JPH07226704A - Adaptive equalizer - Google Patents

Abstract

PURPOSE:To obtain an adaptive diversity equalizer in which an arithmetic quantity is remarkably reduced and a degree of deterioration in a converging speed of a tap coefficient is less even when the number of branches is increased and a total tap number is increased. CONSTITUTION:The adaptive equalizer is provided with an equalization filter section with a tapped delay circuit, at a feedforward section (FF section), a feedback section (FB section) respectively receiving a detection signal of a detection circuit 3 (4) a tap coefficient arithmetic section, plural discrimination feedback adaptive equalizers 33, 34 updating the tap coefficient of the equalization filter section, a synthesis circuit 35 synthesizing the weighting of output signals of the discrimination feedback adaptive equalizers 33, 34, a data discrimination section 36 discriminating an output result of the synthesis circuit 35 to obtain final output data, and each of the plural discrimination feedback adaptive equalizers 33, 34 is provided with a means deciding each tap coefficient based on the output result of a data discrimination section 36.

Description

Detailed Description of the Invention

[0001]

[Industrial application] In high-speed digital mobile communication,
Waveform distortion due to frequency selective fading causes great deterioration in transmission characteristics. The present invention relates to an adaptive diversity equalizer used for suppressing the deterioration of transmission characteristics.

[0002]

2. Description of the Related Art First, a conventional configuration and operation of an adaptive diversity equalizer will be described. As one method of combining diversity and an adaptive equalizer, which follows rapid fluctuations of a transmission line such as mobile communication and overcomes deterioration of transmission characteristics due to level reduction due to fading, for example, Yoshino and Suzuki: "Interference Cancellation Characteristics of DFE Transversal Combining Diversity System for Mobile Radio-Comparison with Metric Combining-" IEICE Transactions BI
I. Vol. J76 B-II. No. 7 pp. 584
There is known an adaptive diversity equalizer using a transversal combining type diversity equalization system as described in -595 (1993.7).

FIG. 6 is a block diagram showing an example of the basic configuration of an adaptive diversity equalizer using the transversal combining type diversity equalization system shown in the above document. In the figure, 1 is an antenna a, 2 is an antenna b, 3 is a detection circuit a for detecting a received signal output from the antenna a, 1 and converting it into a baseband signal, 4 is a received signal output from the antennas b, 2. The detection circuits b and 5 for detecting the signal and converting it into a baseband signal are received signal memories a and 6 for storing one burst of the reception signal received from the antennas a and 1 and converted into the baseband signal by the detection circuits a and 3. Received signal memories b and 7 for storing one burst of the received signals received from b and 2 and converted to baseband signals by the detection circuits b and 4 control the operations of the received signal memories a and 5 and the received signal memories b and 6. A control unit 8 for adjusting the received signal memory a, 5 and an adaptive equalizer for equalizing the received baseband signals stored in the received signal memories b, 6 and 9 for the received signal memories a, 5 Tap interval is a predetermined delay time to an input signal an output Tp (Tp = T / N, T is one symbol time,
(N is an integer) and the number of taps is L. The transversal filter a (FFa unit) of the feedforward unit is 10 and taps with the output of the reception signal memory b, 6 as an input signal with a constant delay time Tp seconds. The transversal filter b (FFb
Part), 11 is a transversal filter (FB part) of the feedback part having a constant delay time T seconds and the number of taps is (M−2L), and 12 is an FFa part 9 and FFb.
An adder that adds the outputs of the sections 10 and FB section 11, 13 is a determiner that identifies the output signal sequence of the adder 12 every T seconds, and makes a hard decision, and 14 is an output signal terminal of the adaptive diversity equalizer. .

FIG. 7 is a block diagram showing the configuration of the adaptive equalizer 8 shown in FIG. 6, and 15 is a reception signal memory a, 5
The input terminal of the received baseband signal stored in the input terminal 16, the input terminal 16 of the received baseband signal stored in the received signal memory b, 6, 17 is the FFa unit 9, the FFb unit 10 and the FB
A tap coefficient updating circuit that determines the tap coefficient of the unit 11 every T seconds, 18 is a switch circuit that switches the input signal sequence of the FB unit 11 between the output signal sequence of the determiner 13 and the reference signal sequence, and 19 is the reference signal sequence input terminal , 20 are adders for obtaining the difference between the output of the adder 12 and the output signal sequence or the reference signal sequence of the determiner 13.

FIG. 8 is a diagram showing an example of a burst format of a signal used in mobile communication and the like. Reference numeral 21 is a unique word (hereinafter referred to as UW) used for training the adaptive equalizer and for frame synchronization, and 22 is a random data portion.

Next, the operation of the conventional adaptive diversity equalizer will be described. In the adaptive diversity equalizer shown in FIG. 6, received signals output from antennas a and 1 and antennas b and 2 are converted into baseband signals by detection circuits a and 3 and detection circuits b and 4, respectively. The output signals of the detection circuits a and 3 are stored in the reception signal memories a and 5, and the output signals of the detection circuits b and 4 are the reception signal memories b and 6.
Stored in. Further, these baseband signals are output to the control unit 7. The control unit 7 outputs a control signal to the reception signal memories a and 5 and the reception signal memories b and 6 after frame synchronization is performed using the UW 21 shown in FIG. 8 at the beginning of each burst. In the reception signal memories a, 5 and the reception signal memories b, 6, the adaptive equalizer 8 receives the control signal,
The received signal corresponding to the burst is output.

