JPH0795715B2 - Equalizer - Google Patents

Description

The present invention relates to an equalizer provided in a receiver used for digital wireless communication, digital mobile communication and the like.

[Conventional technology]

Fig. 3 shows, for example, "Measurement and Control," Vol. 25, No. 12, p22-28.
It is a block diagram which shows the structure of the conventional equalizer shown in (Showa 61, December). In the figure, 1 is a received signal input terminal, 10 to 13 are delay elements for delaying the signal input to the input terminal 1 by time T, and 20 to 23 are signals delayed by the delay elements 10 to 13 and the input signal 1. To the weight a ₀ (n) ~ a
A weighting circuit for multiplying and outputting _N (n) (hereinafter referred to as tap coefficient), 30 is an adder for adding and outputting the outputs of the weighting circuits 20 to 23, and 40 is for outputting the addition result of the adder 30. Output terminal, 50 is a reference signal d (n) of known series
A reference signal input terminal to which is input, 31 is an adder for taking the difference between the reference signal input from the reference signal input terminal 50 and the output of the adder 30, and 60 is an error signal ε for the output of the adder 31.
This is an error signal output terminal for outputting as (n).

The part including the delay elements 10 to 13, the weighting circuits 20 to 23, and the adder 30 is referred to as an equalizer.

Next, the operation will be described.

The received signal χ (n) (n: an argument indicating the discrete time t = n) input to the received signal input terminal 1 is divided into two parts, one is input to the delay element 10, and the other is tap coefficient a ₀ (n). Is input to the weighting circuit 20 having the weights, weighted, and output. Similarly, the output of the delay element 10 becomes the received signal χ (n-1) at t = n-1, one of which is input to the delay element 11 and the other is input to the weighting circuit 21 having the tap coefficient a ₁ (n). And weighted and output. When such an operation is performed for all N delay elements and all N + 1 weighting circuits, the output y (n) of the adder 30 at time n is given by the following equation.

This is a linear time-varying filter, and the Z-transform of the transfer function of this time-varying filter is as follows.

By the way, in the case of an unknown transmission line whose characteristics change every moment such as a fading line, an equalizer is introduced to correct the fading characteristics every moment in order to obtain good transmission quality. There is a need. That is, the tap coefficient a _i (n) (i = 0.1,
..., N) is a function of time n and must be adaptively controlled to be optimal at each time.

Here, as a typical example of the control method, a method using the reference signal d (n) input from the reference signal input terminal 50 shown in FIG. 3 will be shown.

The reference signal is a known series, and while the series is being transmitted, the characteristics of the transmission line are estimated and the distortion of the transmission line is corrected by the equalizer shown in FIG. 3 to realize the ideal transmission characteristics. To do. Usually, it is given by the difference between the reference signal d (n) and the output signal y (n). The tap coefficient is adaptively controlled by the error signal ε (n) = d (n) -y (n) (3). Various adaptive control algorithms have been considered, and in the LMS algorithm, a (n + 1) = a (n) + με (n) χ (n) (4) where μ is a parameter called tap gain ε (n) is (3 ) Is calculated as the value given by However, χ ^T (i) is given as a transposed matrix of χ (i).

There are two equalization methods after correcting the transmission path characteristics with the known series. One is a method of fixing the tap coefficient of the equalizer determined by using a known sequence to equalize the random data part, and the other is an output signal in the random data part by initializing the tap coefficient. The result of determining y (n) is the reference signal d
(N) is a method of performing equalization adaptively.

Regarding the construction method of the equalizer, in addition to the feed-forward type construction shown in FIG. 3, the feedback type shown in FIG.
Also, a decision feedback type configuration method shown in FIG. 5 is also often used, and a decision feedback type configuration method using a combination of FIGS. 3 and 5 together with a feedforward section and a feedback section is also often used. In FIGS. 4 and 5, the same reference numerals as those in FIG. 3 indicate the same or corresponding portions.

Further, reference numeral 70 in FIG. 5 is a judging device. For example, in the case of binary judgment, +1 or -1 is judged depending on which of the binary data (+1, -1) the output result of the adder 30 is, The determination result is used as the output signal y (n) and also as the input signal of the delay element 10 of the feedback section. The feedback type shown in FIG. 4 is used as the input signal of the delay element 10 of the feedback section without making a decision by the decision device shown in FIG.

