CN103856421B - The mpsk signal unbound nucleus signal-noise ratio estimation method that Open-closed-loop combines - Google Patents

Abstract

The invention discloses the mpsk signal unbound nucleus signal-noise ratio estimation method that a kind of Open-closed-loop combines.The method first passes through the two-way orthogonal signalling received and obtains envelope second-order statistics information and first-order statistics information；That receive the second-order statistics information sum of data and first-order statistics information followed by one section and square carry out counting statistics value of consult volume；Then according to the mapping relations one by one between statistical parameter value and snr value, signal to noise ratio is estimated；Finally by a closed loop filter, signal-to-noise ratio (SNR) estimation value is carried out smothing filtering.The present invention need not any assistance data, it is not required that carries out carrier auxiliary in advance；Can quickly obtain signal-to-noise ratio (SNR) estimation value, and estimated accuracy is high, estimated bias is little；Take into full account the requirement of data-measuring when hardware realizes, and hardware implementation complexity is the lowest.The present invention is used in the communication system using the constant enveloped modulation modes such as MPSK.

Description

The mpsk signal unbound nucleus signal-noise ratio estimation method that Open-closed-loop combines

Technical field

The invention belongs to channel estimation technique, the mpsk signal non-data that a kind of Open-closed-loop combines is auxiliary Help signal-noise ratio estimation method.

Background technology

The channel information provided by signal-to-noise ratio (SNR) estimation, can be effectively improved channel transport efficiency, is constantly pursuing High performance today, signal-to-noise ratio (SNR) estimation technology has a wide range of applications.Such as, Tubo code, LDPC code are contour The channel coding method of performance is required for utilizing signal-tonoise information to carry out additional interpretations, and adaptive coding modulation is also Need the foundation that signal-tonoise information selects as coded modulation scheme, this external power control, adaptive equalization with And the technical field such as Dynamic Resource Allocation for Multimedia is also required for utilizing signal-tonoise information to improve performance.

In terms of signal-to-noise ratio (SNR) estimation, forefathers have been carried out numerous studies, and these are studied and can be divided into two big classes: base In data-aided method (DA) and method based on unbound nucleus (NDA).Data-aided method requirement Insert pilot signal or training sequence known to receiving terminal in a stream.Obviously, adopt and can drop in this way Low system transfers efficiency.In a lot of application scenarios, consider from the angle improving spectrum efficiency, unbound nucleus Method is a kind of preferably selection.

According to estimate time the need of knowing carrier parameter, unbound nucleus SNR estimator again can be further It is divided into photocarrier method (CA) and the big class of non-carrier aided algorithm two.Photocarrier method algorithm also has a variety of: online SNR estimator, the signal-to-noise ratio estimation algorithm of iteration.The SNR ranges that online SNR estimator is suitable for (is estimated Meter precision is interval close to the signal to noise ratio of theoretical limit) it is 0 to 6dB, its applicable SNR ranges can be extended To-5 to 12dB.The signal-to-noise ratio estimation algorithm of iteration can reduce the estimated bias when low signal-to-noise ratio further, But the complexity of this algorithm is the highest.But photocarrier method algorithm has a common problem, i.e. by carrier parameter Estimated accuracy impact is bigger, it is necessary to could use after realizing accurate carrier recovery.For overcoming this problem, can To estimate signal to noise ratio according to reception signal envelope, owing to this method is unrelated with carrier parameter, thus the most non- Photocarrier method signal-to-noise ratio estimation algorithm.Non-carrier aided algorithm have second order Fourth-order moment (M2M4) algorithm and single order and Second order estimates SNR method.Wherein single order and second order estimate that the method for signal to noise ratio is than M2M4 arithmetic accuracy Height, when low signal-to-noise ratio, estimated bias is less, but the method has relatively when the symbolic number for estimating is less Big estimated bias.

P.Gao,C.Tepedelenlioglu,“SNR estimation for nonconstant modulus Constellations, " (IEEE Trans.Signal Processing, vol.53, no.3, Pp.865-870,2005) literary composition proposes a kind of single order utilizing signal envelope and second order to estimate signal to noise ratio Method, this method needs to utilize look-up table to realize.This method is higher than M2M4 arithmetic accuracy, at low letter Make an uproar than time estimated bias less.But actual application is limited by look-up table size, during high s/n ratio, is estimated performance Can increase along with signal to noise ratio and decline.Chao Gong,Bangning Zhang,Aijun Liu,Daoxing Guo, " A High Accurate and Low Bias SNR Estimator:Algorithm and Implementation ", (Radioegineering, 2011, Vol.12) literary composition is to proposing a kind of M1M2 The correction algorithm of method, and use the method for segmentation third degree curve matching to replace LUT Method.This side But method precision higher hardware implementation complexity is bigger.Both approaches is all open loop method of estimation, is suitable for The signal-to-noise ratio (SNR) estimation of short burst signal.And continuous signal open loop method of estimation is not being made full use of With, estimated accuracy needs to be improved further.