The adaptive equalizer 8 shown in FIG. 7 estimates the characteristics of the transmission path for the received signal converted into baseband by using the UW 21 at the beginning of each burst and the received signal corresponding to the UW 21 and FFa. The tap coefficients of the unit 9, the FFb unit 10, and the FB unit 11 are converged. (Training mode). At this time, the input signal sequence of the FB unit 11 and the input signal sequence of the adder 20 are reference signal sequences defined by the UW 21 as data having no determination error. Next is the random data section 2
Equalize 2 (tracking mode). At this time, the determiner 13 outputs the output signal sequence of the adder 12 as T
The input signal sequence of the FB unit 11 and the input signal sequence of the adder 20 become the output signal sequence of the determiner 13 by discriminating and making a hard decision every second.

The tap coefficient updating circuit 17 uses the input signal sequence of the adaptive equalizer 8 and the output signal of the adder 20, that is, the error signal, and uses the Kalman filter algorithm (RLS).
Algorithm) etc., the tap coefficients of the FFa unit 9, the FFb unit 10, and the FB unit 2 are updated for each symbol.

Regarding the above tap coefficient updating algorithm, the Kalman filter algorithm (R
An example of the LS algorithm) will be described. Time t = nT
The input signal vector to the equalizer at (n = 0, 1, 2, ...; T is 1 symbol time) is X _M (n), the tap coefficient is C _M (n), and the output of the adder 12 is I. (N), the desired output is d (n), and the error signal (output of adder 20) is e
(N). Where X _M (n), C _M (n), I
(N) and d (n) are complex numbers indicating in-phase and quadrature channels. Further, when the number of taps of the FFa unit 9 and the number of taps of the FFb unit 10 are both L and the total number of taps is M, the relationship between them is as follows.

[0010]

[Equation 1]

Here, * represents a complex conjugate transposed matrix (or vector). Also, y ₁₁ (n), y ₁₂ (n),
..., y _L1 (n) is an input signal of the FFa unit 9, y
₁₂ (n), y ₂₂ (n), ..., y _L2 (n) are FFb unit 1
Represents an input signal of zero. Further, d (n) is an input signal of the FB unit 11, and becomes a reference signal sequence in the training mode, and becomes an output signal sequence in which the decision unit 13 makes a hard decision on the result of the equation 3 in the tracking mode. The error signal e (n) is the output of the adder 20. Then, the tap coefficient C that minimizes the evaluation function ε represented by the following equation
_M (n) becomes the required value.

[0012]

[Equation 2]

Here, λ represents a forgetting factor (0 <λ ≦ 1). C _M (n) that minimizes Equation 5 is as follows.

[0014]

[Equation 3]

[0015] In addition, C _M at time t = (n-1) T
From (n-1) and P (n-1), C at time t = nT
_The algorithm for recursively obtaining _M (n) is as follows.

[0016]

[Equation 4]

Here, K (n) is Kalman gain, P
(N) is an estimation error covariance matrix of tap coefficients, and I is an identity matrix. The tap coefficient updating algorithm is described in S. Heikin, "Translation by Takebe:" Introduction to Adaptive Filters ", Chapter 5, Hyundai Engineering Co. (1987), or J.
G. PROAKIS: "DIGITAL COMMUN
ICATION ”, chapter 6.8, McGRAW-HILL
(1983).

[0018]

Since the conventional adaptive diversity equalizer is configured as described above, when the number of branches of diversity, that is, the number of antennas is large, and when the tap of the FF section of each branch is used. When there are many numbers, in the tap coefficient update algorithm using the Kalman filter algorithm or the like, the amount of calculation required for one tap coefficient update is proportional to the square of the number of taps, and therefore the amount of calculation increases significantly. was there. In addition, it takes a long time for the tap coefficient to converge due to the increase in the number of taps, and in some cases, when the number of symbols of the reference signal in the training mode is small, the tap coefficient does not converge, and as a result, the characteristics may deteriorate. .

The present invention has been made in order to solve the above problems, and the amount of calculation is significantly reduced as compared with the conventional adaptive diversity equalizer without deterioration of the characteristics of the conventional adaptive diversity equalizer. Also, the number of branches has increased,
It is an object of the present invention to obtain an adaptive diversity equalizer with a small degree of deterioration of the tap coefficient convergence speed even when the total number of taps increases.

[0020]

In order to achieve the above object, an adaptive equalizer of the invention according to claim 1 divides an input signal sequence and each divided input signal sequence. As an input, an equalization filter unit having a delay circuit with a tap in a feedforward unit and a feedback unit,
And a plurality of decision feedback adaptive equalizer sections for updating the tap coefficients of the equalization filter section based on the output result of the data decision section according to the tap coefficient update algorithm. A combination circuit that weights and combines the output signals of the plurality of decision feedback adaptive equalizer units, and a data determination unit that determines the output result of the combination circuit and determines the final output data, With the combination circuit, the decision feedback adaptive equalizer block outputs with good equalization characteristics are given a large weight, and those with poor equalization characteristics are given a small weight, and the results are finalized. The equalized output is used.

In order to achieve the above object, the adaptive diversity equalizer of the invention according to claim 2 has a plurality of antennas, a plurality of detection circuits for detecting a plurality of received waves, and each signal after the detection. An equalization filter unit having a feedforward unit and a delay unit with a tap delay circuit in the feedback unit, and a tap coefficient calculation unit are provided as inputs, and the tap coefficient calculation unit is based on the output result of the data determination unit according to the tap coefficient update algorithm. A plurality of decision feedback type adaptive equalizer sections for updating tap coefficients of the equalization filter section, a combining circuit for weighting and combining the output signals of the plurality of decision feedback type adaptive equalizer sections, and The above-mentioned data judging section for judging the output result of the synthesizing circuit and making it the final output data is provided, and among the outputs of the decision feedback type adaptive equalizer section of each branch by the synthesizing circuit. Increasing the weighting for having good equalization characteristics, also synthesized by reducing the weighting for not good equalization characteristic is obtained as its result the final equalized output.

