CN104486778A - Signal system recognition method for heterogeneous networks - Google Patents

Abstract

The invention discloses a signal system recognition method for heterogeneous networks. Firstly, various signal systems of training sample signals are subjected to wavelet transform and fractional Fourier transform, frequencies corresponding to peak values in obtained transform results are counted, frequency regions are chosen as classification and distinguishing standards for the corresponding signal systems according to statistical results, and then a signal received by a recognition terminal is subjected to wavelet transform and fractional Fourier transform as well, so that frequency corresponding to a peak value in a transform result is obtained, the signal system where the frequency is located,, of the frequency region is judged, and thereby a recognition result is obtained. The signal system recognition method for the heterogeneous networks does not need to adopt prior information, is low in implementation complexity, and has a good recognition effect.

Description

A kind of signaling mode recognition methods of heterogeneous network

Technical field

The invention belongs to heterogeneous network technologies field, more specifically say, relate to a kind of signaling mode recognition methods of heterogeneous network.

Background technology

Heterogeneous network (Heterogeneous Network) is the network of a type, and it is the computer produced by different manufacturer, and the network equipment and system form, and operate in most cases in different agreements and support different functions or application.The target of heterogeneous network is to realize the seamless coverage of radio communication at macro network layout basis deploy low power nodes and little coverage node.In LTE network, heterogeneous network has been taken as the effective way improving the network coverage and capacity, can greatly improve LTE network capacity under the cost of relative efficiency.The low-power such as Femcell node, picocell node, via node and among a small circle node deployment covering cavity on increase network coverage rate with improve the availability of frequency spectrum.Meanwhile, the configuration among a small circle of low power nodes can use identical frequency spectrum to transmit greatly to improve reusability, therefore also can improve the capacity of radio communication in different range simultaneously.Wireless frequency spectrum environment of today configures the future development coexisted gradually to multiple network, network configuration is day by day complicated, and corresponding wireless frequency spectrum research difficulty is also increasingly loaded down with trivial details.For the process of cognitive radio networks or terminal, there are two subject matters:

A () cognition network or user carry out perception by its perception to wireless environment and utilize idle frequency spectrum to communicate.This just requires that cognitive terminal must possess efficient and reliable frequency spectrum perception technology;

How (b) cognitive terminal, under opportunistic access spectrum policy, solves cognitive terminal and does not observe rule of communication and disadvantageous " hostile terminal " problem.

Due to the dissimilar network be dispersed with in heterogeneous network, the process that cognition network and terminal can be made to carry out frequency spectrum perception analysis becomes more complicated.The network configured in wireless environment is made up of different transmission rates, through-put power and different signal covers, thus make current existing a lot of frequency spectrum sensing method and be not suitable for these heterogeneous network environments, when particularly facing the scene of low power nodes distribution, spectrum environment sensing efficiency can be lower.In this case, frequency spectrum perception is carried out and signaling mode identification is more difficult.Existing signaling mode recognition methods requires prior information mostly, and this mode is also uneconomical, and is difficult to obtain prior information accurately in a lot of situation.

Summary of the invention

The object of the invention is to overcome the deficiencies in the prior art, a kind of signaling mode recognition methods of heterogeneous network is provided, realize when directly identifying the signal received from heterogeneous network without any when prior information.

For achieving the above object, the signaling mode recognition methods of heterogeneous network of the present invention, comprises the following steps:

S1: the training sample signal building all signaling modes in heterogeneous network, each training sample comprises one group of signal, for each signaling mode training sample, wavelet transformation is carried out to wherein each signal, and then carry out fractional fourier transform, the frequency corresponding to the peak value in the transformation results obtained is added up, and selects frequency separation as the discriminant classification standard of respective signal standard according to statistics;

S2: the signal of reception is carried out wavelet transformation and fractional fourier transform by cognitive terminal, obtains the frequency that peak value in transformation results is corresponding, as differentiation feature;

S3: if the frequency that step S2 obtains belongs to the frequency separation of certain signaling mode, then using this signaling mode as recognition result.

The signaling mode recognition methods of heterogeneous network of the present invention, first by the training sample signal of various types of signal standard through wavelet transformation and fractional fourier transform, the frequency corresponding to the peak value in the transformation results obtained is added up, select frequency separation as the discriminant classification standard of respective signal standard according to statistics, then the signal of reception is carried out wavelet transformation and fractional fourier transform by cognitive terminal equally, obtain the frequency that peak value in transformation results is corresponding, judge that the frequency separation which kind of signaling mode is this frequency be positioned at can obtain recognition result.The present invention without the need to adopting prior information, but carries out Direct Recognition by the signal characteristic after wavelet transformation and fractional fourier transform, and implementation complexity is low and have good recognition effect.