Further, FIG. 6 shows a configuration example of a packet including a known sequence for estimating the transmission path characteristic and a random sequence.

[Problems to be Solved by the Invention]

Since the conventional equalizer is configured as described above, when the change in the characteristics of the transmission line is fast, the transmission line is estimated only by the known sequence, and the tap coefficient is fixed to equalize the data part. If this is done, the quality of the data will decrease as the data part becomes rearward, and if the tap coefficient obtained in the known sequence part is used as the initial value and adaptive equalization is also performed in the data part, the amount of calculation for obtaining the tap coefficient will be enormous. Therefore, there is a problem that the transmission speed of data that can be actually transmitted is limited.

The present invention has been made in order to solve the above problems, and can follow the fluctuation even when the characteristic fluctuation of the transmission line is fast, and can use the tap coefficient obtained in the known sequence section as an initial value in the data section. It is an object of the present invention to obtain an equalizer in which the amount of calculation does not become huge and the transmission speed does not decrease even when adaptive equalization is performed.

[Means for Solving the Problems]

The equalizer according to the present invention, when updating the tap coefficient,
Two or more types of tap coefficient update algorithms are built in, one for each symbol operation, and the other for intermittent operation in a slow cycle.

[Action]

In the present invention, since two or more types of tap coefficient updating algorithms are incorporated, the channel coefficient is estimated by a known sequence to determine the tap coefficient, and the tap coefficient is used as an initial value in the data section once every several symbols. The algorithm that updates the tap coefficient only allows the tap coefficient to follow slow channel fluctuations.

〔Example〕

 An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of an equalizer according to an embodiment of the present invention. In the figure, the same reference numerals as those in FIG. 3 denote the same or corresponding parts, 10 to 15 are delay elements, and 20 to 27 are. Weighting circuits, 30 to 32 are adders, and 70 is a determiner. 100,10
1 is calculated by a certain tap coefficient update algorithm from the equalized output output to the output terminal 40 and the input signal χ (n) indicated by 110 and 111 which is the same as the received signal input to the input terminal 1. This is a calculation unit that outputs tap coefficients C _i (n) and D _i (n). A switch 150 selects either the output result of the determiner 70 or the reference signal d (n).
Further, reference numeral 200 denotes a feedforward section, and 300 denotes a feedback section, which constitute a decision feedback type in which the components of the equalizer shown in FIGS. 3 and 5 are used together.

Next, the operation will be described.

The received signal χ (n) input to the input terminal 1 is filtered by the equation (1) in the description of the conventional example, and is output to the adder 30 output of the feedforward unit 200. Is output as. Where C _i (n) (i = 0, ..., N)
Is a tap coefficient of the feedforward unit 200, and the delay elements 10 to 12 have a delay amount Tp represented by T / Tp = p (integer), where T is the symbol period. And p
The case of = 1 corresponds to the conventional example shown in FIG.

On the other hand, the determiner 70 determines and outputs the equalized output output to the output terminal 40 in the same manner as the conventional method. When the known series is used as the input signal to the feedback section 300, the switch 150 operates so as to adopt the known series input from the reference signal input terminal 50 and adopt the output of the determiner 70 in the data section.

By the way, two types of algorithm operation units 100 and 101 are incorporated as tap coefficient update algorithms. Now, the calculation unit 100 of the tap gain update algorithm (A) estimates the transmission path for each symbol of the known sequence, and the tap coefficient is C (l) = {C at the end time t = 1 of the known sequence.
₀ (l), C ₁ (l), ..., C _N1 (l)}, D (l) = {D ₀
(L), D ₁ (l), ..., D _N2 (l)}. Here, N ₁ +1 and N ₂ +1 are feedforward units 200
And the number of taps of the feedback unit 300.

Next, the arithmetic unit 101 of the tap coefficient update algorithm (B) performs equalization of the data part using the tap coefficient values C (l) and D (l) as initial values, but at this point the tap coefficients are almost transmitted. Since the path estimation has been completed, the data section only needs to correct the gradual fluctuation of the Doppler frequency or the like.
Therefore, it suffices to estimate the rate of change and intermittently update the tap coefficient. That is, assuming that the tap coefficient is updated once every 5 symbols, for example, the 5th symbol, the 10th symbol, the 15th symbol of the data part
······································ Tap coefficient is updated using a symbol of the data part of multiples of 5. That is, since the transmission line does not change rapidly between 2 to 4 symbols, even such an intermittent operation can sufficiently follow the slow transmission line.