In order to achieve the above object, the adaptive diversity equalizer of the invention according to claim 3 has a plurality of antennas, a plurality of detection circuits for detecting a plurality of received waves, and an oversampled post-detection wave. For each signal of, the signals having the same timing at the time of oversampling are input to the respective signals, the equalization filter unit having the delay circuit with taps in the feedforward unit and the feedback unit, and the tap coefficient calculation unit are provided. The arithmetic unit updates the tap coefficients of the equalization filter unit based on the output result of the data determination unit according to the tap coefficient updating algorithm, and the plurality of decision feedback adaptive equalizer units and the plurality of decision feedback adaptive equalizers. The output signal of the above-mentioned synthesis circuit and the synthesis circuit that weights and synthesizes each output signal Serial a data judging unit, the combining circuit, increasing the weighting for having good equalization characteristics in the decision feedback adaptive equalizer unit output of each branch, also,
This is a method in which weighting is applied to those with poor equalization characteristics, and the result is used as the final equalized output.

[0023]

In the adaptive equalizer of the invention according to claim 1 configured as described above, the input signal is divided and tapped independently in the decision feedback type adaptive equalizer unit for each divided input signal sequence. Since the coefficients are updated and the outputs of the decision feedback adaptive equalizer units are weighted and combined, the amount of calculation required to update the overall tap coefficients is reduced and the tap coefficients required for training mode convergence are reduced. The number of symbols becomes shorter. Also, the characteristics of a certain decision feedback adaptive equalizer section are temporarily deteriorated due to fading, etc., and the decision error continues in the output result of the decision feedback adaptive equalizer section alone,
Even if the tap coefficient updating algorithm causes divergence, the output of each decision feedback type adaptive equalizer section is weighted and synthesized, and the tap coefficient of each decision feedback type adaptive equalizer section is updated using the decision result as a reference signal. Therefore, the tap coefficient updating algorithm diverges less frequently.

In the adaptive diversity equalizer of the invention according to claim 2 configured as described above, the decision feedback type adaptive equalizer section of each branch updates the tap coefficient independently, and each decision feedback type. Since the output of the adaptive equalizer unit is weighted and combined, the amount of calculation required to update the overall tap coefficient is reduced, and the number of symbols required to converge the tap coefficient in the training mode is shortened. Also,
Even if the characteristics of a branch are temporarily deteriorated due to fading, etc., and the decision results are continuous in the output result of a single branch, and the tap coefficient updating algorithm causes divergence, etc. Since the outputs are weighted and synthesized and the decision result is used as a reference signal to update the tap coefficient of each decision feedback adaptive equalizer unit, the tap coefficient update algorithm diverges less frequently.

In the adaptive diversity equalizer of the invention according to claim 3 configured as described above, the tap coefficient is updated independently in each decision feedback type adaptive equalizer section, and each decision feedback type adaptive equalizer or the like is updated. Since the output results of the digitizer unit are weighted and combined, the amount of calculation required to update the overall tap coefficient is reduced and the number of symbols required to converge the tap coefficient in the training mode is shortened. In addition, the characteristics of a certain decision feedback adaptive equalizer part are temporarily deteriorated due to fading, etc., and the decision results are continuous in the output result of the decision feedback adaptive equalizer part alone, and the tap coefficient update algorithm diverges. Even if the above occurs, the output of each decision feedback adaptive equalizer section is weighted and synthesized, and the tap coefficient is updated to update the tap coefficient of each decision feedback adaptive equalizer section using the decision result as a reference signal. The algorithm diverges less frequently.

[0026]

【Example】

First Embodiment FIG. 1 is a configuration block diagram showing a first embodiment of the present invention. In the figure, the same parts as those of the conventional example are designated by the same reference numerals and the description thereof is omitted. In FIG. 1, reference numeral 30 denotes a reception signal memory a, 3 for storing one burst of a reception signal received from the antenna a, 1 and converted into a baseband signal by the detection circuit a, 3.
Reference numeral 1 is a reception signal memory b for storing one burst of a reception signal received from the antennas b and 2 and converted into a baseband signal by the detection circuits b and 4, 32 is a reception signal memory a,
30, a control unit for controlling the operation of the reception signal memories b and 31, 33 is a decision feedback type adaptive equalizer a and 34 using the reception baseband signal stored in the reception signal memories a and 30 as an input signal The decision feedback type adaptive equalizers b and 35 having the received baseband signals stored in the memories b and 31 as input signals are the decision feedback type adaptive equalizers a and 33 and the decision feedback type adaptive equalizers b and 34. A weighting synthesis circuit for weighting and synthesizing outputs, 36 is a determiner for performing hard decision by identifying the output signal series of the weighting synthesis circuit 35 every T seconds (one symbol time), and 37 is a decision feedback adaptive equalizer a, 33. , The transversal filter a (FF
a section), 38 is a transversal filter a (FB) of the feedback section in the decision feedback type adaptive equalizer a, 33.
a section), 39 is an adder for adding the output results of the FFa section 37 and the FBa section 38 in the decision feedback type adaptive equalizers a and 33, and 40 is a received signal memory b61 in the decision feedback type adaptive equalizers b and 34. Transversal filter b (FFb unit) of the feed forward unit that uses the output of
Reference numeral 41 is a transversal filter b (FBb section) of the feedback section in the decision feedback type adaptive equalizers b and 34,
An adder 42 adds the output results of the FFb unit 40 and the FBb unit 41 in the decision feedback type adaptive equalizers b and 34.