Accompanying drawing explanation

Fig. 1 is the embodiment flow chart of the signaling mode recognition methods of heterogeneous network of the present invention;

Fig. 2 is the result schematic diagram of three class signals after wavelet transformation and fractional fourier transform in the present embodiment;

Fig. 3 is the characteristic statistics schematic diagram of three class signal training samples under being 0dB situation in signal to noise ratio in the present embodiment;

Fig. 4 is the identification correct probability figure under awgn channel;

Fig. 5 is the identification correct probability figure on the basis of AWGN noise after Rayleigh is weak;

Fig. 6 is the recognition effect figure of the different path delay of time (path delay) under the weak channel of Rayleigh;

Fig. 7 is the identification correct probability figure in mixed signal situation.

Embodiment

Below in conjunction with accompanying drawing, the specific embodiment of the present invention is described, so that those skilled in the art understands the present invention better.Requiring particular attention is that, in the following description, when perhaps the detailed description of known function and design can desalinate main contents of the present invention, these are described in and will be left in the basket here.

Embodiment

Fig. 1 is the embodiment flow chart of the signaling mode recognition methods of heterogeneous network of the present invention.As shown in Figure 1, the signaling mode recognition methods of heterogeneous network of the present invention, comprises the following steps:

S101: obtain discriminant classification standard:

In the present invention, discriminant classification standard is obtained by training, concrete grammar is: the training sample signal building all signaling modes in heterogeneous network, each training sample comprises one group of signal, for each signaling mode training sample, wavelet transformation is carried out to wherein each signal, and then carries out fractional fourier transform, the frequency corresponding to the peak value in the transformation results obtained is added up, and statistical probability is greater than the discriminant classification standard of frequency separation as respective signal standard of predetermined threshold value.

In the present embodiment, be described, comprise 2FSK, BPSK, 16QAM for the signal of three kinds of signaling modes, its expression formula is respectively:

$S_{\mathrm{FSK}}\left(t\right)={\mathrm{AR}}_e\left[\underset{k}{\mathrm{\Σ}}{e}^{j2\mathrm{\π}{f}_c+f_{k}t}g(t-k{T}_s)\right]---\left(1\right)$

In formula (1) $f_k=[i-\frac{M-1}{2}]\mathrm{\Δf},i=\mathrm{0,1},...,M-1.$

$S_{\mathrm{PSK}}\left(t\right)={\mathrm{AR}}_e\left[\underset{k}{\mathrm{\Σ}}{m}_k{e}^{j2\mathrm{\π}{f}_c+f_{k}t}g(t-k{T}_s)\right]---\left(2\right)$

In formula (2) $m_k=\left[e^{j\frac{M-1}{2}}\right],i=\mathrm{0,1},...M-1.$

$S_{\mathrm{QAM}}\left(t\right)={\mathrm{AR}}_e\left[\underset{k}{\mathrm{\Σ}}{m}_k{e}^{j2\mathrm{\π}{f}_c+f_{k}t}g(t-k{T}_s)\right]---\left(3\right)$

M in formula _k=a _k+ jb _k, a _k, b _k=2i-M+1, i=i=0,1 ... M-1.

(1), (2), in (3), A represents amplitude factor, is determined by received signal power; f _k, m _krepresent symbol rate; f _crepresent carrier frequency; T _srepresent mark space; M represents number of modulation levels, k=1,2 ..., M; Δ f represents frequency difference; R _ereal part is asked in [] expression; G (t) represents T _sunit pulse in time.

Reasonable effect can be obtained owing to carrying out high-order time frequency analysis to non-stationary signal, in the present invention, first carry out wavelet transformation to received signal, utilize wavelet transformation to extract the advantage of Characteristics of Mutation, obtain the Characteristics of Mutation of Received signal strength.Consider again the treatment effect that Fourier transform pairs stationary signal is good simultaneously, the result through wavelet transformation again through fractional fourier transform, the time-frequency characteristic of original signal and Characteristics of Mutation are better embodied.

