JPH04268831A - Diversity receiving method - Google Patents

Abstract

PURPOSE:To improve the converging speed of a tap coefficient by initializing the tap coefficient and executing again training in a second stage after the training in a first stage. CONSTITUTION:A memory is installed before input terminals 25, 26, and a received wave is stored in the memory by every branch, and at first, the training is executed for every branch, and a signal-to-noise ratio and the tap coefficient obtained from an equalizer are determined. Next, a transversal composite form is formed, and the tap coefficient multiplied by the square value of the signal-to- noise ratio is set as an initial value for every branch, and the training is executed again from the head of the memory. Namely, the tap coefficient and an equalization error obtained from tap coefficient control circuits 21, 22 in the first stage are transmitted to the tap coefficient control circuit 33, and these are set as the initial values of the circuit. Then, the training as the equalizer is executed for the small number of the taps of every branch, and after converging the tap coefficient to some extent, transversal composite form diversity is synthesized, and the training is executed again, and the converging speed of the tap coefficient is improved.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a diversity reception method used in digital mobile communications.

0002

2. Description of the Related Art Conventionally, a transversal combining diversity equalizer has been used as a method for simultaneously performing equalization processing and diversity combining processing used in digital mobile communications.

FIG. 3 shows a configuration diagram of a conventional decision feedback equalizer to which two-branch transversal composite diversity is applied. The following description also applies to a configuration in which the decision feedback equalizer is replaced with a Viterbi equalizer. In FIG. 3, 1 and 2 are input terminals, 3 and 4 are feedforward filters, 5 and 6 are adders, 7 is a judger, 8 is a feedback filter, 9 is a tap coefficient control circuit, 1
0 is an output terminal. In order to perform diversity combining processing and adaptive equalization processing on the baseband signals input to input terminals 1 and 2, such an equalizer uses a training signal and a feedforward filter 3 using an algorithm such as the recursive least squares method. and 4, calculate the tap coefficients of the feedback filter 8 and perform processing.

0004

Problem to be Solved by the Invention When equalization processing and diversity combining processing are performed simultaneously, the number of tap coefficients that need to be calculated at one time increases. Therefore, the convergence speed of the tap coefficients by the adaptive algorithm using the training signal decreases, resulting in a problem that the equalization characteristics and diversity combining gain deteriorate.

The present invention has been made to solve the above-mentioned problems, and an object of the present invention is to improve the convergence speed of tap coefficients when diversity combining processing and equalization processing are performed simultaneously.

[0006]

[Means for Solving the Problems] The present invention is characterized by first storing received waves in a memory for each branch, first training each branch, and obtaining a signal-to-noise ratio and a tap coefficient obtained from an equalizer. , then form a transversal composite form, initialize the tap coefficients multiplied by the square value of the signal-to-noise ratio for each branch, and perform training again from the beginning of the memory.

[0007]

[Operation] As described above, in the present invention, training is performed as an equalizer for a small number of taps in each branch, and after converging the tap coefficients to some extent, a transversal composite diversity with a large number of taps is formed. By performing training again, the convergence speed of tap coefficients is improved. At this time, the tap coefficients are initially set in proportion to the square value of the signal-to-noise ratio of each branch, so the maximum signal-to-noise ratio can be obtained in transversal synthesis, improving equalization characteristics and diversity characteristics. be able to.

[0008]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIGS. 1A and 1B are block diagrams of an embodiment of the present invention.

FIG. 1A shows a decision feedback equalizer with a two-branch transversal synthesis diversity configuration that implements the method of the present invention, and FIG. 1B shows a decision-feedback equalizer with a two-branch transversal synthesis diversity configuration that implements the method of the present invention. It is something. In FIG. 1A, 25, 26 are input terminals, 13, 14, 2
7, 28 are feedforward filters, 15, 16,
29, 30 are adders, 17, 18, 31 are determiners, 19
, 20, 32 are feedback filters, 21, 22,
33 is a tap coefficient control circuit, and 34 is an output terminal.

FIG. 1B shows an arrangement in which the required number of feedforward filters, determiners, and tap coefficient control circuits is reduced by providing switches S1, S2, and S3, and each reference number indicates the same as in FIG. 1A. Each switch S1,
In the first stage, S2 and S3 are connected to contact (a) to form an equalizer for each branch, and in the second stage, they are connected to contact (b) to form a transversal composite diversity circuit.

[0011] For training purposes, the input terminal 25
, 26 is provided with a memory (not shown).

First, as a first step, a decision feedback equalizer is formed for each branch, and training is performed to obtain respective tap coefficient vectors W1 and W2. Furthermore, the mean square value within one burst of the equalization error, which is the difference between the signal after equalization and the signal before equalization, is determined from the determiners 17 and 18,
The reciprocal number is the signal-to-noise ratio, and each square value is r1
and r2.

Next, as a second step, transversal composite diversity is formed. The tap coefficient of each branch is initially calculated by multiplying the tap coefficient obtained in the first step by the square value of the signal-to-noise ratio of each branch, and then multiplying it by a coefficient K such that the combined power is equal to the case of one branch. Set. That is, the tap coefficients and equalization errors obtained in the first stage are transferred from the tap coefficient control circuits 21 and 22 to the tap coefficient control circuit 3.
3 and set it as the initial value of the circuit. Assuming that the feedforward parts of W1 and W2 are F1 and F2, and the feedback parts are B1 and B2, the feedforward filter tap coefficient vectors F'1 and F'2 and the feedback filter tap coefficient vector B', which are initialized in the second stage, are , given by the following formula. F'1=K×F1×r1 F'2=K×F2×r2 B'=K×(B1×r1+B2×r2)

By using these F'1, F'2, and B' as initial values in the second stage, convergence can be made faster.

[0015] If the initial value of the second stage is only the result of the training of the first stage, when the received signal is very weak, the initial value will become meaningless and the operation in the second stage will be confused. , r1, r2 to stabilize the operation.

[0016]

As described above, according to the present invention, in transversal composite diversity with a large number of taps, after the first stage training, the second stage training is performed as shown in FIG.
By initializing the tap coefficients and performing training again in the step, it is possible to improve the convergence speed of the tap coefficients, so diversity synthesis processing and adaptive equalization processing can be performed simultaneously, resulting in diversity etc. with good characteristics. can be realized.

[Brief explanation of the drawing]

FIG. 1A is a configuration diagram of a decision feedback equalizer with a two-branch transversal composite diversity configuration that implements the present invention.

FIG. 1B is a block diagram of another embodiment of the present invention.

FIG. 2 is a diagram showing a state of convergence of tap coefficients.

FIG. 3 is a configuration diagram of a decision feedback equalizer to which conventional two-branch transversal composite diversity is applied.

[Explanation of symbols]

1, 2, 11, 12, 25, 26 input terminals 3, 4,
13, 14, 27, 28 Feedforward filter 5, 6, 15, 16, 29, 30 Adder 7, 17,
18, 31 Determiner 8, 19, 20, 32 Feedback filter 9,
21, 22, 33 tap coefficient control circuit 10, 23,
24, 34 Output terminal

Claims (1)

[Claims] Claim 1: In a diversity reception method in which received signals of multiple branches are stored in a memory for each branch and then combined by a transversal filter, and training for diversity combining processing and equalization processing is performed simultaneously, equalization is performed for each branch. Forming a vessel and training,
A diversity reception method characterized in that an initial value of a diversity synthesis filter is set in proportion to the square value of the signal-to-noise ratio obtained from the equalization error and the result of the training.