FIG. 2 is a block diagram showing the configuration of the decision feedback adaptive equalizers a and 33. In the figure, the same parts as those shown in the conventional example and FIG.
In FIG. 2, 43 is a received signal input terminal to which the output signals of the received signal memories a and 30 are input, 44 is a determination output signal input terminal to which the output result of the determiner 36 is input, and 45 is a weighted result of the operation of the adder 39. An equalization filter output terminal for outputting to the synthesizing circuit 35, 46 is a tap coefficient updating circuit a which determines the tap coefficient of the FFa section 37 and the FBa section 38 for each symbol, and 47 is a difference between the calculation result of the adder 39 and the desired signal. It is an adder that generates a take error signal.

FIG. 3 is a block diagram showing the configuration of the weighting synthesis circuit 35. 48 is a decision feedback type adaptive equalizer a, 3
The input terminal a to which the output signal of 3 is input, 49 is the input terminal b to which the output signal of the decision feedback type adaptive equalizer b and 34 is input,
50 is a multiplier a that multiplies the output signals of the decision feedback adaptive equalizers a and 33 by a weighting coefficient a, and 51 is a multiplier b that multiplies the output signals of the decision feedback adaptive equalizers b and 34 by a weighting coefficient b, Reference numeral 52 is a coefficient calculation circuit that determines the weighting coefficient a and the weighting coefficient b, and 53 is a multiplier a, 50 and a multiplier b,
An adder for adding the output result of 51, an output terminal 54 for outputting the operation result of the adder 53 to the determiner 36, and an addition for generating an error signal by taking the difference between the operation result of the adder 53 and the desired signal. 56, a switch circuit for switching the input signal sequence of the adder 55 between the output signal sequence of the determiner 36 and the reference signal sequence, 57 the determination output signal input terminal to which the output result of the determiner 36 is input, 58 the reference signal It is a series input terminal.

The operation of the adaptive diversity equalizer according to the first embodiment will be described with reference to FIGS. In the adaptive diversity equalizer shown in FIG. 1, received signals output from antennas a and 1 and antennas b and 2 are converted into baseband signals by detection circuits a and 3 and detection circuits b and 4, respectively. . The output signals of the detection circuits a and 3 are stored in the reception signal memories a and 30, and the output signals of the detection circuits b and 4 are stored in the reception signal memories b and 31. Further, these baseband signals are output to the control unit 32.
In the control unit 32, the UW2 shown in FIG.
After performing frame synchronization using 1, the control signal is output to the reception signal memories a and 30 and the reception signal memories b and 31. The reception signal memories a and 30 are controlled by this control signal.
The received signal corresponding to the burst is output to the decision feedback type adaptive equalizers a and 33. The received signal memories b and 31 are also
With this control signal, the received signal corresponding to the burst is output to the decision feedback type adaptive equalizers b and 34.

Decision feedback type adaptive equalizers a and 33 shown in FIG.
Then, with respect to the received signal converted into the baseband, the characteristics of the transmission path are estimated using the UW 21 at the beginning of each burst and the received signal corresponding to the UW 21, and the FFa unit 37 and the FBa unit 3
The tap coefficient of 8 is converged (training mode).
At this time, the input signal sequence of the FBa unit 38 and the adder 4
The input signal sequence of No. 7 is data having no determination error, that is, a reference signal sequence determined from the known signal sequence of UW21. Next, the random data section 22 is equalized (tracking mode). At this time, the input signal series of the FBa unit 38 and the input signal series of the adder 47 become the output signal series of the determiner 36 input from the determination output signal input terminal 44.

In the tap coefficient updating circuits a and 46, the input signal series of the decision feedback type adaptive equalizers a and 33 and the adder 47 are added.
Of the FFa unit 37 for each symbol according to the tap coefficient update algorithm using the output signal sequence of
The tap coefficient of the FBa unit 38 is updated. When the Kalman filter algorithm shown in the section of the description of the conventional example is used for this tap coefficient updating algorithm, the input signal of the decision feedback type adaptive equalizers a and 33 is given by the following equation.

[0032]

[Equation 5]

Further, when the tap coefficients of the decision feedback type adaptive equalizers a and 33 are made to correspond to the two equations, Ca _M (n) is obtained.
The output of the decision feedback type adaptive equalizer a, 33 at time nT (n = 0, 1, 2, 3 ...) is given by the following equation.

[0034]

[Equation 6]

Similarly, the output of the decision feedback adaptive equalizers b and 34 is given by the following equation.

[0036]

[Equation 7]

In the weighting synthesis circuit 35 shown in FIG. 3, the output signals of the decision feedback adaptive equalizers a and 33 input from the input terminals a and 48 are multiplied by the coefficient a determined by the coefficient calculation circuit 52. Multiply by a and 50 to perform weighting.
Further, the coefficient calculation circuit 52 is applied to the output signals of the decision feedback type adaptive equalizers b and 34 input from the input terminals b and 49.
The coefficient b determined by is multiplied by multipliers b and 51 to perform weighting. In the adder 53, the multipliers a and 50 and the multipliers b and 5
When the output result of 1 is added and the received signal corresponding to the random data 22 is being equalized, the result is output from the output terminal 54 to the determiner 36.