Summary of the invention

It is an object of the invention to provide the mpsk signal unbound nucleus signal-to-noise ratio (SNR) estimation that a kind of Open-closed-loop combines Method, this method speed is fast, precision is high, hardware implementation complexity is low.

The technical solution realizing the object of the invention is: the mpsk signal non-data that a kind of Open-closed-loop combines is auxiliary Helping signal-noise ratio estimation method, step is as follows:

The first step, envelope information obtains, i.e. according to two-way orthogonal signalling r received_InAnd r_Qn, determine that it wraps Network second order information u_n, then recycle u_nDetermine envelope first moment information v_n；

Second step, statistical parameter calculates, i.e. one section N number of first moment information v receiving signal_nSum square M₁, and second moment information u_nSum M₂, pass through M₁And M₂Determine statistical parameter γ；

3rd step, signal-to-noise ratio (SNR) estimation, i.e. utilize the mapping relations one by one between the λ of statistical parameter γ and signal to noise ratio, According to γ-value, signal to noise ratio λ is estimated；

4th step, closed loop smothing filtering, i.e. utilize a second-order loop filter to above using open-loop method Obtain signal to noise ratio to be worth according to a preliminary estimate and smooth, improve estimated accuracy further, the letter obtained after smothing filtering Compared estimate value of making an uproar is as final output result ρ.

In the first step, envelope second moment information u_nFormula is:

$u_n=\mathrm{floor}((r_{\mathrm{In}}^2+r_{\mathrm{Qn}}^2)/2^q)$

Wherein q is for receiving data-measuring figure place.

In the first step, envelope second moment information v_nFormula is:

$v_n=\mathrm{round}(\sqrt{u_n}\×2^{q/2})$

Wherein q is for receiving data-measuring figure place.

In second step, statistical parameter γ formula is:

$\mathrm{\γ}=2^q(1-\mathrm{\δ})(M_1^2-{\mathrm{\δM}}_2)/M_2$

Wherein δ is corrected parameter, and 0≤δ≤1.

In 3rd step, signal-to-noise ratio (SNR) estimation value λ is as follows with the mapping relations of statistical parameter γ:

$\mathrm{\γ}=f\left(\mathrm{\λ}\right)2^q(1-\mathrm{\δ})(\frac{(N-1)\mathrm{\π}{e}^{-\mathrm{\λ}}}{4N(1+\mathrm{\λ})}{((1+\mathrm{\λ})I_0\left(\frac{\mathrm{\λ}}{2}\right)+{\mathrm{\λI}}_1\left(\frac{\mathrm{\λ}}{2}\right))}^2+\frac{1}{N}-\mathrm{\δ})$

I in formula₀() is zero Bessel function；I₁() is first-order bessel function.

In 4th step, the initial value of closed loop smoothing filter output register obtains for receiving data by first paragraph Open loop signal-to-noise ratio (SNR) estimation value.

The present invention compared with prior art, its remarkable advantage: (1) estimating speed is fast, first carries out owing to have employed The method that open loop is estimated, it is only necessary to tens symbols are obtained with one and are worth according to a preliminary estimate；(2) estimated accuracy Height, the open loop estimation equation used has an advantage that estimated bias is little, and uses the method for loop filtering can To improve precision further；(3) hardware implementation complexity is low；(4) present invention is by receiving signal carrier parameter The impact of estimated accuracy；(5) present invention need not any auxiliary information；(6) this method has taken into full account quantization Impact, be therefore especially suitable for hardware realize.

Accompanying drawing explanation

Fig. 1 is that entirety of the present invention realizes block diagram.

Fig. 2 is statistical parameter of the present invention and signal to noise ratio mapping relations curve.

Fig. 3 is that second-order loop filter of the present invention realizes schematic diagram.

Detailed description of the invention

Below in conjunction with the accompanying drawings the present invention is described in further detail.

The entirety of the present invention realizes flow process as it is shown in figure 1, be described in detail each step below:

(1) envelope information obtains

Simple mpsk signal has the feature of constant-envelope, and the actual signal that receives all comprises noise.Noise The mpsk signal envelope received can be made to produce rise and fall, it is big that the change severe degree of envelope has reacted signal to noise ratio Little.Therefore, it can the envelope utilizing mpsk signal to estimate signal to noise ratio.

