JPH0490614A - Equalizer - Google Patents

Abstract

PURPOSE:To improve followup characteristics by finding the magnitude of the absolute value of the parameter of an adaptive algorithm after equalization processing for a prescribed period, setting the optimum number of taps according to the condition of a propagation path, and performing the equalization processing again. CONSTITUTION:After the equalization processing is performed against a reception signal stored in a reception signal memory 11 by an equalization processing part 13, the optimum value of the number of taps of an equalization processing part is decided from the magnitude of the parameter decided by an adaptive algorithm parameter calculation circuit 15 by a optimum tap number decision circuit 17. In this case, when the optimum tap number is not changed, the equalization processing part 13 continues the above-mentioned equalization processing. When the optimum tap number is changed, it performs the equalization processing once more against the identical reception signal stored in the reception signal memory 11. Thus, the followup characteristics can be improved against the propagation path fluctuation.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an equalizer that eliminates transmission distortion, which is one of the major technical issues in high-speed digital communications.

In particular, the present invention relates to an adaptive equalizer that has excellent tracking characteristics for channel fluctuations in high-speed digital mobile communications.

[Conventional technology]

For example, in high-speed digital mobile communications of several +k b/s or more, overcoming deterioration of transmission characteristics (delay distortion) due to frequency selective fading caused by propagation delay time dispersion has become a major technical issue. An equalizer uses a delay element and a variable weighting circuit to eliminate transmission distortion generated in such a propagation path.

By the way, the operation of the equalizer involves initializing the tap coefficients of the multiplier (variable weighting is a circuit (tap)) based on a known signal sequence (training signal sequence) inserted into the received signal sequence in advance. There is a training section to be set, and a tracking section in which the tap coefficients are controlled based on the initial values according to the propagation path characteristics that change over time to follow changes in the propagation path characteristics. Note that as an adaptive control method for tracking tap coefficients according to fluctuations in propagation path characteristics, LM
S (Least Mean 5quares) algorithm, RL S (Recursive Leases)
t5quares) algorithm and others are known.

By the way, LMS and RLS algorithms are publicly known,
For example, the document rsIMON HAYKIN: Adapti
veFiltering Theory”, Pren
tice-Hall, 1986J, “chapter
5 “Adaptive Transversal F
filter using Gradient-Vecto
r Estimation”, chapter 8
”Adaptive Transversal F
iter using RecursiveLe
ast 5quares J.

As shown in FIG. 5, the transmission signal is configured as a burst signal to which a signal known to the receiving side is added as a training signal sequence, and a frame is configured by collecting a plurality of N burst signals.

FIG. 6 is a block diagram showing the configuration of a mobile communication receiver including an equalizer.

In the figure, a signal received by a receiving antenna 61 is amplified by a radio frequency (RF) amplifier 62, and then multiplied by a signal from a local oscillator 64 by a mixer 63 and converted to an intermediate frequency. The received signal converted to an intermediate frequency is band-limited by a band pass filter (BPF) 6.5, and then amplified via an intermediate frequency (IF) amplifier 66 and input to a quadrature detector 67, where the in-phase component is and converted into a baseband analog signal with orthogonal components. Note that the quadrature detector 67 includes a local oscillator 71, a π/2 phase shifter 72, and two mixers 73.

The received signal, which has been quadrature-detected and divided into in-phase and quadrature components, is passed through a corresponding ronos filter (
After being band-limited by an LPF (LPF) 75, it is converted into a digital signal by sampling and quantization via an analog/digital converter (A/D) 76, and then sent to an equalizer 7.
7Q is input.

FIG. 7 is a block diagram showing the configuration of a symbol tap interval decision feedback equalizer used in a mobile communication receiver.

In the figure, the decision feedback equalizer includes a delay element (T) 81
, a multiplier 82 and an adder 8 that performs Calo calculation on the output of each multiplier.
3, the input signal to the equalizer (A/D shown in Figure 6)
It consists of a feedforward transversal filter 84 that takes in (output) IN, a delay element (T) 81, a multiplier 82, and an adder 83 that adds up the outputs of each multiplier, and takes in the output signal OUT of the equalizer. ftno\nk transversal filter 85, an adder 86 that adds each fill output, a determiner 87 that determines the sign of the adder output y(i), and sends it as the equalizer output signal OUT, an adder An error detector 88 that compares the output y (i) and the determiner output (output signal 0UT) and detects the output error e (i) of the equalizer, a training signal memory 89 that stores the training signal, and The above-mentioned adaptive algorithm is operated from the output error e (i) of the equalizer and the input/output signals x (i + α) and d (i - β) of the delay element 81 of each filter to calculate the taps of each multiplier 82. Coefficient W(i)
It consists of a tap coefficient calculation circuit 90 that calculates .