In the decision unit 36 shown in FIG. 1, the output terminal 54
Hard decision is performed on the output signal from the decision feedback adaptive equalizer a, 33, and decision feedback adaptive equalizer b, 3
4, output to the weighting synthesis circuit 35.

In the switch circuit 56 in FIG. 3, the UW
21 is equalized, the reference signal sequence input from the reference signal sequence input terminal 58 is output, and the received signal corresponding to the random data 22 is equalized. , And outputs the output signal sequence of the determiner 36 input from the determination output signal input terminal 57.

The adder 55 takes the difference between the calculation result of the adder 53 and the output of the switch circuit 56, that is, the desired signal, generates an error signal, and outputs the result to the coefficient calculation circuit 52.

The coefficient calculation circuit 52 uses the outputs of the decision feedback adaptive equalizers a and 33 and the decision feedback adaptive equalizers b and 34 and the output signal sequence of the adder 53, that is, the error signal, to update the tap coefficient. For each symbol, a coefficient a and a coefficient b of the multipliers a and 50 and multipliers b and 51 are calculated using an algorithm or the like.
To update. When the Kalman filter algorithm shown in the description of the conventional example is used for this tap coefficient updating algorithm, the input signal of the weighting / synthesizing circuit 35 is given by the following equation.

[0042]

[Equation 8]

Further, when the coefficients a and b are made to correspond to the two expressions, the following expressions are obtained.

[0044]

[Equation 9]

Then, time nT (n = 0, 1, 2, 3
The output of the adder 53 in () is given by the following equation.

[0046]

[Equation 10]

During random data equalization, the output of this adder is hard-decided by the decision device 36, and the desired signal d
(N). At this time, the decision feedback type adaptive equalizer a, 3
3, the error signals of the decision feedback type adaptive equalizers b and 34, and the weighting synthesis circuit 35 are given by the following equations.

[0048]

[Equation 11]

Each tap coefficient updating circuit, based on each error signal, updates the tap coefficient Ca after updating.
Define _M (n), Cb _M (n), and Cc (n).

In the above embodiment, the Kalman filter is used as the tap coefficient updating algorithm. However, this is performed by using a ZF (zero forcing) algorithm, an LMS (Least mean square) algorithm, and an RLS (Recursive least) algorithm.
A tap coefficient updating algorithm such as a square algorithm may be used.

In the above embodiment, the case where the adaptive equalizer of each branch is the decision feedback type adaptive equalizer has been described. However, the adaptive equalizer of each branch is a linear adaptive equalizer having no FB section. But good.

As described above, in the first embodiment, the decision feedback type adaptive equalizers a and 33 and the decision feedback type adaptive equalizers b and 3 are used.
4. The tap coefficient is updated independently of each other. Therefore, for example, in the case of the Kalman filter algorithm described in the present embodiment, the calculation amount is proportional to the square of the number of taps, but the number of branches is K, the number of taps in the FF section of each branch is L ₁ , and the number of taps in the FB section is When the number of taps is L ₂ , the total calculation amount is expressed by the following equation.

[0053]

[Equation 12]

Here, A represents a proportional multiplier. In contrast,
In the adaptive diversity equalizer shown in the conventional example, the total calculation amount is represented by the following equation.

[0055]

[Equation 13]

From the above equation, for example, L ₁ = 4, L ₂ = 1 and K
In the case of = 2, comparing the present embodiment with the conventional example, the calculation amount in the present embodiment can be set to 2/3 of the conventional example from the following equation.

[0057]

[Equation 14]

Second Embodiment FIG. 4 is a block diagram showing the configuration of a second embodiment of the present invention. In the figure, the same parts as those in the conventional example and the first embodiment are designated by the same reference numerals and the description thereof will be omitted. In FIG. 4, reference numeral 60 is a reception signal memory a for storing one burst of a reception signal received from the antennas a and 1 and converted into baseband signals by the detection circuits a and 3, and 61 is a detection circuit b received from the antennas b and 2. , A reception signal memory b for accommodating one burst of the reception signal converted into a baseband signal by 62, 62 is a control unit for controlling the operations of the reception signal memories a, 60, and reception signal memories b, 61, and 63 is a reception signal memory a, 60 and a decision feedback type adaptive equalizer a, 64 using the received baseband signals stored in the received signal memories b, 61 as input signals
Are received signal memories a and 60 and received signal memories b and 6
The decision feedback type adaptive equalizers b and 65 which use the received baseband signal stored in 1 as an input signal are the received signal memories a and 60 and the received signal memories b and 61 in the decision feedback type adaptive equalizers a and 63. Transversal filter a (FFa
Part), 66 is a decision feedback type adaptive equalizer a, 63,
Transversal filter a (FB of the feedback section
a section), 67 is an adder for adding the output results of the FFa section 65 and the FBa section 66 in the decision feedback type adaptive equalizers a and 63, and 68 is a received signal memory in the decision feedback type adaptive equalizers b and 64. a, 60 and a transversal filter b (FFb) in the feedforward section, which uses the outputs of the received signal memories b and 61 as input signals, and 69 is a transversal filter in the feedback section in the decision feedback adaptive equalizers b and 64. b (FBb unit) and 70 are FFb unit 68 and FBb unit 6 in decision feedback type adaptive equalizers b and 64.
9 is an adder that adds the output results of 9.

