CN114580470B

CN114580470B - OFDM and UFDM multi-carrier signal identification method based on non-uniform quantization map features

Info

Publication number: CN114580470B
Application number: CN202210168667.7A
Authority: CN
Inventors: 胡国兵; 李鹏; 吴珊珊; 赵嫔姣; 杨莉; 罗荣华
Original assignee: Jinling Institute of Technology
Current assignee: Jinling Institute of Technology
Priority date: 2022-02-21
Filing date: 2022-02-21
Publication date: 2024-09-06
Anticipated expiration: 2042-02-21
Also published as: CN114580470A

Abstract

Aiming at the problem that the time-frequency domain similarity of two multi-carrier signals of OFDM and UFMC is high and is difficult to distinguish by the traditional method, the invention provides an OFDM and UFMC multi-carrier signal identification method based on the non-uniform quantization map feature. The method comprises three steps: firstly, performing DFT (discrete Fourier transform) on a power spectrum of a signal to be identified to obtain a secondary spectrum of the signal; then, converting the secondary spectrum of the signal into a map domain through nonuniform quantization; and finally, detecting the complete connectivity of the diagram, if the diagram is the complete diagram, judging the observation signal as UFMC, otherwise, judging the observation signal as OFDM. The invention avoids the dependence of the identification method on training samples and can realize the effective identification of the multicarrier signals with slight differences.

Description

OFDM and UFDM multi-carrier signal identification method based on non-uniform quantization map features

Technical Field

The invention belongs to the field of signal identification and processing, and particularly relates to an OFDM and UFDM multi-carrier signal identification method based on non-uniform quantization map features.

Background

The multi-carrier technology is one of key technologies of the physical layer of the new generation mobile communication system. The recognition of the multi-carrier signals is one of the main tasks in intelligent communication such as electronic reconnaissance and cognitive radio, and is also an important precondition for the recognition and demodulation processing of subsequent information modulation. However, because the waveform difference between the time domain and the frequency domain of some multicarrier signals is very small, and the influence of noise, the characteristic separability based on the Euclidean space is poor, and great difficulty is brought to identification. However, the existing recognition method based on deep learning requires a large number of identified training samples, and when the signal-to-noise ratio varies in a large dynamic range, the samples need to be subjected to accurate signal-to-noise ratio estimation in advance to be used for training, so that the samples can be paired with the test samples under the corresponding signal-to-noise ratios in the later period for recognition. If under the uncooperative condition of electronic reconnaissance, the marked training sample set is not easy to acquire in reality under the signal to noise ratio of a large dynamic range due to the difficulty of signal interception. Therefore, new signal representation and feature extraction methods need to be studied to accommodate the identification requirements of such signals.

The developed domain signal processing method in recent years provides a new idea for solving the problems. The basic idea is to transform signal samples (time domain, frequency domain or transform domain) into a specific graph through uniform quantization, and the vertices and edges of the graph are determined by the relation of sampling points in the signal domain before transformation and the mapping rule. At present, related domain signal processing is applied to signal detection and classification, and mainly signal detection is realized based on complete graph characteristics. That is to say, in terms of the signal classification problem, the time-domain form of the signal to be identified or its frequency spectrum has to satisfy the following constraints: one class must exhibit the same-distribution noise characteristic, and the count value of all vertexes in the histogram of the uniformly quantized sample is greater than 0; the other class must have non-uniformly distributed noise characteristics, wherein the count value of at least one vertex in the histogram of the uniformly quantized samples is 0. This requirement limits the range of applications of the full graph method in signal recognition, and in some cases it is difficult to ensure that statistics of the signal to be recognized meet the above conditions. Therefore, it is necessary to combine the characteristics of the signal and the characteristics of the domain processing method to make necessary corrections, and to relax the restrictions of the above conditions, thereby expanding the application range of such methods.

Disclosure of Invention

Aiming at the defects in the prior art, the invention provides an OFDM and UFDM multi-carrier signal identification method based on non-uniform quantization map features. According to the method, the uniform quantization mode in the domain transformation is changed into non-uniform quantization according to the nuances of the secondary spectrums of the OFDM and the UFMC, the identification problem is skillfully converted into complete graph detection, and the low-complexity characteristic of the average degree of the graph is utilized as the detection characteristic to finish the identification of two types of signals.

In order to achieve the above purpose, the present invention adopts the following technical scheme:

The OFDM and UFDM multi-carrier signal identification method based on the non-uniform quantization map features is characterized by comprising the following steps:

Step 1: performing DFT conversion on the power spectrum of the signal to be identified, and taking the modulus value of the power spectrum to obtain a secondary spectrum of the signal;

Step 2: converting the secondary spectrum of the signal into a map domain in a non-uniform quantization mode to form a map;

Step 3: and detecting the complete connectivity of the diagram, if the diagram is the complete diagram, judging the diagram as the UFMC signal, and if the diagram is the complete diagram, judging the diagram as the OFDM signal.