Further, the switch 91 in the figure is switched depending on the training section and the tracking section.

That is, in the training period, the output of the training signal memory 89 is sent to the error detector 88, and the output from the adder 86 is sent to the error detector 88.
The difference between the output of the training signal memory 89 and the output of the training signal memory 89 is defined as the output error e (i) of the equalizer. In addition, in the tracking period, the output of the determiner 87 (output signal 0UT) is sent to the error detector 88, and the difference between the output of the adder 86 and the output of the determiner 87 is set as the output error e (i) of the equalizer. .

Based on this configuration, the equalizer initializes the tap coefficients by operating an adaptive algorithm using the training signal sequence extracted from the received signal, and after completing the training, the equalizer initializes the tap coefficients according to the set tap coefficients. Equalization processing is performed on the transmission signal sequence.

[Problems to be Solved by the Invention] In mobile communications, as the receiver moves, propagation path conditions such as the absolute delay time of the received signal, which is proportional to the distance between the transmitter and receiver, and the difference in propagation time between the leading wave and the delayed wave, change. , the magnitudes of the absolute values of the leading wave and the delayed wave are constantly changing.

To deal with fluctuations in absolute delay time, it is necessary to establish a technique to generate a stable frame synchronization signal, but in reality this technique is difficult, so by increasing the number of taps of the equalizer, the frame synchronization signal can be generated stably. Measures are being taken to absorb the fluctuations in

Also, when the propagation time difference between the leading wave and the delayed wave is large, or when the propagation path state is non-minimum phase, it is necessary to increase the number of taps of the equalizer to reduce residual code amount interference. . On the other hand, under propagation path conditions in which almost no delayed waves occur, the number of taps can be significantly reduced.

Under such conditions, in order to fix the number of taps of the equalizer to a predetermined value, it is necessary to set the number of taps to the maximum assuming the above-mentioned worst condition. However, increasing the number of taps increases the amount of computational processing required by the adaptive algorithm.
It was inevitable that the tracking characteristics of the adaptive algorithm would deteriorate as the propagation path characteristics changed.

In other words, when the number of taps is fixed, RL with excellent tracking characteristics
Even if the equalizer was controlled using the S algorithm, it was difficult to obtain sufficient tracking characteristics.

In high-speed digital communications (e.g. mobile communications) where propagation path conditions constantly change, the present invention improves the ability to follow changes in propagation path characteristics and provides equalization that can minimize the amount of computational processing of adaptive algorithms. The purpose is to provide equipment.

[Means to solve the problem]

The present invention includes a received signal memory for storing received signals, a training signal sequence read out from the received signals stored in the received signal memory to initialize tap coefficients, and then a transmission signal sequence read out to set the tap coefficients. In the equalizer, the equalization processing section can increase or decrease the number of taps, and the equalization processing section can increase or decrease the number of taps according to the set number of taps. an adaptive algorithm parameter calculation circuit configured to perform equalization processing on a predetermined received signal stored in a signal memory, and to calculate the magnitude of an adaptive algorithm parameter obtained by the equalization processing of the equalization processing section; and an optimal tap number determining circuit that determines the optimal value of the tap number based on the magnitude of the parameter of the adaptive algorithm, and sets the tap number based on the optimal value in the equalization processing section.

[Function] The present invention performs equalization processing on the received signal stored in the received signal memory in the equalization processing section, and then calculates the optimal tap from the size of the parameter determined by the adaptive algorithm parameter calculation circuit. A number determining circuit determines the optimum number of taps for the equalization processing section.

Here, when the optimal number of taps does not change, the equalization processing section continues the previous equalization process, and when the optimal number of taps changes, it re-equals the same received signal stored in the received signal memory. When equalizing a received signal with a large delay time of a delayed wave, or a received signal in a non-minimum phase state, the number of taps of the equalizer is increased to reduce the delay of the delayed wave. When equalizing a received signal that takes a short time, the number of taps of the equalizer can be reduced, and the equalization process can be executed in an optimal state. Therefore, it is possible to improve the tracking characteristics with respect to propagation path fluctuations, and it is also possible to minimize the amount of calculation processing of the adaptive algorithm.

[Example] Hereinafter, an example of the present invention will be described in detail based on the drawings.

FIG. 1 is a block diagram showing the configuration of an embodiment of the equalizer of the present invention.

As shown in FIG. 5, the transmitter adds a training signal sequence to the beginning of the burst and transmits it, and the receiver converts the received signal into a baseband signal and sends it to the equalizer.

In FIG. 1, a received signal memory 1 stores received signals converted into Hespant signals. The equalization processing section 13, which has the same configuration as the conventional equalizer shown in FIG.
A training signal sequence is read out from one burst of received signals stored in the received signal memory 11, an adaptive algorithm is operated to set initial values of tap coefficients, and then a transmission signal sequence is read out and equalized.