FIG. 5 is a block diagram showing the configuration of the decision feedback type adaptive equalizers a and 63. In the figure, the same parts as those shown in the conventional example and FIG.
In FIG. 5, reference numeral 71 is a reception signal input terminal a to which the output signals of the reception signal memories a and 60 are input, and 72 is a reception signal input terminal b to which the output signals of the reception signal memories b and 61 are input,
Reference numeral 73 denotes a tap coefficient updating circuit a which determines tap coefficients of the FFa unit 65 and the FFb unit 66 for each symbol.

The operation of the adaptive diversity equalizer of the second embodiment will be described. In the adaptive diversity equalizer shown in FIG. 4, the reception signals output from the antennas a and 1 and the antennas b and 2 are converted into baseband signals by the detection circuits a and 3 and the detection circuits b and 4, respectively. The output signals of the detection circuits a and 3 are stored in the reception signal memories a and 60, and the output signals of the detection circuits b and 4 are stored in the reception signal memories b and 61. Further, these baseband signals are output to the control unit 62. The control unit 62 outputs a control signal to the reception signal memories a and 60 and the reception signal memories b and 61 after frame synchronization is performed using the UW 21 shown in FIG. 8 at the beginning of each burst. The reception signal memories a and 60 output the reception signals corresponding to the bursts to the decision feedback type adaptive equalizers a and 63 and the decision feedback type adaptive equalizers b and 64 by this control signal. The received signal memories b and 61 also output received signals corresponding to bursts to the decision feedback adaptive equalizers a and 63 and the decision feedback adaptive equalizers b and 64 by this control signal. At this time, the received signal memories a and 60 are the decision feedback type adaptive equalizers a and 63.
For example, at time t = nT (n = 0, 1, 2,
3, ...) sampling data is output. Also, sampling data at different times, for example, at time t = (n + 1/2) T is output to the decision feedback adaptive equalizers b and 64. Similarly to the reception signal memories a and 60, the reception signal memories b and 61 also output sampling data at the time t = nT to the decision feedback adaptive equalizers a and 63. For the decision feedback adaptive equalizers b and 64,
The sampling data at different times t = (n + 1/2) T is output.

Decision feedback type adaptive equalizers a and 63 shown in FIG.
Then, the received signals sampled at the same time in each diversity branch output from the received signal memories a and 60 and the received signal memories b and 61 are transmitted using the UW 21 at the beginning of each burst and the corresponding received signals. The characteristics of the road are estimated and the tap coefficients of the FFa unit 65 and the FBa unit 66 are converged (training mode). At this time, the input signal series of the FBa unit 66 and the input signal series of the adder 47 are data without a judgment error, that is, UW2.
It is a reference signal sequence defined from one known signal sequence. Next, the random data section 22 is equalized (tracking mode). At this time, the input signal sequence of the FBa unit 66 and the input signal sequence of the adder 47 become the output signal sequence of the determiner 36 input from the determination output signal input terminal 44.

In the tap coefficient updating circuits a and 73, the input signal series of the decision feedback type adaptive equalizers a and 63 and the adder 47 are added.
Of the FFa unit 65 for each symbol according to the tap coefficient update algorithm using the output signal sequence of
The tap coefficient of the FBa unit 66 is updated. When the Kalman filter algorithm shown in the section of the description of the conventional example is used for this tap coefficient updating algorithm, the input signal of the decision feedback type adaptive equalizers a and 63 is given by the following equation.

[0063]

[Equation 15]

Here, equations 26 and 27 show double oversampling reception data for one burst accumulated in the reception signal memories a and 60 and the reception signal memories b and 61.

Further, when the tap coefficients of the decision feedback type adaptive equalizers a and 63 are made to correspond to the two equations and Ca _M (n), the decision feedback at time nT (n = 0, 1, 2, 3 ...) The outputs of the shape adaptive equalizers a and 63 are 14 equations.

The decision feedback type adaptive equalizers b and 64 have the same configuration and operation as the decision feedback type adaptive equalizers a and 63.
The input signal here is delayed by 1/2 symbol from the input signal of the decision feedback type adaptive equalizers a and 63.
Then, the output becomes the expression 15.

In the weighting / synthesizing circuit 35 shown in FIG.
The output signals of the decision feedback type adaptive equalizers a and 63 input from the input terminals a and 48 are multiplied by the coefficient a determined by the coefficient calculation circuit 52 by the multipliers a and 50, and weighted.
Also, the coefficient calculation circuit 52 is applied to the output signals of the decision feedback type adaptive equalizers b and 64 input from the input terminals b and 49.
The coefficient b determined by is multiplied by multipliers b and 51 to perform weighting. In the adder 53, the multipliers a and 50 and the multipliers b and 5
When the output result of 1 is added and the received signal corresponding to the random data 22 is being equalized, the result is output from the output terminal 54 to the determiner 36.

The decision unit 36 makes a hard decision on the output signal from the output terminal 54, and the results thereof are decision feedback type adaptive equalizers a and 63, decision feedback type adaptive equalizers b and 64, and a weighting synthesis circuit. To 35.

The switch circuit 56 outputs the reference signal sequence input from the reference signal sequence input terminal 58 to equalize the received signal corresponding to the random data 22 while equalizing the received signal corresponding to the UW 21. During equalization, the output signal series of the determiner 36 input from the determination output signal input terminal 57 is output.

The adder 55 takes the difference between the calculation result of the adder 53 and the output of the switch circuit 56, that is, the desired signal, generates an error signal, and outputs the result to the coefficient calculation circuit 52.