The first step of the present invention seeks to obtain signal envelope information.The mpsk signal received is divided into a length of N Some sections, process the most piecemeal.The mpsk signal received have passed through quadrature frequency conversion, can be divided into two Road signal r_InAnd r_Qn, n represents one section of nth data received in signal, and span is 0 to N-1. Hardware is required for logarithm value in realizing and carries out quantification treatment, represents quantization digit with q, and at this, we take q=8.First calculate and receive signal envelope second moment information u_n, method is as follows:

$u_n=\mathrm{floor}((r_{\mathrm{In}}^2+r_{\mathrm{Qn}}^2)/2^q)$

In formula, floor represents and rounds to 0 direction.When hardware realizes, square operation can cause data bits to add Times.R in this example_InAnd r_QnThe number obtained after quadratic sum is 16 quantizations, for keep fixed quantisation figure place need into Row cut position processes, and removes least-significant byte.Therefore, in formula after quadratic sum computing by the value that obtains divided by 2^q.Real Now, by r_InAnd r_QnThe low q position of quadratic sum value cuts out, and gets final product in above formula divided by 2^qAnd round to 0 direction Computing.The u obtained by said method_nIt it is the positive integer of a q position quantization.

Then according to u_nObtain envelope first moment information v_n, computational methods are as follows.

$v_n=\mathrm{round}(\sqrt{u_n}\×2^{q/2})---\left(2\right)$

Open radical sign computing due to above formula existence and open radical sign computing, it is achieved this computing can use multiple method, this example The method utilizing look-up table realizes.u_nBeing the positive integer of a q position quantization, possible value number has 2^qIndividual. By each u_nThe v that value is corresponding_nValue one look-up table of write, writing address is u_n.For keeping quantization digit, v_nValue It is also adopted by q position to quantify.It is thus desirable to setting up a size is 2^qThe look-up table of × q realizes u_nTo v_nOne by one Map.Use u_nV just can be directly read as address_nValue.

(2) statistical parameter calculates

This step can be subdivided into again two steps.First, one section of N number of first moment information v receiving signal_n Square M of sum₁, and second moment information u_nSum M₂, as follows:

$M_1={\left(\underset{n=0}{\overset{N-1}{\mathrm{\Σ}}}{v}_n\right)}^2---\left(3\right)$

$M_2=\underset{n=0}{\overset{N-1}{\mathrm{\Σ}}}{u}_n^2---\left(4\right)$

Then according to M₁And M₂, further counting statistics parameter γ according to the following formula:

γ=2^q(1-δ)(M₁/N²-δ(M₂/N))(M₂/ N) (5)

δ is modifying factor, and span is 0 to 1, takes δ=0.8 in this example.

(3) signal to noise ratio is searched

Exist between signal to noise ratio and statistical parameter γ and be shown below relation:

$\mathrm{\γ}=f\left(\mathrm{\λ}\right)2^q(1-\mathrm{\δ})(\frac{(N-1)\mathrm{\π}{e}^{-\mathrm{\λ}}}{4N(1+\mathrm{\λ})}{((1+\mathrm{\λ})I_0\left(\frac{\mathrm{\λ}}{2}\right)+{\mathrm{\λI}}_1\left(\frac{\mathrm{\λ}}{2}\right))}^2+\frac{1}{N}-\mathrm{\δ})---\left(6\right)$

In formula, λ is snr value, function I₀(·)、I₁() is respectively zero Bessel function and single order Bezier Function.This is a monotonically increasing function, and therefore statistical parameter γ and signal to noise ratio are worth between λ according to a preliminary estimate There are mapping relations one by one.This is also a nonlinear function simultaneously, it is difficult to obtained the Analytical Expression of λ by γ Formula.For by statistical parameter γ obtain signal to noise ratio be worth λ according to a preliminary estimate can in a number of different ways, such as piecewise fitting Method, approximation abbreviation method, loop up table etc..This example uses and is suitable for the LUT Method that FPGA realizes, due to letter Make an uproar and represent than usual decibels, therefore realize inverse function by look-up table λ_dB=10log₁₀(λ)=10log₁₀f^-1(γ).Its mapping relations are as shown in Figure 2.Signal-to-noise ratio (SNR) estimation precision is 0.1dB, signal-to-noise ratio (SNR) estimation interval is-2dB to 12dB, and 8 quantizations can be used to represent.γ is 8 amounts Changing, therefore look-up table size is 256*8.γ-value i.e. can find out the signal to noise ratio of correspondence directly as address.