Here, the configuration that characterizes the present invention is that after the tap coefficients have sufficiently converged in the equalization processing unit 13, its operation is temporarily stopped, and the parameter used in the adaptive algorithm (for example, the set tap coefficient W ( i)) and calculates the square of the magnitude of its absolute value.The adaptive algorithm parameter calculation circuit 15 takes in the output and noise information that quantifies the noise in the propagation path, and determines the optimal number of taps. The optimum tap number determining circuit 17 is provided to send data to the equalization processing section 13.

In addition, various parameters used in the adaptive algorithm (
The values of the tap coefficient W(i) and the gain vector K(i) in the RLS algorithm are complex numbers. Therefore, the adaptive algorithm parameter calculation circuit 15 calculates the square of the magnitude of the absolute value, and the real part (
It is composed of a multiplier 21 that squares Re[W(i)]) and the imaginary part (Im[W(i)]), respectively, and an adder 23 that adds the outputs of each multiplier.

The optimal number of taps determination circuit 17 uses the magnitude IW(i)l'' of the square of the absolute value of the parameter outputted by the adaptive algorithm parameter calculation circuit 15 and noise information 10-r/10 that quantifies the noise in the propagation path. and a tap number determiner 27 that determines the most common value of the number of taps based on the comparison result.

Note that it is appropriate to detect the noise information by using the field strength information of the received signal of the receiver, or by using the output signal of the error detector of the equalizer after the tap coefficients have been sufficiently converged. .

The operation of this embodiment will be described below with reference to output examples of the adaptive algorithm parameter calculation circuit 15 shown in FIGS. 2 to 4.

Figure 2 (a) shows a decision feedback equalizer with a feedforward transversal filter of 5 taps and a feedback transversal filter of 4 taps, in which the operation of the equalizer is stopped at a predetermined point. The magnitude of the square of the absolute value of the tap coefficient W(i) at the time IW(i)12
is plotted on the vertical axis.

As an example, if the noise information obtained by using the field strength information of the received signal is 7 dB, then TdB
Smaller signals are considered below the noise level.

Therefore, the magnitude IW (i
) M is larger than the true value IQ-r/10 of the noise information rdB as the effective tap existence range. In the example shown in FIG. 2(a), the tap coefficients W(1) and W(
0) and W(-1) are the effective tap existence range.

Next, one extra tap is provided outside the effective tap existing range in consideration of the effects of zinc on the clock signal and the frame signal, and the total number of these taps is set as the optimal number of taps. Here, the optimal number of taps is 5 taps.

In this way, the adaptive algorithm parameter calculation circuit 15
Then, the optimal number of taps obtained by the optimal number of taps determining circuit 17 is set in the equalization processing unit 13, and after the number of taps is reduced, the received signal of the same burst stored in the received signal memory 11 is trained from the beginning. Execute the equalization process again including the operation. By following such a procedure, it is possible to improve the tracking characteristics of the equalizer and to reduce the amount of calculation processing of the adaptive algorithm.

FIG. 2(b) shows the magnitude W(i)l'' of the square of the absolute value of all tap coefficients in a similar decision feedback equalizer.
This is a case where the true value of the noise information TdB exceeds 10 by 10.

At this time, the tap coefficients W(4) and W(
-4) are both larger than IQ-r/10, and additional taps are required outside these taps. Therefore, additional taps are provided outside of those taps for a total of 11 taps, and the received signal memory 11 is
Equalization processing is performed again on the received signals of the same Hurst stored in , including the training operation from the beginning. As shown in Figure 2 (C), the new tap coefficients W(5) and W(-5) are both smaller than 10-r/1G, so the number of taps is further increased from 11 taps. There is no need to increase it. In other words, it is sufficient to restart the operation from the point at which the operation of the other person is stopped, such as after being stopped for a certain period of time.

By the way, when the RLS algorithm is used to operate the equal-other sin, there is a gain vector K (i) as an internal parameter of the adaptive algorithm, but the absolute value of the elements of this gain vector K (i) is large. indicates that the rate of change in the tap coefficient of the corresponding tap is large. In other words, even if the absolute value of the tap coefficient is small at a certain point in time, if the absolute value of the gain vector is large, this indicates that the absolute value of the tap coefficient may suddenly increase thereafter.

Therefore, in the crime of using the RLS algorithm,
It is a good idea to judge whether to increase or decrease the number of taps by including not only the absolute value of the tap coefficient but also the gain vector and other parameters of the adaptive algorithm.