FIG. 2 shows a packet configuration example in this operation.

By the way, as a tap coefficient update algorithm, it is necessary to use a known sequence that is as short as possible for channel estimation to improve the data transmission efficiency, and therefore an algorithm that converges at high speed is required. For example, the Kalman filter shown in the conventional example is a typical example. This algorithm converges extremely quickly, but the number of calculations becomes enormous. However, if this algorithm is used for each symbol only in the case of a short known sequence and the data part is made to perform an intermittent operation once for n symbols, the calculation amount per symbol is 1 / n. The amount does not significantly affect the data transmission rate.

In this embodiment, the error signal corresponding to the equation (3) is calculated from the equalized output and its judgment value in the arithmetic units 100 and 101 of the tap coefficient updating algorithm, which is the judgment unit 70 in FIG. It can be obtained by using the result of.

Further, in the above embodiment, the case where the Kalman filter is used for the algorithm is shown, but the algorithm used is not limited to this, for example, the LMS method,
It may be an adaptive algorithm such as a learning identification method.

Further, in the above-described embodiment, two arithmetic units for performing the tap coefficient updating algorithm are arranged in parallel and the respective coefficients are respectively in charge of updating the coefficient. ) Is used, the same effect can be obtained by changing only the software with the same birdware.

Further, in the above embodiment, two arithmetic units of the tap coefficient updating algorithm are arranged, but the number of arithmetic units arranged is not limited to this, and a plurality of arithmetic units may be arranged. In this case, one of them may be arranged. It may be selected appropriately.

In addition, in the above embodiment, the equalizer including both the feedforward unit and the feedback unit is shown.
It is not always necessary to provide both of them, and even if only one of them is provided, the same effect is obtained.

Further, in the above-mentioned embodiment, the configuration using the baseband transmission as a model is shown, but the configuration is extended to a modulation system such as orthogonal modulation, and a two-dimensional baseband model orthogonal to the in-phase is considered. The same effect can be achieved by configuring an equalization unit that takes into consideration interference from the in-phase component to the quadrature component and interference from the quadrature component to the in-phase component.

〔The invention's effect〕

As described above, according to the equalizer according to the present invention, a plurality of arithmetic units having different arithmetic cycles are provided in the arithmetic unit for performing the tap coefficient updating algorithm of the equalizer, and the arithmetic operation for each symbol and the intermittent operation are performed. Since the calculation with the slow cycle is performed, it is possible to follow the fluctuation of the transmission path, further reduce the number of calculations, and have an effect of being able to cope with high-speed transmission.

[Brief description of drawings]

FIG. 1 is a block diagram of an equalizer according to an embodiment of the present invention, FIG. 2 is a timing chart showing the operation of the present invention, and FIG. 3 is a block diagram of a conventional feedforward equalizer. FIG. 4 is a block diagram of a conventional feedback equalizer, FIG. 5 is a block diagram of a conventional decision feedback equalizer, and FIG. 6 is a diagram showing a packet configuration example. 1 is a signal input terminal, 10 to 15 are delay elements, 20 to 27 are weighting circuits, 30 to 32 are adders, 40 is an equalization output terminal, 50 is a reference signal input terminal, 70 is a decision unit, and 100 and 101 are first And a calculation unit of the second tap coefficient updating algorithm, 150 is a switch, 200 is a feedforward unit, and 30 is a feedback unit. The same reference numerals in the drawings indicate the same or corresponding parts.

Claims (1)

[Claims] 1. An equalizer comprising a plurality of delay elements, a weighting circuit, and an adder, which multiplies a signal input from an input terminal a plurality of times according to a set tap coefficient, and an output of the equalizer. Based on an input signal and an output signal of the determiner, which receives the signal and makes a determination, a switch means for selecting one of the output of the determiner and a reference signal which is a known signal sequence, A first calculation unit that estimates a transmission channel for each symbol of a known sequence to calculate a tap coefficient, and a transmission channel is estimated intermittently for each predetermined symbol using the tap coefficient calculated by the first calculation unit. And an arithmetic unit for calculating a tap coefficient by performing the above.