The coefficient calculation circuit 52 uses the outputs of the decision feedback adaptive equalizers a and 63 and the decision feedback adaptive equalizers b and 64 and the output signal series of the adder 53, that is, the error signal, to update the tap coefficient. For each symbol, a coefficient a and a coefficient b of the multipliers a and 50 and multipliers b and 51 are calculated using an algorithm or the like.
To update. When the Kalman filter algorithm shown in the section of the description of the conventional example is used for this tap coefficient updating algorithm, the input signal of the weighting synthesis circuit 35 becomes 16 equations when corresponding to 1 equation. Further, when the coefficient a and the coefficient b are associated with the two expressions, the expression 17 is obtained. Then, at time nT (n =
The output of the adder 53 in 0, 1, 2, 3, ... At the time of random data equalization, the output of this adder is hard-decided by the decision device 36 and is used as the desired signal d (n). At this time, the error signals of the decision feedback type adaptive equalizers a and 63, the decision feedback type adaptive equalizers b and 64, and the weighting synthesis circuit 35 are equations 19, 20 and 21, respectively.

Each tap coefficient updating circuit is based on each error signal, and each of the updated tap coefficients Ca _M (n),
Define Cb _M (n) and Cc _M (n).

In the above embodiment, the Kalman filter is used as the tap coefficient updating algorithm. However, this is performed by using a ZF (zero forcing) algorithm, an LMS (Least mean square) algorithm, and an RLS (Recursive least) algorithm.
A tap coefficient updating algorithm such as a square algorithm may be used.

In the above embodiment, the case where the adaptive equalizer of each branch is the decision feedback type adaptive equalizer has been described. However, the adaptive equalizer of each branch is a linear adaptive equalizer having no FB section. But good.

As described above, in the second embodiment, the decision feedback type adaptive equalizers a and 63 and the decision feedback type adaptive equalizers b and 6 are used.
4. The tap coefficient is updated independently of each other. Therefore, for example, in the case of the Kalman filter algorithm described in the present embodiment, the calculation amount is proportional to the square of the number of taps, but the number of branches is K, the number of symbols in the FF section of each branch is L _1s , and the number of symbols in the FB section is The number of taps is L ₂ ,
When the number of oversamples is S, the total calculation amount is expressed by the following equation.

[0076]

[Equation 16]

Here, A represents a proportional multiplier. In contrast,
In the adaptive diversity equalizer shown in the conventional example, the total calculation amount is represented by the following equation.

[0078]

[Equation 17]

From the above equation, for example, L _1s = 2, L ₂ = 1 and K
= 2 and S = 2, a comparison between the present embodiment and the conventional example shows that from Equation 24, the calculation amount in the present embodiment is 2/3 that of the conventional example.
Can be

In the second embodiment, since the decision feedback adaptive equalizer sections combine received signals of different branches sampled at the same timing, the probability that the signal level drops due to fading becomes small. In addition, the characteristics of each decision feedback type adaptive equalizer unit are improved,
As a result, good characteristics can be obtained.

[0081]

As described above, according to the invention of claim 1, the input signal is divided, and the tap coefficients are independently updated in the decision feedback adaptive equalizer unit for each divided input signal sequence. Since the output of each decision feedback adaptive equalizer unit is weighted and combined, the amount of calculation required to update the overall tap coefficient is reduced, and the number of symbols required to converge the tap coefficient in the training mode is short. Become. In addition, the characteristics of a decision feedback adaptive equalizer block are temporarily deteriorated due to fading, etc., and the decision results are continuous in the output results of the decision feedback adaptive equalizer block alone, and the tap coefficient update algorithm diverges. Even if the above occurs, the output of each decision feedback adaptive equalizer section is weighted and synthesized, and the tap coefficient is updated to update the tap coefficient of each decision feedback adaptive equalizer section using the decision result as a reference signal. The frequency of divergence of the algorithm is reduced and the followability of each decision feedback type adaptive equalizer unit is improved, and as a result, an adaptive equalizer having good equalization characteristics can be obtained.

As described above, according to the invention of claim 2, the tap coefficient is updated independently in each decision feedback type adaptive equalizer section of each branch, and the output of each decision feedback type adaptive equalizer section is updated. In order to perform weighted synthesis, the number of symbols required for tap coefficient convergence in the training mode is shorter than that of the conventional adaptive diversity equalizer, and the number of symbols required for tap coefficient convergence in the training mode is reduced. An equalizer can be obtained. In addition, even if the characteristics of a certain branch is temporarily deteriorated due to fading, etc., and the judgment error continues in the output result of the single branch, and the tap coefficient update algorithm causes divergence, etc.
The output of each decision feedback type adaptive equalizer section is weighted and synthesized, and the tap coefficient update algorithm diverges in order to update the tap coefficient of each decision feedback type adaptive equalizer section using the decision result as a reference signal. As the number decreases, the followability of each decision feedback type adaptive equalizer unit is improved, and as a result, an adaptive diversity equalizer having good equalization characteristics can be obtained.

