CN106878211A - A kind of satellite channel multimode blind equalization algorithm - Google Patents

Abstract

A kind of satellite channel multimode blind equalization algorithm disclosed by the invention, first, generates discrete-time signal, and discrete-time signal is calculated into decision error absolute value after satellite-signal is carried out into equalization processing；Secondly, when decision error absolute value>During default worst error threshold value, decision error absolute value is processed using VS CMA algorithms；Again, when decision error absolute value<During default minimal error threshold value, decision error absolute value is processed using DD LMS algorithms；Finally, when default minimal error threshold value<Decision error absolute value<During default worst error threshold value, bimodulus is weighted using VS CMA algorithms, DD LMS algorithms to decision error absolute value and is processed.The present invention controls the step-length of CMA algorithms by introducing nonlinear residual error function, to realize the CMA algorithms of variable step size.Method disclosed by the invention has smaller residual error and faster convergence rate, and causes that intersymbol interference is smaller, and the trace ability for time varying channel is strengthened.

Description

A kind of satellite channel multimode blind equalization algorithm

Technical field

The present invention relates to satellite channel balancing technique, and in particular to a kind of satellite channel multimode blind equalization algorithm.

Background technology

Satellite communication transmission distance, is not limited by any complicated geographical conditions of communication point-to-point transmission, is not appointed by point-to-point transmission What natural calamity and the influence of man induced event, are widely used in the aspects such as military communication, satellite television broadcasting.But in relaying In satellite communication system, frequency selective fading is larger, and Channel propagation delay is more long, and Doppler frequency shift is serious, and these shortcomings exist The transmission quality of signal largely be have impact on so that intersymbol interference is extremely serious, or even influences the normal work of system.For Acquisition preferable communication efficiency reduces intersymbol interference, it is therefore desirable to which channel transfer characteristic is carried out, it is necessary to improve transmission characteristic Compensation and correction.

Balanced device is a kind of wave filter of the redeeming for being applied to receiving terminal of communication system, and it can be to transmission channel Compensate, reduce intersymbol interference, improve the accuracy of data transfer.

Constant modulus algorithm (CMA) complexity is low, it is easy to accomplish, however step-length it is smaller when the slow convergence precision of convergence rate it is high, step Fast convergence rate but convergence precision is relatively low when length is larger.

The convergence rate of decision-making direction calculation (DD) is significantly larger than Bussgang class algorithms, can be good at improving balanced device Constringency performance and error performance.But if channel variation is more, then judgement error rate can be caused higher, and then cause algorithm Cannot restrain, therefore decision-making direction calculation is often applied to tend to convergent equalizing signal.

The content of the invention

It is an object of the invention in order to improve the transmission performance of satellite channel, improve the constringency performance and error of balanced device Performance；A kind of satellite channel multimode blind equalization algorithm is provided.

In order to achieve the above object, the present invention is achieved through the following technical solutions：

A kind of satellite channel multimode blind equalization algorithm, the algorithm includes：

Discrete-time signal z (n) is generated after satellite-signal y (n) is carried out into equalization processing, and the discrete time is believed Number calculate decision error absolute value | d (n) |；

As decision error absolute value | the d (n) |>Default worst error threshold value C_maxWhen, it is exhausted to the decision error Processed using variable step size constant modulus algorithm being worth；

As decision error absolute value | the d (n) |<Default minimal error threshold value C_minWhen, it is exhausted to the decision error Processed using decision-making sensing least mean square algorithm being worth；

As the default minimal error threshold value C_min<Decision error absolute value | the d (n) |<The default maximum Error threshold value C_maxWhen, the decision error absolute value is pointed to using the variable step size constant modulus algorithm, the decision-making minimum Mean square algorithm is weighted bimodulus treatment.

It is preferred that described carry out satellite-signal to generate discrete-time signal after equalization processing, and to it is described discrete when Between signal of change decision error absolute value the step of, specifically include：

Satellite-signal y (n) is demodulated discrete-time signal z (n) is obtained after equilibrium treatment；

Discrete-time signal z (n) is estimated into signal by being obtained after maximum likelihood decision

Decision error absolute value | the d (n) | is calculated according to following formula：

It is preferred that described as decision error absolute value | the d (n) |>Default worst error threshold value C_maxWhen, to institute The step of stating decision error absolute value and processed using variable step size constant modulus algorithm, specifically include：

The step of being processed using variable step size constant modulus algorithm the decision error absolute value using following formula：

F (n+1)=f (n)-μ (n)_VS-CMAe(n)_VS-CMAy^*(n) (2)；

Wherein, f (n) is the tap coefficient of balanced device, ()^*Represent the conjugation of plural number, step size controlling functionFirst error function e (n)_VS-CMA=z (n) (| z (n) |²- R),K is used to control the scope of μ (n) values, and α and β is used for the shape of control function.