In the present embodiment, the wavelet basis of wavelet transformation adopts the wavelet basis of Daubechies5, and its expression formula is:

${\mathrm{\ψ}}_{(a,b)}\left(t\right)=\frac{1}{\sqrt{\left|a\right|}}\mathrm{\ψ}\left(\frac{t-b}{a}\right)---\left(4\right)$

Wherein, ψ _a,bt () represents wavelet basis, a, b are respectively zoom factor and shift factor, and R is real number field, and ψ () represents wavelet mother function.

Its Fourier transform is expressed as:

${\mathrm{\ψ}}_{(a,b)}\left(t\right)=\frac{1}{\sqrt{\left|a\right|}}\underset{-\∞}{\overset{\∞}{\∫}}\mathrm{\ψ}\left(\frac{t-b}{a}\right)e^{-\mathrm{j\ωt}}\mathrm{dt}=-\frac{a}{\sqrt{\left|a\right|}}{e}^{-\mathrm{jb\ω}}\mathrm{\ψ}\left(\mathrm{a\ω}\right)--\left(5\right)$

Vector discrete Fourier transform be,

$\stackrel{\→}{d}=W\stackrel{\→}{g}---\left(6\right)$

Wherein, W represents Orthogonal Wavelets,

Fractional fourier transform is defined as:

$\left(F_s^{\mathrm{\α}}f\right)\left(x\right)=\underset{m=0}{\overset{+\∞}{\mathrm{\Σ}}}{B}_m{\mathrm{\λ}}_m(s;\mathrm{\α}){\mathrm{\Φ}}_m\left(x\right)---\left(7\right)$

In formula $f\left(x\right)=\underset{m=0}{\overset{+\∞}{\mathrm{\Σ}}}{B}_m{\mathrm{\Φ}}_m\left(x\right),$

The fractional fourier transform of cycle signal x (t) is:

$x_a\left(u\right)=P_c{\hat{x}}_{\mathrm{\β}}\left(u\frac{{\mathrm{\σ}}_{\mathrm{\β}}}{{\mathrm{\σ}}_{\mathrm{\α}}}\right)---\left(8\right)$

In formula $\mathrm{\β}={\tan }^{-1}(\tan \mathrm{\α}/{\mathrm{\σ}}^2),$

$P_c=\sqrt{{\mathrm{\σ}}_{\mathrm{\β}}}\sqrt{\frac{1-j\cot \mathrm{\α}}{1-j{\mathrm{\σ}}^2\cot \mathrm{\α}}}\·\exp \left(j\frac{u^2}{2}\cot \mathrm{\α}(1-\frac{{\cos }^2\mathrm{\β}}{{\cos }^2\mathrm{\α}})\right).$

Fractional fourier transform has two advantages, and one is fractional fourier transform is the generalization of Fourier transform, and theoretical property is strong and flexibility ratio is high; Two is be easy to realize.

Fig. 2 is the result schematic diagram of three class signals after wavelet transformation and fractional fourier transform in the present embodiment.As shown in Figure 2, after wavelet transformation and fractional fourier transform, the feature (distribution of amplitude peak in frequency) of 2FSK, BPSK, 16QAM tri-every class signal in class signal is all different, and these features can be used for carrying out the classification of signal.The present invention is by obtaining discrimination standard to the characteristic statistics of training sample handshaking result.

Fig. 3 is the characteristic statistics schematic diagram of three class signal training samples under being 0dB situation in signal to noise ratio in the present embodiment.As shown in Figure 3, the range value of 2FSK signal sampling point is distributed in 50-150 and 1150-1175 place, and the amplitude peak of bpsk signal sampled point is distributed in whole interval, 16QAM signal sampling point amplitude peak be distributed in be less than 50 or 600-650 interval in.Frequency separation be set according to the distribution of the amplitude peak in this figure in frequency and three class signals are classified, being greater than the frequency separation of predetermined threshold value by statistical probability as discriminant classification standard.The minimum frequency that general employing statistical probability is greater than predetermined threshold value is interval, and namely all statistical probabilities are greater than the frequency separation that in the frequency separation of predetermined threshold value, width is minimum.Environment according to heterogeneous network determines, when network environment is good, frequency separation can arrange relatively narrower, and when network environment is bad, frequency separation setting is relatively wider.

S102: extract signal characteristic:

Cognitive terminal Received signal strength, carries out wavelet transformation and fractional fourier transform, obtains the frequency that peak value in transformation results is corresponding, as differentiation feature.