(4) closed loop smothing filtering

Step 1 achieves signal to noise ratio open loop to 3 and estimates.In the case of not considering that signal to noise ratio changes over, open Ring estimated accuracy depends on the size of N, and N the biggest estimation progress is the biggest.But under practical situation, signal to noise ratio Changing often, can only be considered as fixed value at short notice, therefore data length N can not be excessive, otherwise Within this segment data persistent period, signal to noise ratio may have occurred and that large change.Which limits in actual application and open The precision of ring method of estimation.For improving estimation progress further, loop filter can be used to estimate with open loop The method that device combines.This example employing one obtains second-order loop filtering as shown in Figure 3 and carries out value according to a preliminary estimate Filtering.In figure, D1 and D2 is two depositors.When obtaining first paragraph snr of received signal estimated value, cut Changing switch placed below, it is placed in top in the time.Thus make the output register in loop filter D2 initial value is first time open loop signal-to-noise ratio (SNR) estimation value, and remaining time is filtered estimated value.The most permissible Realize loop filter quickly enters lock, ensure that again enough output accuracies.Filter coefficient α and β Can require to be adjusted according to signal to noise ratio change speed and estimated accuracy.This example takes α=0.016 He β=0.025。

By using the present invention, signal-to-noise ratio (SNR) estimation error can be made less than 0.1dB, there is estimating speed fast, essence Spend high advantage.

Claims (6)

1. the mpsk signal unbound nucleus signal-noise ratio estimation method that an Open-closed-loop combines, it is characterised in that Step is as follows:

The first step, envelope information obtains, i.e. according to two-way orthogonal signalling r received_InAnd r_Qn, determine that it wraps Network second moment information u_n, then recycle u_nDetermine envelope first moment information v_n；

Second step, statistical parameter calculates, i.e. one section N number of first moment information v receiving signal_nSum square M₁, and second moment information u_nSum M₂, pass through M₁And M₂Determine statistical parameter γ；

3rd step, signal-to-noise ratio (SNR) estimation, i.e. utilize the mapping relations one by one between statistical parameter γ and signal to noise ratio λ, According to γ-value, signal to noise ratio λ is estimated；

4th step, closed loop smothing filtering, i.e. utilize a second-order loop filter to above using open-loop method Obtain signal to noise ratio to be worth according to a preliminary estimate and smooth, improve estimated accuracy further, the letter obtained after smothing filtering Compared estimate value of making an uproar is as final output result ρ.

The mpsk signal unbound nucleus signal-to-noise ratio (SNR) estimation side that Open-closed-loop the most according to claim 1 combines Method, it is characterised in that: in the first step, envelope second moment information u_nFormula is:

$u_n=floor\left(\right(r_{In}^2+r_{Qn}^2)/2^q)$

Wherein q is for receiving data-measuring figure place.

The mpsk signal unbound nucleus signal-to-noise ratio (SNR) estimation side that Open-closed-loop the most according to claim 1 combines Method, it is characterised in that: in the first step, envelope first moment information v_nFormula is:

$v_n=round(\sqrt{u_n}\×2^{q/2})$

Wherein q is for receiving data-measuring figure place.

The mpsk signal unbound nucleus signal-to-noise ratio (SNR) estimation side that Open-closed-loop the most according to claim 1 combines Method, it is characterised in that: in second step, statistical parameter γ formula is:

$\mathrm{\γ}=2^q(1-\mathrm{\δ})(M_1^2-{\mathrm{\δM}}_2)/M_2$

Whereinδ is corrected parameter, and 0≤δ≤1；Q is for receiving number According to quantization digit.

The mpsk signal unbound nucleus signal-to-noise ratio (SNR) estimation side that Open-closed-loop the most according to claim 1 combines Method, it is characterised in that: in the 3rd step, signal-to-noise ratio (SNR) estimation value λ is as follows with the mapping relations of statistical parameter γ:

$\mathrm{\γ}=f\left(\mathrm{\λ}\right)=2^q(1-\mathrm{\δ})(\frac{(N-1){\mathrm{\πe}}^{-\mathrm{\λ}}}{4N(1+\mathrm{\λ})}{\left(\right(1+\mathrm{\λ}\left)I_0\right(\frac{\mathrm{\λ}}{2})+{\mathrm{\λI}}_1(\frac{\mathrm{\λ}}{2}\left)\right)}^2+\frac{1}{N}-\mathrm{\δ})$

I in formula₀() is zero Bessel function；I₁() is first-order bessel function；Q is for receiving data volume Changing figure place, δ is corrected parameter.

The mpsk signal unbound nucleus signal-to-noise ratio (SNR) estimation that Open-closed-loop the most according to claim 1 combines Method, it is characterised in that: in the 4th step, the initial value of closed loop smoothing filter output register is by first Section receives data and obtains open loop signal-to-noise ratio (SNR) estimation value.