It is preferred that described as decision error absolute value | the d (n) |<Default minimal error threshold value C_minWhen, to institute State decision error absolute value using decision-making point to least mean square algorithm processed the step of, specifically include：

Least mean square algorithm is pointed to using equation below using decision-making to the decision error absolute value to process：

F (n+1)=f (n)-μ_DD-LMSe(n)_DD-LMSy^*(n) (3)；

Wherein, the second error functionIteration step length μ_；D-LMSIt is constant.

It is preferred that described as the default minimal error threshold value C_min<Decision error absolute value | the d (n) |<Institute State default worst error threshold value C_maxWhen, to the decision error absolute value using the variable step size constant modulus algorithm, described The step of decision-making points to least mean square algorithm and is weighted bimodulus and processes, specifically includes：

Bimodulus is weighted to the decision error absolute value using equation below to process：

F (n+1)=f (n)-[w (n) μ (n)_VS-CMAe(n)_VS-CMA+(1-w(n))μ_DD-LMSe(n)_DD-LMS]y^*(n) (4)；

Wherein, w (n) is weighting function, specific as follows：

On the basis of common sense in the field is met, above-mentioned each optimum condition can be combined, and obtain final product each preferable reality of the present invention Example.

Positive effect of the invention is：

A kind of satellite channel multimode blind equalization algorithm disclosed by the invention, first, equalization processing is carried out by satellite-signal After generate discrete-time signal, and the discrete-time signal is calculated into decision error absolute value；Secondly, when the decision error Absolute value>During default worst error threshold value, to the decision error absolute value using variable step size constant modulus algorithm at Reason；Again, when the decision error absolute value<During default minimal error threshold value, the decision error absolute value is used Decision-making is pointed to least mean square algorithm and is processed；Finally, when the default minimal error threshold value<The decision error is absolute Value<During the default worst error threshold value, to the decision error absolute value using the variable step size constant modulus algorithm, The decision-making points to least mean square algorithm and is weighted bimodulus treatment.The present invention is compared to prior art to traditional constant modulus algorithm (CMA, Constant Modulus Algorithm) is improved, and is controlled by introducing nonlinear residual error function The step-length of CMA algorithms, to realize the CMA algorithms of variable step size.Method disclosed by the invention has smaller residual error and more Fast convergence rate, and cause that intersymbol interference is smaller, the trace ability for time varying channel is strengthened.

Brief description of the drawings

Fig. 1 is a kind of overall flow schematic diagram of satellite channel multimode blind equalization algorithm of the invention.

Fig. 2 is a kind of structure chart of satellite channel multimode blind equalization algorithm of the invention.

Specific embodiment

Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete Site preparation is described, it is clear that described embodiment is only a part of embodiment of the invention, rather than whole embodiments.It is based on Embodiment in the present invention, it is every other that those of ordinary skill in the art are obtained under the premise of creative work is not made Embodiment, belongs to the scope of protection of the invention.

A kind of satellite channel multimode blind equalization algorithm as shown in Figure 1 and Figure 2, including：

S1, generates discrete-time signal z (n), and discrete time is believed after satellite-signal y (n) is carried out into equalization processing Number calculate decision error absolute value | d (n) |.The step is specific as follows：

As shown in Fig. 2 x (n) is transmission sequence in blind equalization systems, h (n) is the shock response of channel, and N (n) is noise Sequence, satellite-signal y (n) is receiving sequence while being again the input signal of blind equalization systems.

Then,

Discrete-time signal z (n) is obtained after satellite-signal y (n) is demodulated into equilibrium treatment.

Then,

Discrete-time signal z (n) is estimated into signal by being obtained after maximum likelihood decision

Decision error absolute value | d (n) | is calculated according to following formula：

Method disclosed by the invention, at algorithm iteration initial stage or larger changes in channel characteristics, use larger step-length with Ensure convergence rate；After algorithmic statement, reduce step-length to ensure convergence precision.According to this principle, we are walked using a kind of Nonlinear function between the factor long and residual error controls step-length.

S2, error in judgement absolute value | d (n) | and default worst error threshold value C_max, default minimal error threshold value C_minBetween relation, correspondence performs step S2.1, S2.2 and S2.3.