In order to optimize the technical scheme, the specific measures adopted further comprise:

further, in the step 1, the signal to be identified is r (k),

r(k)＝s(k)+n(k)，0≤k≤K-1

Its secondary spectrum is

R(m)＝|DFT{|DFT[r(k)]|²}|，0≤m≤K-1

Where s (K) represents a multicarrier signal, n (K) represents additive white gaussian noise, K represents a time series value, K represents a number of signal samples, R (K) is a noise-contaminated time-domain signal, R (m) is a frequency-domain version thereof, and m represents a subscript of a discrete frequency.

Further, in the step 2, the secondary spectrum R (m) of the signal is normalized to obtain a normalized spectrum U (m):

let the quantization level number be N, carry on the non-uniform quantization to the normalized frequency spectrum:

Wherein Q (m) represents a non-uniformly quantized spectrum;

Finally, Q (m) is converted to the graph domain, constituting a graph G (V, E), the vertex set V of which represents a mapping of quantization levels {1,2,..n }, v= { V ₁,v₂,...v_N }; the edge set e= { E _α,β|ν_α∈V,v_β∈V},e_α,β of the graph represents an edge between two vertices of the graph.

Further, the specific implementation of the composition graph G (V, E) is: traversing the level relation between each quantized sample Q (m) and Q (m+1) one by one, and connecting two vertexes when level jumps from v _α to v _β exist, wherein e _α,β = 1; otherwise, two vertices are not connected, e _α,β =0.

Further, in the step 3, the recognition of the two types of signals of OFDM and UFDM is converted into a complete graph detection of the graph G (V, E) obtained in the step 2, where the complete graph detection depends on the verification of the average degree characteristic of the graph G (V, E).

Further, the average degree characteristic of the graph is checked, specifically as follows:

The degree vector of the graph G (V, E) is d, the average degree is defined as κ=average (d), if κ=n-1, G (V, E) is the complete graph, and the signal decision is UFMC signal; otherwise, G (V, E) is an incomplete graph, and the signal decision is an OFDM signal.

The beneficial effects of the invention are as follows: the invention recognizes the secondary spectrum of OFDM and UFMC signals by means of non-uniform quantized graph feature transformation, converts the multi-carrier signals with extremely small difference between the characteristics of two types of time domains and frequency domains into the detection of the complete connectivity of the graph generated by the secondary spectrum, and defines the low-complexity complete graph detection feature based on average degree. Compared with the existing algorithm, the method changes uniform quantization into non-uniform quantization, provides possibility for graph domain identification modeling of OFDM and UFMC signals to complete graph identification, changes traditional high-complexity complete graph detection characteristics based on Laplacian eigenvalues into graph-based average degree, avoids characteristic analysis of a high-dimensional matrix, reduces complexity of identification processing, and has better application prospect under non-cooperative conditions.

Drawings

FIG. 1 is a flow chart of an identification method of the present invention.

Fig. 2a to 2d are signal envelope, power spectrum, secondary spectrum and generation diagram of the OFDM signal, respectively.

Fig. 3a to 3d are signal envelopes, power spectrums, secondary spectrums, and generation diagrams of UFMC signals, respectively.

Fig. 4 is a schematic diagram of average recognition accuracy of the recognition method according to the present invention.

Detailed Description

The invention will now be described in further detail with reference to the accompanying drawings.

In a multi-carrier signal, the time-frequency domain forms of the two signals of the OFDM and the UFMC are very similar, and are difficult to effectively distinguish by using generalized characteristics (such as time domain and frequency domain), especially in the case of strong noise. And the characteristics of other types of multi-carrier signals are greatly different and are easy to identify. Therefore, in the recognition of the multi-carrier signal, after the other multi-carrier signals are recognized, two signals, i.e. OFDM and UFMC, are often left to be difficult to recognize. For this reason, the present invention proposes an OFDM and UFDM multi-carrier signal identification method based on the non-uniform quantization map feature as shown in fig. 1, which specifically includes the following matters.

1. Acquisition of secondary spectra

The multi-carrier signal r (k) to be identified can be expressed as:

r(k)＝s(k)+n(k)，0≤k≤K-1

where s (K) represents a multicarrier signal, n (K) is additive white gaussian noise, K represents a time-series value, and K represents the number of signal samples. Consider here two types of multicarrier signals, OFDM and UFMC, whose secondary spectrum is:

R(m)＝|DFT{|DFT[r(k)]|²}|，0≤m≤K-1

2. non-uniform quantization based quadratic spectrogram domain transformation

1. Normalization: normalizing the secondary spectrum R (m) to obtain a normalized spectrum thereof:

2. non-uniform quantization: let the quantization level number be N, carry on the non-uniform quantization to the normalized frequency spectrum:

3. Map mapping: converting Q (m) to a graph domain, constituting a graph G (V, E), wherein a set of vertices V of the graph represent a mapping of quantization levels {1,2,..n }, v= { V ₁,v₂,...v_N }; the edge set e= { E _α,β|ν_α∈V,v_β∈V},e_α,β of the graph represents an edge between two vertices of the graph; the specific practice of constructing the graph G (V, E) is: traversing the level relation between each quantized sample Q (m) and Q (m+1) one by one, and connecting two vertexes when level jumps from v _α to v _β exist, wherein e _α,β = 1; otherwise, two vertices are not connected, e _α,β =0.