As described above, according to the invention of claim 3, the tap coefficient is updated independently in the decision feedback adaptive equalizer section of each branch, and the output of each decision feedback adaptive equalizer section is updated. Since the results are weighted and combined, the amount of calculation required to update the overall tap coefficient is reduced, and
It is possible to obtain an adaptive diversity equalizer that requires a short number of symbols to converge tap coefficients in the training mode. In addition, the characteristics of a certain decision feedback adaptive equalizer part are temporarily deteriorated due to fading, etc., and the decision results are continuous in the output result of the decision feedback adaptive equalizer part alone, and the tap coefficient update algorithm diverges. Even if the above occurs, the output of each decision feedback adaptive equalizer section is weighted and synthesized, and the tap coefficient is updated to update the tap coefficient of each decision feedback adaptive equalizer section using the decision result as a reference signal. The frequency of divergence of the algorithm is reduced and the followability of each decision feedback type adaptive equalizer unit is improved, and as a result, an adaptive diversity equalizer having good equalization characteristics can be obtained.
Further, in each decision feedback type adaptive equalizer section, and in the second embodiment, in each decision feedback type adaptive equalizer section, in order to synthesize the reception signals of different branches sampled at the same timing, fading is performed. Since the probability that the signal level will drop is small, the characteristics of each decision feedback type adaptive equalizer unit are improved, and as a result, good characteristics can be obtained.

[Brief description of drawings]

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

FIG. 2 is a block diagram showing an example of the internal configuration of the decision feedback adaptive equalizer shown in FIG.

FIG. 3 is a block diagram showing an example of the internal configuration of the weighting synthesis circuit shown in FIG.

FIG. 4 is a second embodiment of the adaptive diversity equalizer of the present invention.
It is a configuration block diagram showing.

5 is a block diagram showing an internal configuration example of the decision feedback adaptive equalizer shown in FIG. 4. FIG.

FIG. 6 is a configuration block diagram showing a conventional adaptive diversity equalizer.

7 is a block diagram showing an example of the internal configuration of the decision feedback adaptive equalizer shown in FIG.

FIG. 8 is a diagram showing an example of a burst format used in mobile communication.

[Explanation of symbols]

1 antenna a 2 antenna b 3 detection circuit a 4 detection circuit b 5 reception signal memory a 6 reception signal memory b 7 control unit 8 adaptive equalizer 9 transversal filter a (FFa unit) of feedforward unit 10 of feedforward unit Transversal filter b (FFb section) 11 Transversal filter (FB section) of feedback section 12 Adder 13 Judgmentor 14 Output signal terminal 15 Input terminal 16 Input terminal 17 Tap coefficient update circuit 18 Switch circuit 19 Reference signal sequence input terminal 20 Adder 21 Unique word 22 Random data part 30 Received signal memory a 31 Received signal memory b 32 Control part 33 Decision feedback type adaptive equalizer a 34 Decision feedback type adaptive equalizer b 35 Weighting synthesis circuit 36 Judgmenter 37 Feedforward Department of Transbar Rufiruta a (FFa portion) of 38 feedback unit transversal filter a
(FBa unit) 39 Adder 40 Transversal filter b of feedforward unit (FFb unit) 41 Transversal filter b of feedback unit
(FBb section) 42 Adder 43 Received signal input terminal 44 Judgment output signal input terminal 45 Equalization filter output terminal 46 Tap coefficient update circuit a 47 Adder 48 Input terminal a 49 Input terminal b 50 Multiplier a 51 Multiplier b 52 Coefficient calculation circuit 53 Adder 54 Output terminal 55 Adder 56 Switch circuit 57 Judgment output signal input terminal 58 Reference signal sequence input terminal 60 Received signal memory a 61 Received signal memory b 62 Control unit 63 Decision feedback adaptive equalizer a 64 Decision feedback type adaptive equalizer b 65 Transversal filter a (FFa unit) of the feedforward unit 66 Transversal filter a of the feedback unit
(FBa section) 67 Adder 68 Transversal filter b of feedforward section (FFb section) 69 Transversal filter b of feedback section
(FBb section) 70 adder 71 received signal input terminal a 72 received signal input terminal b 73 tap coefficient update circuit a

Claims (3)

[Claims] 1. A means for dividing an input signal sequence, an equalization filter unit having a feed-forward unit and a feedback unit each having a tapped delay circuit as an input, and an input signal sequence after division, and a tap coefficient calculation unit. And a plurality of decision feedback adaptive equalizer units for updating the tap coefficients of the equalization filter unit based on the output result of the data judgment unit according to the tap coefficient update algorithm, A decision feedback adaptive equalizer section is provided with a combining circuit for weighting and combining the output signals, and a data judging section for judging the output result of the combining circuit and making the final output data. Adaptive equalizer. 2. An equalization filter having a plurality of antennas, a plurality of detection circuits for detecting a plurality of received waves, and each of the detected signals as an input and having a tapped delay circuit in a feedforward section and a feedback section. A plurality of decision feedback adaptive equalizers for updating the tap coefficients of the equalization filter section based on the output result of the data decision section according to the tap coefficient update algorithm. Section, a synthesizing circuit for weighting and synthesizing the output signals of the plurality of decision feedback adaptive equalizer sections, and a data deciding section for deciding an output result of the synthesizing circuit to obtain final output data. An adaptive equalizer characterized in that 3. A plurality of antennas, a plurality of detection circuits for detecting a plurality of received waves, and a signal having the same oversampling timing as each input with respect to each signal after oversampled detection, An equalization filter unit having a tapped delay circuit in a feedforward unit and a feedback unit, and a tap coefficient calculation unit are provided, and the tap coefficient calculation unit is based on the output result of the data determination unit according to a tap coefficient update algorithm. Decision feedback type adaptive equalizer section for updating the tap coefficient of each section, a combining circuit for weighting and combining each output signal of the plurality of decision feedback type adaptive equalizer sections, and an output result of the combining circuit The adaptive equalizer is characterized by including a data determination unit that determines the final output data.