Figure 3 shows the tap coefficient W(i) and gain vector K when the equal-other sin operation is stopped at a predetermined point in time in a decision feedback equalizer using the RLS algorithm.
The magnitude of the square of the absolute value of (i) W(i) l ”
IK(i) I 2 is taken on the vertical axis.

In FIG. 3(a), for example, tap coefficients W(3), W
Since the magnitude of (-3) is small, that tap is considered unnecessary when judged from only the tap coefficients in the above procedure. However, as shown in FIG. 3(b), the tap coefficient W
(3) , gain vector K(3) corresponding to W(-3)
) and K(-3) show large values, it can be expected that the tap coefficients W(3) and W(-3) will increase rapidly in the future.

Therefore, in this example, it can be determined that the number of taps can be reduced based only on the magnitude of the tap coefficient, but if the magnitude of the gain vector is also taken into account, it is not desirable to reduce the number of taps.

That is, in the decision feedback equalizer using the RLS algorithm, the configuration of the adaptive algorithm parameter calculation circuit 15 and the optimum number of taps determination circuit 17 must take this into consideration.

FIG. 4 shows the tap coefficient W(i
), where the vertical axis is the magnitude IW (iNz) of the square of the absolute value of .

Incidentally, such a state occurs when the frame synchronization circuit in the receiving stage is unable to follow fluctuations in the absolute delay time between the transmitter and receiver, and here, as shown in FIG. 4(a), the tap coefficient W An example in which the absolute value of (3) is the largest is shown.

In this case, when the received signal is read out from the received signal memory 11 again and the second equalization process is executed, the readout position is shifted and adjusted to shift to the state shown in FIG. 4(b). be able to. In other words, by making such adjustments, it is possible to reduce the number of taps,
The computational processing amount of the adaptive algorithm can be reduced.

〔Effect of the invention〕

As described above, the present invention calculates the magnitude of the absolute value of the parameter of the adaptive algorithm after equalization processing for a predetermined period, sets the optimal number of taps according to the propagation path conditions,
The tracking characteristics can be improved by performing the equalization process again. Further, the amount of calculation processing of the adaptive algorithm can be reduced, and power consumption can be reduced.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention. FIGS. 2 to 4 are diagrams showing output examples of the adaptive algorithm parameter calculation circuit. FIG. 5 is a diagram showing the frame structure of a transmission signal. FIG. 6 is a block diagram showing the configuration of a mobile communication receiver including a mobile communication receiver. FIG. 7 is a block diagram showing the configuration of a symbol tap interval decision feedback equalizer used in a mobile communication receiver. 11... Received signal memory, 13... Equalization processing unit, 1
5...Adaptive algorithm parameter calculation circuit, 17.
... Optimal number of taps determining circuit, 21... Multiplier, 23.
...Adder, 25...Comparator, 27...Tap number determiner, 61...Receiving antenna, 62...Radio frequency (RF) amplifier, 63...Mixer, 64...Local Oscillator, 65...Band pass filter (BPF),
66...Intermediate frequency (IF) amplifier, 67...Quadrature detector, 71...Local oscillator 71.72...π/2
Phase shifter, 73...Mixer 73.75...Low pass filter (LPF), 76...Analog/digital converter (A'/D), 77...Other crimes, 81...Delay Element (T), 82... Multiplier, 83... Adder, 8
4... Feedforward transversal filter, 85... Feedback transversal filter, 86... Adder, 87... Determiner, 88...
・Error detector, 89...Training signal memory, 9
0...Tap coefficient calculation circuit, 91...Switch. (a) (b) Figure 1 (C) Figure 2 En (4) W (2) En (1) W (-1) W (-2) W (-3) 猷-滲 (a) K( 4) K(3) K(2) K(1) K(0) K(-1) K(-2) K(-3) K(=Diagram-C2) W(0) Enemy(-1 ) W(-2) W(-3) W(-滲(a) W(2) W(-1) W(-2) W(-3) W(-滲C'b
) Figure

Claims (1)

[Claims] (1) A received signal memory that stores received signals; a training signal sequence is read out from the received signal stored in the received signal memory to initialize tap coefficients, and then a transmission signal sequence is read out and the tap coefficients are set. In an equalizer equipped with an equalization processing unit that automatically adjusts and performs equalization processing, the equalization processing unit is capable of increasing or decreasing the number of taps, and the received signal is adjusted according to the set number of taps. an adaptive algorithm parameter calculation circuit configured to perform equalization processing on a predetermined received signal stored in a memory, and to obtain the magnitude of an adaptive algorithm parameter obtained by the equalization processing of the equalization processing section; Equalization characterized by comprising: an optimal number of taps determining circuit that determines the optimal value of the number of taps based on the magnitude of the parameter of the adaptive algorithm, and sets the number of taps in the equalization processing section based on the optimal value. vessel.