S103: signaling mode identification:

If the frequency that step S102 obtains belongs to the frequency separation of certain signaling mode, then using this signaling mode as recognition result.

In order to beneficial effect of the present invention is described, under Matlab environment, carries out the performance of emulation experiment to signaling mode recognition methods of the present invention carry out compliance test result and assessment.To 2FSK, BPSK, 16QAM tri-class signal carry out simulation results show analysis, simulation result and performance evaluation are divided into four aspect: AWGN (Additive White Gaussian Noise, additive white Gaussian noise) single-signal identification under channel, under the weak lower single-signal identification of Rayleigh, the weak environment of Rayleigh, propagation delay time is on the identification of mixed signal under the impact of Signal analysis, awgn channel.In simulating, verifying, suppose that signal transmission rate is 40Kbps, sampling frequency is 800KHz, and carrier frequency is 100KHz, produces 100 symbols and ask recognition accuracy as signal to be identified under each signal to noise ratio, and training sample sequence length is 10000.

Fig. 4 is the identification correct probability figure under awgn channel.Bpsk signal correct recognition rata under signal to noise ratio is-9dB is greater than 90%.When the signal to noise ratio of signal equals 0dB, the probability that three class signals are correctly validated all is greater than 96%.

Because confusion matrix can describe the identification probability of a kind of signal type and other signal clearly, be used for stating Signal analysis problem so be often taken as a kind of visual instrument.Table 1 is the identification probability of three class signals under signal to noise ratio is-5dB.As can be seen from Table 1, algorithm can obtain good recognition effect to bpsk signal and 16QAM signal.

Table 1

Fig. 5 is the identification correct probability figure on the basis of AWGN noise after Rayleigh is weak.This emulation down-sampling input interval is 10 ^-5, the path delay of time is 10 ^-6, maximum doppler frequency is 130Hz.As shown in Figure 5, recognition accuracy comparatively awgn channel compares some decline, but recognition correct rate still can reach good level on the whole.Table 2 is when signal to noise ratio is for-5dB, the identity confusion matrix of signal after Rayleigh is weak.As table 2 can find out the recognition effect that the present invention can reach good.

Table 2

Fig. 6 is the recognition effect figure of the different path delay of time (path delay) under the weak channel of Rayleigh.As shown in Figure 6, time delay is less, and recognition efficiency is higher.The present invention is less than 10 in time delay ^-5good recognition efficiency can be reached.

In the communication environment of heterogeneous network, in spectrum environment, often there is polytype signal.Being identified in real world applications of mixed signal occupies very important status.Therefore the recognition correct rate of mixed signal under this algorithm is also demonstrated herein.Suppose only there is AWGN noise in spectrum environment in this simulation process, and having two primary user's signals (two kinds in three kinds of signals) and an interference signal (remaining a kind of signal), Fig. 7 is the identification correct probability figure in mixed signal situation.As shown in Figure 7, when only having two signals to exist, recognition effect well, and when signal to noise ratio is 5dB, the recognition correct rate of signal is more than 80%.When the poor effect of carrying out identifying after three class signal mixing.

Known according to above emulation, the present invention, when carrying out mono signal identification and mixed signal identification, can reach good effect.

Although be described the illustrative embodiment of the present invention above; so that those skilled in the art understand the present invention; but should be clear; the invention is not restricted to the scope of embodiment; to those skilled in the art; as long as various change to limit and in the spirit and scope of the present invention determined, these changes are apparent, and all innovation and creation utilizing the present invention to conceive are all at the row of protection in appended claim.

Claims (2)

1. a signaling mode recognition methods for heterogeneous network, is characterized in that, comprise the following steps:

S1: the training sample signal building all signaling modes in heterogeneous network, each training sample comprises one group of signal, for each signaling mode training sample, wavelet transformation is carried out to each signal wherein, and then carry out fractional fourier transform, the frequency corresponding to the peak value in the transformation results obtained is added up, and selects frequency separation as the discriminant classification standard of respective signal standard according to statistics;

S2: the signal of reception is carried out wavelet transformation and fractional fourier transform by cognitive terminal, obtains the frequency that peak value in transformation results is corresponding, as differentiation feature;

S3: if the frequency that step S2 obtains belongs to the frequency separation of certain signaling mode, then using this signaling mode as recognition result.

2. signaling mode recognition methods according to claim 1, is characterized in that, the wavelet basis of described wavelet transformation adopts the wavelet basis of Daubechies5.