CN104270328B - A kind of signal to noise ratio real-time estimation method - Google Patents

Abstract

The present invention relates to a kind of real-time estimation method, receiving terminal receives passband data signal and is, output result is converted to by quadrature frequency conversion, according to some specific input dataMean powerAnd output dataMean power, obtain receivingkThe estimate of signal to noise ratio during individual dataFor：；The present invention receives signal total mean power and useful signal mean power and the relation of noise average power, solves useful signal mean power and noise average power, and then estimate signal to noise ratio using before and after quadrature frequency conversion；The present invention is not only suitable for single carrier digital modulation signals, is also applied for multi-carrier digital modulated signal；Mass data need not be cached, the features such as amount of calculation is small by having compared with other methods of estimation, with complexity is low, wide adaptability, efficiency high the features such as, it is particularly suitable for hardware realization, it is adaptable to the technical field such as cooperative communication, the analysis of non-co-operation signal, electromagnetic environment monitor.

Description

A kind of signal to noise ratio real-time estimation method

Technical field

The present invention relates to the method for estimation of signal to noise ratio, a kind of specifically signal to noise ratio of digital modulation signals side of estimation in real time Method.

Background technology

Signal to noise ratio is an important parameter for weighing channel quality, and many applications are required for signal to noise ratio as priori, Such as the adaptive coding and modulating in cooperative communication, bit error rate estimation, Turbo decodings；And analyzed for non-co-operation signal, letter Make an uproar than being the important references parameter for selecting Modulation Identification method, and be the important evidence for assessing data after demodulating confidence level；Separately Outside, the normalized premise of signal power in the accurate estimation of signal to noise ratio or demodulating process, only realizes the normalization of signal, The correct judgement of symbol can be realized.

At present, signal-noise ratio estimation method mainly has the method for estimation based on symbol square, segmentation symbol square signal-to-noise ratio (SNR) estimation, base Method of estimation, maximum Likelihood in Subspace Decomposition, and auto-relativity function method etc..These methods are carrying out noise It is required for caching substantial amounts of data during compared estimate, and estimated accuracy requires that more high required data buffer storage space is bigger, this Undoubtedly add hard-wired complexity；In addition, the above method is only suitable for estimating single carrier number in addition to digital signal processing The signal to noise ratio of word modulated signal, and digital signal processing computation complexity is very big, hardware is realized difficult.

The content of the invention

The present invention proposes a kind of digital modulation signals signal to noise ratio real-time estimation method, directly utilizes natural birth in demodulating process Raw signal, estimates signal to noise ratio in real time, only needs to cache five data in estimation procedure, so as to overcome above method needs A large amount of data cached shortcomings, and method proposed by the invention there is no particular requirement that to digital modulation signals, both be applicable Multi-carrier digital modulated signal is also applied in single carrier digital modulation signals.

Technical scheme is as follows：

A kind of signal to noise ratio real-time estimation method, it is characterised in that real-time estimating step is as follows：

(1) what receiving terminal was received is that the passband data signal crossed by noise pollution is x (n), wherein,

N=i+1, i+2 ..., i+L；The L is data length, the i.e. length of window for signal-to-noise ratio (SNR) estimation, and i is window Sequence number；When the passband data signal received is k-th of input data x (k), the k ∈ n；

(2) so x (k) mean power is：

Wherein P_x(k-1) it is the mean power of x (k-1) when receiving -1 data of kth；

(3) after for received passband data signal x (n), carry out quadrature frequency conversion and passband signal be converted into baseband signal, Quadrature frequency conversion output result y (n)；After influence of the LPF to baseband signal is ignored, the y (n) can be with equivalently represented For：Y (n)=b (n)+v (n) * h (n), wherein b (n) are complex baseband signal, and v (n) is white complex gaussian noise, under h (n) is orthogonal The shock response of low pass filter in conversion process, wherein h (n) data length are K；Therefore, when receiving k-th of input number After x (k), obtained output data is y (k)；

(4) so y (k) mean power is：

(5) according to x (k) mean power P_x(k) with y (k) mean power P_y(k) believe when obtaining receiving k-th of data Make an uproar than estimateFor：

Wherein

(6) k=k+1 is taken, (1)-(5) is repeated, until k=L, takesEstimation procedure terminates.

Further, the quadrature frequency conversion described in step (3) includes two steps of shift frequency and LPF, LPF Effect be to filter out the high-frequency signal produced during shift frequency.

From above-mentioned estimation flow, the present invention only needs additional buffered c, k, P_x(k)、P_y(k) andFive mediants According to, and because quadrature frequency conversion is the general process of signal demodulation, so complexity increase is several as caused by algorithm for estimating It can ignore.

Beneficial effects of the present invention are as follows：

The present invention is put down using signal total mean power before and after quadrature frequency conversion, is received with useful signal mean power and noise The relation of equal power, solves useful signal mean power and noise average power, and then estimate signal to noise ratio；The present invention was both fitted For single carrier digital modulation signals, multi-carrier digital modulated signal is also applied for；Have compared with other methods of estimation and be not required to Cache mass data, the features such as amount of calculation is small, with complexity is low, wide adaptability, efficiency high the features such as, be particularly suitable for hardware Realize, it is adaptable to the technical field such as cooperative communication, the analysis of non-co-operation signal, electromagnetic environment monitor.

