are transmitted through a Gaussian channel, the signal-to-noise ratio is changed to 9dB from-6 dB by taking 3dB as a step, an OVSF sequence with the spreading factor of 32 is adopted as a spreading sequence, the chip rate is 3.84MHz, the dynamic range of the wave beam is simulated by a random function randn (1, M), and the signals coupled to the wave beam by adjacent wave beams are attenuated by 30 dB. FIG. 8 is a graph showing the output SNR as a function of the input SNR for a spread spectrum signal according to the present invention. As can be seen, the method provided by the invention can provide reliable output while effectively compressing the quantization bit number.

Next, the effect of compressing the satellite multipath non-spread spectrum signals is explained: in the specific embodiment of the present invention, the experimental conditions are set as follows: the beam dimension is M-32, QPSK signals are transmitted through a Gaussian channel, the signal-to-noise ratio is changed from-6 dB to 9dB by taking 3dB as a step size, and the signal rate is 16 Kbaud. The dynamic range of the beam is modeled by a random function randn (1, M), and beam aliasing is not considered because the signal in a non-spread spectrum communication system is frequency band separable. Fig. 9 is a graph showing the output snr and the transformation with the input snr for the non-spread spectrum signal according to the method of the present invention. Similar to the spread spectrum situation, the method provided by the invention not only can effectively compress the quantization bit number, but also can provide better output.

The simulation experiment result fully shows that the dynamic range compression method for the satellite load multi-beam sampling data provided by the invention has good applicability.

Claims

1. A satellite load multi-beam sampling data compression method is characterized by comprising the following steps:

in the on-satellite forwarding process:

(2) extracting the maximum value of each row of elements of the quantized signal, and constructing a linear transformation matrix after performing negation and diagonalization treatment in sequence;

(3) the amplitude dynamic range compression is realized by a matrix multiplication mode, namely, a quantized signal is multiplied by a linear transformation matrix to obtain a compression matrix;

in the ground station receiving process:

(6) and recovering the received data forwarded on the satellite into the initial user signal by utilizing a sparse reconstruction algorithm under the dictionary matrix phi.

2. A satellite load multi-beam sampling data compression method according to claim 1, wherein the step (5) comprises the steps of:

[501] the received ideal noiseless signal is processed in a blocking mode;

3. A method for compression of satellite borne multi-beam sampled data according to claim 2, characterized by the step [503]When the iteration precision is smaller than a set threshold or the iteration times meet the condition that C is larger than or equal to WQ, terminating the iteration, wherein C is the iteration times, Q is the number of parts of partitioning an ideal noise-free signal, and W is larger than 1 and smaller than or equal to WQ

Is an integer of (1).

4. A satellite load multi-beam sampling data compression method according to claim 1, wherein the step (6) comprises the steps of:

[601] recovering received on-satellite forwarded data into multi-path block data by utilizing a sparse reconstruction algorithm under a dictionary matrix phi;

5. A method for compressing data according to claim 1, wherein the step (1) is that the multi-beam aliasing signal is in the form of spread spectrum signal or non-spread spectrum signal; the modulation mode is a phase modulation signal, a frequency modulation signal or an amplitude modulation signal.

6. The method according to claim 1, wherein the ideal noise-free signal in step (5) is identical to the multi-beam aliased signal in modulation.

7. The method according to claim 1, wherein the dictionary learning method in step (5) is K-SVD, optimal dictionary learning method or Fisher discriminant dictionary learning method.

8. A satellite load multi-beam sampling data compression method according to claim 1, wherein the sparse reconstruction algorithm in step (6) is a matching pursuit algorithm, a convex optimization algorithm or an iterative non-convex algorithm with fast convergence property.