Blind equalizer implementation method for high-order quadrature amplitude modulation signals

Abstract

The present invention discloses a blind equalizer implementation method suitable for high-order quadrature amplitude modulation signals, comprising: S1, normalizing the input signal of the equalizer so that the peak values ​​of the in-phase dimension and the quadrature phase dimension are within a predetermined dynamic range; S2, based on the characteristics of the multi-mode value of the high-order quadrature amplitude modulation signal MQAM, presetting three predetermined mode values ​​in the normalized constellation diagram, and calculating the equalization error between the equalizer output signal and the three mode values ​​through an error formula; S3, based on the LMS principle, the equalizer tap coefficient iteration method is used to drive the equalizer coefficient to iteratively update with the equalization error calculated in S2, so that the algorithm converges quickly. The present invention provides a blind equalizer implementation method suitable for high-order quadrature amplitude modulation signals, so that the blind equalizer has a simple design method, high calculation efficiency, and flexibility in parameter adjustment on the basis of maintaining an adjustable phase capability, and has strong applicability for engineering application development.

Description

Technical Field

[0001] The present invention relates to the technical field of channel blind equalization digital signal processing in communication, and more specifically, to a blind equalizer implementation method suitable for high-order quadrature amplitude modulation signals. Background Art

[0002] The realization of high-order QAM digital transceivers in the field of wired transmission meets the current demand for high-speed and large-capacity communications. This modulation method uses amplitude and phase to carry information at the same time, and has the characteristics of fully utilizing bandwidth and strong noise resistance. When the modulation order is increased, the significant characteristic of high frequency band utilization can be fully utilized.

[0003] However, as the modulation order increases, the distance and phase difference between the signal constellation points decrease at the same time. Affected by channel noise and inter-symbol interference, the system performance deteriorates sharply, seriously affecting the communication quality. The usual approach to solving this problem is to design an equalizer in the receiver so that the characteristics of the equalizer are exactly opposite to the channel characteristics, so that the equalizer can accurately and adaptively compensate for the non-ideal characteristics of the channel.

[0004] The adaptive equalization scheme based on the traditional method of sending training sequences requires additional bandwidth, which increases the system overhead. Especially in some non-cooperative application fields, the unknown nature of the training sequence makes the traditional equalization scheme undesirable. Blind equalization technology does not require the use of training sequences. It can recover the transmission sequence by using only the prior information of the received sequence, which can save bandwidth and improve the utilization of the frequency band. It is an efficient adaptive equalization technology. Therefore, the design of blind equalizers suitable for high-order QAM signals has become a key issue in high-speed and high-quality communications.

[0005] Looking further, the essence of blind equalization digital signal processing technology is a type of adaptive filtering algorithm. Its basic implementation idea is to establish a finite length filter (also called an equalizer) with adjustable coefficients in the receiver, use the received signal output by the filter and its prior information (i.e. statistical characteristics) to calculate the error, and then use the error to update the filter coefficient taps. Through this error-driven iterative update of the filter coefficient taps, the filter output signal is finally realized to meet its prior characteristics. At this time, the filter coefficient is the inverse response of the uncertainty channel. The core of this blind equalization technology lies in the error calculation method.

[0006] There are many kinds of blind equalization error calculation methods in the current academic field. The classic constant modulus algorithm (CMA) has simple calculation and good convergence performance, and has a wide range of applications. The improved constant modulus blind equalization error algorithm (MCMA) guides the convergence of the algorithm based on the statistics of the modulated in-phase branch and the orthogonal branch. The error function contains both amplitude information and phase information, so the phase deviation can be eliminated. The multi-mode blind equalization algorithm (MMA) divides the high-order constellation diagram into small sub-regions and assigns each sub-region a separate error function to obtain a smaller steady-state residual error.

[0007] In summary, although the existing blind equalization error has been incorporated into the design of the equalizer to solve related problems, its calculation method still has the following shortcomings:

[0008] First, the classic constant modulus blind equalization error algorithm can converge when applied to high-order QAM signals with multi-mode points, but the steady-state residual error after convergence is large and there is phase rotation. The higher the modulation order, the worse the equalization effect.