Brief description of the drawings

Fig. 1 is estimation steps flow chart of the invention；

Fig. 2 is that the signal to noise ratio in the specific embodiment of the invention estimates dynamic process figure in real time.

Embodiment

A kind of signal to noise ratio real-time estimation method disclosed by the invention, as shown in figure 1, its real-time estimating step is as follows：

(1) what receiving terminal was received is that the passband data signal crossed by noise pollution is x (n), wherein,

N=i+1, i+2 ..., i+L；The L is data length, the i.e. length of window for signal-to-noise ratio (SNR) estimation, and i is window Sequence number；When the passband data signal received is k-th of input data x (k), the k ∈ n；

(2) so x (k) mean power is：

Wherein P_x(k-1) it is the mean power of x (k-1) when receiving -1 data of kth；

(3) after for received passband data signal x (n), carry out quadrature frequency conversion and passband signal be converted into baseband signal, Quadrature frequency conversion output result y (n)；After influence of the LPF to baseband signal is ignored, the y (n) can be with equivalently represented For：Y (n)=b (n)+v (n) * h (n), wherein b (n) are complex baseband signal, and v (n) is white complex gaussian noise, under h (n) is orthogonal The shock response of low pass filter in conversion process, wherein h (n) data length are K；Therefore, when receiving k-th of input number After x (k), obtained output data is y (k)；

(4) so y (k) mean power is：

(5) according to x (k) mean power P_x(k) with y (k) mean power P_y(k) believe when obtaining receiving k-th of data Make an uproar than estimateFor：

Wherein

(6) k=k+1 is taken, (1)-(5) is repeated, until k=L, takesEstimation procedure terminates.

Specifically above-mentioned signal-noise ratio estimation method is verified by taking a kind of 16QAM signals as an example.

Simulation parameter is：Character rate 1/16sps；Carrier frequency 1/4Hz；Sample rate 1Hz；Forming filter is root liter Cosine filter, form factor is 0.35；Data length is 16000 (1000 symbols)；Signal to noise ratio is 10dB；Low pass filter H (n) is FIR filter, and exponent number is 16, is 3/8Hz by frequency, specific coefficient is：0.0000,0.0048,0.0080- 0.0089, -0.0429, -0.0290,0.0973,0.2834,0.3745,0.2834,0.0973, -0.0290, -0.0429, - 0.0089 0.0080,0.0048,0.0000.

The real-time estimating step of signal to noise ratio is as follows：

(1) P is taken_x(0)=P_y(0)=0, L=16000, utilizeWherein

C=0.3255 is calculated, and takes k=1

(2) data x (k) is received；

(3) calculate

(4) k-th of output data y (k) after quadrature frequency conversion is received；

(5) calculate

(6) calculate

(7) k=k+1 is made, (1)-(6) are obtained until k=L

As shown in Figure 2, it is shown that the dynamic process that signal to noise ratio is estimated in real time.

Claims (2)

1. a kind of signal to noise ratio real-time estimation method, it is characterised in that real-time estimating step is as follows：

(1) what receiving terminal was received is that the passband data signal crossed by noise pollution is x (n), wherein,

N=i+1, i+2 ..., i+L；The L is data length, the i.e. length of window for signal-to-noise ratio (SNR) estimation, and i is window sequence Number；When the passband data signal received is k-th of input data x (k), the k ∈ n；

(2) so x (k) mean power is：

$P_x\left(k\right)=\frac{1}{k}\underset{n=1}{\overset{k}{\Σ}}{\left|x\left(n\right)\right|}^2=\frac{k-1}{k}\[\frac{1}{k-1}\underset{n=1}{\overset{k-1}{\Σ}}{\left|x\left(n\right)\right|}^2+\frac{{\left|x\left(k\right)\right|}^2}{k-1}\]=\frac{1}{k}\[(k-1)P_x(k-1)+{\left|x\left(k\right)\right|}^2\],$

Wherein P_x(k-1) it is the mean power of x (k-1) when receiving -1 data of kth；

(3) after for received passband data signal x (n), carry out quadrature frequency conversion and passband signal is converted into baseband signal, it is orthogonal Down coversion output result y (n)；After influence of the LPF to baseband signal is ignored, the y (n) is equivalently represented to be：Y (n)= B (n)+v (n) * h (n), wherein b (n) are complex baseband signal, and v (n) is white complex gaussian noise, and h (n) is during quadrature frequency conversion The data length of the shock response of low pass filter, wherein h (n) is K；Therefore, after k-th of input data x (k) is received, Obtained output data is y (k)；

(4) so y (k) mean power is：

(5) according to x (k) mean power P_x(k) with y (k) mean power P_y(k) signal to noise ratio when obtaining receiving k-th of data EstimateFor：

$\widehat{\mathrm{\ρ}}\left(k\right)=10\lg \frac{P_y\left(k\right)-{\mathrm{cP}}_x\left(k\right)}{P_x\left(k\right)-P_y\left(k\right)},$

Wherein

(6) k=k+1 is taken, (1)-(5) is repeated, until k=L, takesEstimation procedure terminates.

2. a kind of signal to noise ratio real-time estimation method according to claim 1, it is characterised in that：Described in step (3) just Down coversion is handed over to include two steps of shift frequency and LPF, the effect of LPF is to filter out the high frequency letter produced during shift frequency Number.