S2.1, as decision error absolute value | d (n) |>Default worst error threshold value C_maxWhen, it is absolute to decision error Value is processed using variable step size constant modulus algorithm (VS-CMA).

In this step, concrete operations are as follows：

Decision error d (n) is：

Order

We can release：

Wherein ζ (n) be independent same distribution and average be zero Gaussian noise.It can thus be derived that：

Wherein w (n) is weighted error vector.

In VS-CMA implementation processes,Gradually it is intended to f^TN (), such weighted error vector w (n) also tends to Zero, residual error can also reduce.After algorithmic statement, d (n) is adapted to control the value of step-length.Yet with the presence of ζ (n), D (n) is very sensitive for interference signal.For some unstable channels, d (n) probably due to certain larger interference signal Generation acute variation.In this case, we directly can not go to control step-length using d (n), and otherwise step-length will be excessive so that calculates Method convergence cannot be guaranteed.Therefore, a nonlinear function is used in the present invention, and d (n) is turned into the one of nonlinear function The individual factor, to ensure that step sizes can be influenceed by residual error, while controlling step-length to be changed in a rational scope.

Using following tap coefficients iterative formula to decision error absolute value using variable step size constant modulus algorithm at The step of reason：

F (n+1)=f (n)-μ (n)_VS-CMAe(n)_VS-CMAy^*(n) (2)；

Wherein, f (n) is the tap coefficient of balanced device, ()^*Represent the conjugation of plural number, step size controlling functionFirst error function e (n)_VS-CMA=z (n) (| z (n) |²- R),

K is used to control the scope of μ (n) values so that μ (n) can adjust convergence of algorithm speed.α and β is used for controlling letter Several shapes, when α and β take different value, μ (n) can realize different radius of curvature.

In the present embodiment, when d (n) values reduce, then the value of μ (n) can diminish；When the value of d (n) increases, then μ N the value of () can also become big.

S2.2, as decision error absolute value | d (n) |<Default minimal error threshold value C_minWhen, it is absolute to decision error Value is pointed to lowest mean square (DD-LMS) algorithm and is processed using decision-making.

The general principle of DD-LMS algorithms is identical with least mean square algorithm (LMS).Work as known training sequenceWhen, use Mean square error between the output of filter and expected reception value as cost function, i.e.,：

Define R=E [y^T(n) y (n)] it is the L*L dimension autocorrelation matrixes for representing equalizer input sequence；DefinitionTo represent the cross-correlation matrix of balanced device, then cost function (3-1) is changed into：

According to minimum mean square error criterion so that J (n) is 0 to the gradient of f (n), therefore is had：

Can then release optimalShould meet：

LMS algorithm uses the square error between equalizer output sequence and ideal response sequence to replace mean square error conduct Cost function.So that：

Using steepest descent method, i.e., equalizer tap coefficient vector is adjusted along the opposite direction of cost function gradient vector. The iterative formula of tap coefficient is：

The error of DD-LMS algorithms is defined as the difference that decision device receives between signal and the output signal of judgement.

Then according to formula (3-6), decision error absolute value is pointed to using decision-making using the iterative formula of following tap coefficient Least mean square algorithm is processed：

F (n+1)=f (n)-μ_DD-LMSe(n)_DD-LMSy^*(n) (3)；

Wherein, the second error functionIteration step length μ_DD-LMSIt is constant.

Present invention use DD-LMS algorithms are implemented in combination with VS-CMA algorithms, when equilibrium treatment starts, due to judgement Error Absolute Value is more than default worst error threshold value C_max, then the constringency performance of VS-CMA can be utilized well.When equal After weighing apparatus work a period of time, channel eye pattern opens, then changeable use DD-LMS algorithms obtain faster convergence rate and more Small steady-state error.

S2.3, as default minimal error threshold value C_min<Decision error absolute value | d (n) |<Default worst error door Limit value C_maxWhen, decision error absolute value is weighted using variable step size constant modulus algorithm, decision-making sensing least mean square algorithm double Mould treatment.It is specific as follows：

Bimodulus is weighted to decision error absolute value using equation below to process：

F (n+1)=f (n)-[w (n) μ (n)_VS-CMAe(n)_VS-CMA+(1-w(n))μ_DD-LMSe(n)_DD-LMS]y^*(n) (4)；

Wherein, w (n) is weighting function, specific as follows：

Above-mentioned weighting function is a nonlinear function on residual error absolute value.It is met when decision error is exhausted To value | d (n) | close to lower limit C_minWhen w (n) tend to 0, weighted errorMiddle e (n)_DD-LMSProportion tends to 1；Work as judgement Error Absolute Value | d (n) | is close to higher limit C_maxWhen w (n) level off to 1-e^-γ, weighted errorMiddle e (n)_VS-CMAProportion Tend to 1-e^-γ.Weighting function ensure that the conversion smoothed between both of which, so that significantly more efficient combination both of which is excellent Point.

Then the present invention combines switching bimodulus and the final tap coefficient iterative formula of weight bimodulus is as follows：

Although the foregoing describing specific embodiment of the invention, it will be appreciated by those of skill in the art that these It is merely illustrative of, protection scope of the present invention is defined by the appended claims.Those skilled in the art is not carrying on the back On the premise of principle of the invention and essence, various changes or modifications can be made to these implementation methods, but these are changed Protection scope of the present invention is each fallen within modification.

Claims (5)

1. a kind of satellite channel multimode blind equalization algorithm, it is characterised in that the algorithm includes：

Discrete-time signal z (n) is generated after satellite-signal y (n) is carried out into equalization processing, and to the discrete-time signal meter Calculate decision error absolute value | d (n) |；

As decision error absolute value | the d (n) |>Default worst error threshold value C_maxWhen, to the decision error absolute value Processed using variable step size constant modulus algorithm；

As decision error absolute value | the d (n) |<Default minimal error threshold value C_minWhen, to the decision error absolute value Least mean square algorithm is pointed to using decision-making to be processed；

As the default minimal error threshold value C_min<Decision error absolute value | the d (n) |<The default worst error Threshold value C_maxWhen, lowest mean square is pointed to using the variable step size constant modulus algorithm, the decision-making to the decision error absolute value Algorithm is weighted bimodulus treatment.

2. satellite channel multimode blind equalization algorithm as claimed in claim 1, it is characterised in that described to carry out satellite-signal The step of discrete-time signal being generated after weighing apparatusization treatment, and decision error absolute value is calculated to the discrete-time signal, specifically Comprising：

Satellite-signal y (n) is demodulated discrete-time signal z (n) is obtained after equilibrium treatment；

Discrete-time signal z (n) is estimated into signal by being obtained after maximum likelihood decision

Decision error absolute value | the d (n) | is calculated according to following formula：

$\left|d\left(n\right)\right|=|\hat{x}\left(n\right)-z\left(n\right)|---\left(1\right).$

3. satellite channel multimode blind equalization algorithm as claimed in claim 2, it is characterised in that described when the decision error is exhausted To value | d (n) |>Default worst error threshold value C_maxWhen, variable step size constant modulus algorithm is used to the decision error absolute value The step of being processed, specifically includes：

The step of being processed using variable step size constant modulus algorithm the decision error absolute value using following formula：

F (n+1)=f (b)-μ (n)_VS-CMAe(n)_Vs-CMAy^*(n) (2)；

Wherein, f (n) is the tap coefficient of balanced device, ()^*Represent the conjugation of plural number, step size controlling functionFirst error function e (n)_VS-CMA=z (n) (| z (n) |²- R),K is used to control the scope of μ (n) values, and α and β is used for the shape of control function.

4. satellite channel multimode blind equalization algorithm as claimed in claim 3, it is characterised in that described when the decision error is exhausted To value | d (n) |<Default minimal error threshold value C_minWhen, lowest mean square is pointed to using decision-making to the decision error absolute value The step of algorithm is processed, specifically includes：

Least mean square algorithm is pointed to using equation below using decision-making to the decision error absolute value to process：

F (n+1)=f (n)-μ_DD-LMSe(n)_DD-LMsy^*(n) (3)；

Wherein, the second error functionIteration step length μ_；D-LMSIt is constant.

5. satellite channel multimode blind equalization algorithm as claimed in claim 4, it is characterised in that described when the default minimum Error threshold value C_min<Decision error absolute value | the d (n) |<The default worst error threshold value C_maxWhen, sentence to described Certainly Error Absolute Value points to least mean square algorithm and is weighted bimodulus treatment using the variable step size constant modulus algorithm, the decision-making The step of, specifically include：

Bimodulus is weighted to the decision error absolute value using equation below to process：

F (n+1)=f (n)-[w (n) μ (b)_VS-CMAe(n)_vS-CMA+(1-w(n))μ_DD-LMSe(n)_DD-LMS]y^*(n) (4)；

Wherein, w (n) is weighting function, specific as follows：

$w\left(n\right)=1-\exp (-\mathrm{\&gamma;}\frac{\left|d\left(n\right)\right|-C_{min}}{C_{max}-C_{min}})---\left(5\right).$