[0009] Second, the blind equalization error acquisition algorithm based on the idea of ​​regional division of high-order constellation diagrams pays too much attention to the convergence of steady-state errors, and even adds various process controls during the blind equalization process, which increases the complexity of the algorithm, consumes a lot of resources and has a large amount of calculation in the design and implementation.

[0010] Third, the algorithm is not flexible enough. Commonly used blind equalization designs usually only have one equalization step parameter factor to adjust the equalization effect.

[0011] The original intention of blind equalizer design is to quickly open the eye diagram of the signal affected by channel interference. At present, various blind equalization designs have certain focuses, but in practical applications, it is impossible to take into account the algorithm complexity, parameter adjustment flexibility, steady-state error, convergence speed and design ease of implementation of the equalizer design. Summary of the invention

[0012] An object of the present invention is to solve at least the above problems and / or disadvantages and to provide at least the advantages which will be described hereinafter.

[0013] In order to achieve these purposes and other advantages of the present invention, a method for implementing a blind equalizer applicable to a high-order quadrature amplitude modulation signal is provided, wherein the blind equalizer implementation is based on the premise that the occurrence of all points in the signal constellation diagram is approximately equally probable, and comprises:

[0014] S1, normalizing the input signal of the equalizer so that the peak values ​​of the in-phase dimension and the quadrature-phase dimension are within a predetermined dynamic range;

[0015] S2, based on the multi-mode value characteristics of the high-order quadrature amplitude modulation signal MQAM, three predetermined mode values ​​are preset in the normalized constellation diagram, and the equalization error between the equalizer output signal and the three mode values ​​is calculated through the error formula;

[0016] S3, an iterative method of equalizer tap coefficients based on the LMS principle, uses the equalization error calculated in S2 to drive the equalizer coefficients for iterative update, so that the algorithm converges quickly.

[0017] Preferably, in S2, three constant predetermined modulus values ​​are pre-selected to respectively approximate small modulus value signals, medium modulus value signals and large modulus value signals in the constellation space to adapt to a variety of high-order QAM signals, and the predetermined modulus value is obtained in the following manner:

[0018] In the MQAM signal, set M to {M = 2 n , n≥4, and n is an even number}, the normalization coefficient is

[0019] Then after the 16QAM constellation is normalized, the in-phase and quadrature-phase coordinates are {±1 / 4, ±3 / 4};

[0020] After the 64QAM constellation is normalized, the in-phase and quadrature-phase coordinates are {±1 / 8, ±3 / 8, ±5 / 8, ±7 / 8};

[0021] After the 256QAM constellation is normalized, the in-phase and quadrature-phase coordinates take values ​​of {±1 / 16, ±3 / 16, ±5 / 16, ±7 / 16, ±9 / 16, ±11 / 16, ±13 / 16, ±15 / 16};

[0022] Based on the above coordinates, the three selected predetermined modulus values ​​are R1=0.5, R2=0.25, and R3=0.75, respectively, to adapt to the coordinates of constellation points of different modulation orders.

[0023] Preferably, in S2, the error calculation method based on the balanced output signal and the three predetermined modulus values ​​is:

[0024] Carry out error calculation in two dimensions: in-phase and quadrature-phase.

[0025] When calculating the error, the absolute value of the signal is used instead of the square operation;

[0026] A fine-tuning factor {α, 0<α≤1} is introduced in the error calculation to fine-tune the steady-state error;

[0027] Among them, the error formula is configured as follows:

[0028] e(k)=e I (k)+je Q (k)

[0029] e I (k)=yI(k)×{α×(R1-|yI(k)|)×|(R2-|yI(k)|)|×|(R3-|yI(k)|)|}

[0030] e Q (k)=yQ(k)×{α×(R1-|yQ(k)|)×|(R2-|yQ(k)|)|×|(R3-|yQ(k)|)|}

[0031] Among them, e(k) represents the blind equalization calculation error, j represents the complex number label, and e I (k) represents the in-phase dimension error, e Q (k) represents the orthogonal phase dimension error, α represents the fine-tuning factor, and its value range is 0<α≤1. R1, R2, and R3 represent three predetermined module values ​​respectively. yI(k) represents the in-phase dimension signal output by the equalizer, and yQ(k) represents the orthogonal phase dimension signal output by the equalizer.

[0032] Preferably, the theoretical value of the fine-tuning factor in the error calculation formula is selected as any value in the interval (0, 1], and the theoretical value is approximated to the value {{1 / (2 k ),k≥0}}, to replace the multiplication operation with the shift operation.

[0033] The present invention includes at least the following beneficial effects: the blind equalizer design method (MMMA) of the present invention has the characteristics of simple algorithm, high flexibility in parameter adjustment, large steady-state error and small error, and easy design implementation on the basis of ensuring the original adjustable phase capability, and has strong applicability in engineering development.

[0034] Other advantages, objectives and features of the present invention will be embodied in part through the following description, and in part will be understood by those skilled in the art through study and practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 Schematic diagram of the implementation of the blind equalizer in the present invention;

[0036] Figure 2 It is a schematic diagram of the division of the normalized constellation diagram under 16QAM modulation of the present invention;

[0037] Figure 3 It is a schematic diagram of the division of the normalized constellation diagram under 64QAM modulation of the present invention;

[0038] Figure 4 It is a schematic diagram of the division of the normalized constellation diagram under 256QAM modulation of the present invention;

[0039] Figure 5It is an engineering implementation diagram of the improved blind equalization error calculation method of the present invention;

[0040] Figure 6 The figure is a simulation diagram of verifying the superiority of the steady-state error of the improved blind equalization error calculation method (MMMA) of the present invention and two types of classic blind equalization algorithms CMA and MCMA under 16QAM modulation;

[0041] Figure 7 The figure is a simulation diagram of verifying the superiority of the steady-state error of the improved blind equalization error calculation method (MMMA) of the present invention and two types of classic blind equalization algorithms CMA and MCMA under 64QAM modulation;

[0042] Figure 8 The figure is a simulation diagram of verifying the superiority of the steady-state error of the improved blind equalization error calculation method (MMMA) of the present invention and two types of classic blind equalization algorithms CMA and MCMA under 256QAM modulation;

[0043] Fig. 9 It is a simulation schematic diagram of verifying the flexibility of adjustment factors of the improved blind equalization error calculation method (MMMA) of the present invention and two types of classic blind equalization algorithms CMA and MCMA under 64QAM modulation. DETAILED DESCRIPTION

[0044] The present invention is further described in detail below in conjunction with the accompanying drawings so that those skilled in the art can implement the invention with reference to the description.

[0045] A blind equalizer implementation method suitable for high-order quadrature amplitude modulation signals (M-ary Quadrature Amplitude Modulation, referred to as MQAM), such as Figure 1 The design idea of ​​the present invention is to normalize the modulated signal after passing through the channel before entering the equalizer at the receiving end to adapt to the predetermined three constant modulus values; then the output of the equalizer and the predetermined three modulus values ​​are used to calculate the error information in the in-phase dimension and the orthogonal dimension according to a given error algorithm; and then the calculated error is driven to update the equalizer coefficient through the commonly used coefficient iteration equation based on the least mean square error (Least Mean Square, referred to as LMS), so that the signal eye diagram opens quickly. Furthermore, in the calculation of the error, the fine-tuning factor in the error calculation can be changed to obtain an acceptable steady-state residual error after the algorithm converges.

[0046] Specifically, the blind equalizer implementation of the high-order QAM signal of the present invention is based on the premise that the appearance of all points in the signal constellation diagram is approximately equally probable. First, the input signal of the equalizer is normalized so that the peak values ​​of the in-phase dimension (marked as I) and the quadrature-phase dimension (marked as Q) are within the dynamic range (-0.8 to +0.8). In this process, the input data of the equalizer is normalized, and a margin of 0.2 is retained when the step size parameter of the equalizer is fixed, so as to have better adaptability.

[0047] Secondly, based on the multi-mode value characteristics of high-order QAM signals, three predetermined modulus values ​​are preset in the normalized constellation diagram. Here, each point on the constellation is connected to the coordinate center as a modulus value, and three modulus values ​​with good balance in the constellation are taken to ensure good adaptability and controllable errors, and the equalization errors between the equalizer output signal and the three modulus values ​​are calculated.

[0048] Again, based on the Least Mean Square (Simple LMS) equalizer tap coefficient iteration method, the calculated equalization error is used to drive the iterative update of the equalizer coefficient, so that the algorithm converges quickly, that is, the signal eye diagram also opens quickly. In this scheme, the high-order QAM modulation constellation diagram with two-dimensional modulation characteristics is normalized, and error statistics are performed in two dimensions (in-phase and orthogonal phase) to maintain its ability to correct phase deviation. In each dimension, R1, R2, and R3 are used to represent three predetermined modulus values, that is, three convergence circle radii (convergence circle radius R1 is centered on the coordinate origin in this dimension, with a radius of 0.5; convergence circle radius R2 is centered on the coordinate origin in this dimension, with a radius of 0.25; convergence circle radius R3 is centered on the coordinate origin in this dimension, with a radius of 0.75) to obtain a smaller steady-state residual error.

[0049] Furthermore, by introducing an adjustment factor (α) in each dimension, the steady-state residual error can be further reduced. Based on this error calculation method, the equalizer can ensure the adjustable phase capability and flexibly adjust the residual steady-state error acceptance level.

[0050] In the specific implementation, the method of the present invention is further described through the following four aspects:

[0051] 1. Prerequisites for blind equalizer design and implementation

[0052] a) When implementing this, the approximate equiprobability of each point in the signal constellation diagram needs to be transmitted. The high-order MQAM signal is a typical two-dimensional modulation signal with multiple modes. The approximate equiprobability of each point in the signal constellation diagram needs to be transmitted, so that the equalizer can accurately adapt to the channel;

[0053] b) Ensure the dynamic range of the equalizer input signal at different MQAMs. Normalize the equalizer input signal to ensure the dynamic range of the input signal, which is conducive to the adaptability of the step size parameters and the pre-selected three modulus parameters when equalizing different MQAM signals.

[0054] 2. Predetermine the selection method of 3 kinds of module values

[0055] Based on different MQAM signals, three identical predetermined modulus values ​​are selected to obtain a smaller steady-state residual error. The method of obtaining three predetermined modulus values ​​with reference to the standard constellation diagram is as follows: three predetermined modulus values ​​are selected in the normalized standard constellation diagram to adapt to a variety of high-order QAM signals. Based on the characteristics of the multi-modulus values ​​of the high-order QAM signal and the compromise consideration of the algorithm implementation complexity, three constant modulus values ​​are pre-selected to respectively approximate the small modulus value signal, the medium modulus value signal and the large modulus value signal in the constellation space to ensure that there is a smaller steady-state error after the blind equalization converges.

[0056] The specific implementation process includes: for the selected high-order quadrature amplitude modulation (MQAM) signal, M can be {M = 2 n , n≥4, and n is an even number}, the normalization coefficient is

[0057] like Figure 2-4 ,After normalizing the standard MQAM constellation diagram, the common 16QAM constellation is normalized, the in-phase and quadrature-phase coordinates are {±1 / 4,±3 / 4};

[0058] After the 64QAM constellation is normalized, the in-phase and quadrature-phase coordinates are {±1 / 8, ±3 / 8, ±5 / 8, ±7 / 8};

[0059] After the 256QAM constellation is normalized, the in-phase and quadrature-phase coordinates are {±1 / 16, ±3 / 16, ±5 / 16, ±7 / 16, ±9 / 16, ±11 / 16, ±13 / 16, ±15 / 16}, etc. According to the characteristics of the above coordinates, three predetermined modulus values ​​can be selected, namely R1=0.5, R2=0.25, and R3=0.75, to respectively adapt to the coordinates of the constellation points of the medium modulus, the smaller modulus, and the larger modulus in two dimensions.

[0060] 3. Error calculation implementation

[0061] Based on the error acquisition algorithm with simple algorithm, high calculation efficiency and flexible parameters. The engineering implementation of blind equalization error calculation designed by the present invention is as follows Figure 5As shown in the figure, an error calculation method based on the balanced output signal and three predetermined modulus values ​​is proposed. First, the error calculation is carried out in the two dimensions of in-phase and quadrature phase, which can maintain the algorithm's ability to correct phase deviation. Secondly, the absolute value of the signal is used to replace the traditional square operation during error calculation, which reduces the hardware processing overhead during implementation and improves the calculation efficiency. Finally, a fine-tuning factor {α, 0<α≤1} is introduced in the error calculation to achieve fine-tuning of the steady-state error to obtain an acceptable error level during implementation.

[0062] The error formula is configured as follows:

[0063] e(k)=e I (k)+je Q (k)

[0064] e I (k)=yI(k)×{α×(R1-|yI(k)|)×|(R2-|yI(k)|)|×|(R3-|yI(k)|)|}

[0065] e Q (k)=yQ(k)×{α×(R1-|yQ(k)|)×|(R2-|yQ(k)|)|×|(R3-|yQ(k)|)|}

[0066] Wherein, e(k) represents the blind equalization calculation error, and j represents a complex number label. I (k) represents the in-phase dimension error, e Q (k) represents the orthogonal phase dimension error, α represents the fine-tuning factor, and its value range is 0<α≤1. R1, R2, and R3 respectively represent three predetermined modulus values ​​(radius of the convergence circle). yI(k) represents the in-phase dimension signal output by the equalizer, and yQ(k) represents the orthogonal phase dimension signal output by the equalizer.

[0067] 4. Implementation and parameter adjustment of equalizer

[0068] According to the design requirements, a filter with adjustable coefficients is constructed in Simulink or FPGA environment. The filter coefficients are updated based on the minimum equalization error algorithm. Load the received data and first adjust the equalizer step factor to make the algorithm converge. After the algorithm converges, adjust the value of the fine-tuning factor so that the steady-state residual error reaches the application requirement.

[0069] 5. Solution verification

[0070] In order to verify the effectiveness of the solution of the present invention, the solution verification is divided into two contents.

[0071] 1) Verify the superiority of the steady-state error of the algorithm proposed in this paper in high-order QAM modulation;

[0072] 2) Verify the flexibility of the adjustment factors proposed by the algorithm during debugging.

[0073] For the verification of the superiority of the improved algorithm with steady-state error, in order to verify the effectiveness of the blind equalization error algorithm proposed in this paper, two types of classic blind equalization algorithms in Table 1 are selected for comparison.

[0074] Table 1 Algorithm comparison

[0075]

[0076]

[0077] The algorithm simulation parameters are shown in Table 2.

[0078] Table 2 Algorithm simulation parameter settings

[0079]

[0080] The results after algorithm simulation are as follows Figure 6-8 As shown, from Figure 6-8 It can be seen from the simulation results of 16 / 64 / 256QAM modulation that the improved blind equalization error algorithm basically coincides with the coordinates and has a better steady-state convergence error.

[0081] For the verification of the flexibility of the adjustment factor, in order not to lose generality, the 64QAM modulation mode is selected for simulation verification. Although the theoretical value α of the adjustment factor ranges from 0<α≤1, it involves multiplication when participating in the operation, which consumes a lot of resources in engineering implementation. In order to realize the multiplication function through simple shift operations, α needs to be reasonably planned. Generally speaking, any value in the interval (0, 1] is usually selected to meet the acceptable steady-state error level, but in engineering development, we approximate the theoretical value to the value {{1 / (2 k ),k≥0}}, at this time, the multiplication operation can be replaced by the shift operation, so as to further reduce the hardware calculation amount and improve the speed and efficiency of the operation without losing the adjustment flexibility.

[0082] In this simulation, the values ​​of α are selected as (0.125, 0.25, 0.5, 1), and the simulation parameters are shown in Table 3:

[0083] Table 3 Parameter settings for simulations with different fine-tuning factors

[0084]

[0085]

[0086] The simulation structure is as follows Fig. 9 As shown, from Fig. 9It can be seen from the simulation result diagram that the present invention has flexibility in fine-tuning the steady-state error by using different adjustment factors.

[0087] The above solution is only an illustration of a preferred embodiment, but is not limited thereto. When implementing the present invention, appropriate replacement and / or modification can be performed according to user needs.

[0088] The number of devices and processing scales described here are used to simplify the description of the present invention. Applications, modifications and variations of the present invention will be obvious to those skilled in the art.

[0089] Although the embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and the embodiments. It can be fully applied to various fields suitable for the present invention, especially cardinality QAM signal modulation and CAP signal modulation technology. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and the illustrations shown and described here.

Claims

1. A method for implementing a blind equalizer for high-order quadrature amplitude modulation signals, wherein the blind equalizer is implemented based on the premise that all points in the signal constellation diagram have approximately equal probability of appearing, and is characterized in that: include: S1, normalizing the input signal of the equalizer so that the peak values ​​of the in-phase dimension and the quadrature-phase dimension are within a predetermined dynamic range; S2, based on the multi-mode value characteristics of the high-order quadrature amplitude modulation signal MQAM, three predetermined mode values ​​are preset in the normalized constellation diagram, and the equalization error between the equalizer output signal and the three mode values ​​is calculated through the error formula; S3, an iterative method of equalizer tap coefficients based on the LMS principle, uses the equalization error calculated in S2 to drive the equalizer coefficients for iterative update, so that the algorithm converges quickly; In S2, three constant predetermined modulus values ​​are pre-selected to respectively approximate small modulus value signals, medium modulus value signals and large modulus value signals in the constellation space to adapt to a variety of high-order QAM signals. The predetermined modulus value is obtained as follows: In the MQAM signal, set M to , take the normalization coefficient as ; Then after the 16QAM constellation is normalized, the in-phase and quadrature-phase coordinates are {±1 / 4,±3 / 4}; After the 64QAM constellation is normalized, the in-phase and quadrature-phase coordinates are {±1 / 8, ±3 / 8, ±5 / 8, ±7 / 8}; After the 256QAM constellation is normalized, the in-phase and quadrature-phase coordinates are {±1 / 16, ±3 / 16, ±5 / 16, ±7 / 16, ±9 / 16, ±11 / 16, ±13 / 16, ±15 / 16}; Based on the above coordinates, the three predetermined modulus values ​​selected are R1=0.5, R2=0.25, and R3=0.75, respectively, to adapt to the coordinates of constellation points of different modulation orders; In S2, the error calculation method based on the balanced output signal and three predetermined modulus values ​​is: Carry out error calculation in two dimensions: in-phase and quadrature-phase. When calculating the error, the absolute value of the signal is used instead of the square operation; A fine-tuning factor is introduced in the error calculation , make fine adjustments to the steady-state error; Among them, the error formula is configured as follows: Among them, e(k) represents the blind equalization calculation error, j represents the complex number label, and e I (k) represents the in-phase dimension error, e Q (k) represents the orthogonal phase dimension error, α represents the fine-tuning factor, and the value range is 0<α≤1, R1, R2, and R3 represent three predetermined modulus values, yI(k) represents the in-phase dimension signal output by the equalizer, and yQ(k) represents the orthogonal phase dimension signal output by the equalizer; In the error calculation formula, the theoretical value of the fine-tuning factor is selected as any value in the interval (0, 1]. During engineering development, the theoretical value is approximated to the value , to replace the multiplication operation with a shift operation.

Images