3. Identification based on complete graph detection:

1. feature definition: a degree vector of fig. G (V, E) is d, and an average degree of κ=average (d) is defined as an identification feature quantity;

2. identifying a rule: if the average degree kappa=n-1 of the graph, G (V, E) is the complete graph, and the signal decision is UFMC signal; otherwise, G (V, E) is an incomplete graph, and the signal decision is an OFDM signal.

Fig. 2a-2d and fig. 3a-3d are the envelope, power spectrum, secondary spectrum waveform and generation diagrams of the OFDM signal and UFMC signal, respectively. Signal to noise ratio in simulation of 10dB, the parameters of OFDM are: the number of signal blocks is 50, the number of subcarriers in each signal block is 12, the cyclic prefix length is 72 points, the modulation mode is 16QAM, and the filter length is 513 points. The side lobe attenuation of the UFMC signal is 40, the filter adopts Dolph-Chebyshev, and the filter order is 43. The graph shows that the time domain waveforms of the two types of signals are random, the change rule of the power spectrum is not great, the main lobe of the secondary spectrum has slight difference, the graph structure generated after nonuniform quantization has obvious difference, the generated graph of the OFDM signal is an incomplete graph, and the UFMC is a complete graph.

Fig. 4 is a graph of average recognition accuracy versus signal-to-noise ratio for the present recognition method. The change range of the signal to noise ratio in the simulation is 0 to 10dB, the step length is 2dB, each signal to noise ratio is simulated 1000 times, two signals are respectively identified, and finally the average value of the performance is taken. As can be seen from the graph, when the signal-to-noise ratio is greater than 4dB, the average recognition accuracy can reach 100%.

It should be noted that the terms like "upper", "lower", "left", "right", "front", "rear", and the like are also used for descriptive purposes only and are not intended to limit the scope of the invention in which the invention may be practiced, but rather the relative relationship of the terms may be altered or modified without materially altering the teachings of the invention.

The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above examples, and all technical solutions belonging to the concept of the present invention belong to the protection scope of the present invention. It should be noted that modifications and adaptations to the invention without departing from the principles thereof are intended to be within the scope of the invention as set forth in the following claims.

Claims

1. The OFDM and UFDM multi-carrier signal identification method based on the non-uniform quantization map features is characterized by comprising the following steps:

Step 2: converting the secondary spectrum of the signal into a map domain in a non-uniform quantization mode to form a map; in the step 2, the secondary spectrum R (m) of the signal is normalized to obtain a normalized spectrum U (m):

Wherein Q (m) represents a non-uniformly quantized spectrum;

Finally, Q (m) is converted to the graph domain, constituting a graph G (V, E), the vertex set V of which represents a mapping of quantization levels {1,2,..n }, v= { V ₁,v₂,...v_N }; the edge set e= { E _α,β|ν_α∈V,ν_β∈V},e_α,β of the graph represents an edge between two vertices of the graph;

2. The method for identifying the OFDM and UFDM multicarrier signals based on the non-uniform quantization map feature as claimed in claim 1, wherein: in the step 1, the signal to be identified is r (k),

r(k)＝s(k)+n(k),0≤k≤K-1

Its secondary spectrum is

R(m)＝|DFT{|DFT[r(k)]|²}|,0≤m≤K-1

Where s (K) represents one of two multi-carrier signals, where the specific type of the multi-carrier signal is unknown before identification, n (K) represents additive white gaussian noise, K represents a value of a time sequence, K represents a number of signal samples, R (K) is a time domain signal polluted by noise, R (m) is a frequency domain form thereof, and m represents a subscript of a discrete frequency.

3. The method for identifying the OFDM and UFDM multicarrier signals based on the non-uniform quantization map feature as claimed in claim 1, wherein: the specific implementation of the composition diagram G (V, E) is as follows: traversing the level relation between each quantized sample Q (m) and Q (m+1) one by one, and connecting two vertexes when level jumps from v _α to v _β exist, wherein e _α,β = 1; otherwise, two vertices are not connected, e _α,β =0.

4. The method for identifying the OFDM and UFDM multicarrier signals based on the non-uniform quantization map feature as claimed in claim 1, wherein: in the step 3, the recognition of the two signals of the OFDM and the UFDM is converted into a complete graph detection of the graph G (V, E) obtained in the step 2, wherein the complete graph detection depends on the verification of the average degree characteristic of the graph G (V, E).

5. The method for identifying the OFDM and UFDM multicarrier signals based on the non-uniform quantization map feature according to claim 4, wherein: the average degree characteristic of the graph is checked, and the average degree characteristic of the graph is specifically as follows: