JPH11274991A - Estimation method for transmission line of equalizer - Google Patents

Abstract

PROBLEM TO BE SOLVED: To eliminate wrong transmission and to improve the code error rate by estimating a transmission line with the impulse response of the start state of a block defined as the initial value of the impulse response of the next block, if the block is decided incorrectly. SOLUTION: A data section is divided into M pieces of blocks, and a block quality calculation process is carried out after a transmission series estimation process, and the quality for every block is calculated. If the calculated block quality is satisfactory, the impulse response of the end state of the block is defined as the initial value of impulse response for the next block. If the block quality is poor, the impulse response of the shaft state of the block is defined as the initial value of impulse response of the next block and a transmission line is estimated. In such as estimation method for a transmission line, wrong transmissions of an adaptive equalizer can be eliminated.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for estimating a transmission path of an adaptive equalizer for removing waveform distortion caused by multipath fading in a high-speed burst transmission environment.

[0002]

2. Description of the Related Art A conventional transmission path estimation method will be described below. The conventional adaptive equalizer employs maximum likelihood sequence estimation (MLS).
FIG. 11E is an equalizer to which transmission path estimation is applied. FIGS. 11 and 12 show flowcharts (flowcharts) of a conventional transmission path estimation method, and FIG. 13 shows a burst format. The adaptive equalizer estimates the state of the transmission path at time t = jT (T is a symbol period, hereinafter simply referred to as time j) by using a quantity called an impulse response.

The impulse response is represented by a tap coefficient vector of (L + 1) taps, and the k-th tap coefficient is indicated as h (j, k) (k = 0 to L). However, i is used as a time index in the training process, and j is used in subsequent processes. First, a known PN sequence signal having a length (Ntr + L) symbol is received and a training process is performed to determine an initial value h (0, k) of an impulse response.

[0004] The initial value h (0, k) is "Equation 1" (Fig. 11).
(1), a training sequence ut (ik) composed of a PN sequence shifted by k symbols from the received signal r (1).
From the amount of correlation with Here, * represents a complex conjugate. In the data section next to the training process, a transmission channel estimation process, a replica generation process, and a transmission sequence estimation process are performed, and a determination signal is output.

[0005]

(Equation 1)

The LMS algorithm will be described as an example of the transmission path estimation process. The impulse response h (j,
k) is updated by “Equation 2” (Equation (2) in FIG. 12) using the determination signal u (j) and the error signal e (j).
μ is a step size parameter indicating the magnitude of correction when the tap coefficient is repeatedly updated.

[0007]

(Equation 2)

Next, in the process of generating a replica, the sequence IS (j is determined as a transmission symbol by using the tap coefficient h (j, k) of the impulse response at time j obtained in the transmission channel estimation process.
−k) Assuming that｝ (k = 0-L) has been transmitted, a replica χ (j) of the received signal is created using “Equation 3” (Equation (3) in FIG. 12). Replica is all possible IS
(J−k)｝ is calculated.

[0009]

(Equation 3)

Next, the Viterbi algorithm used in the transmission sequence estimation process will be described. First, the state of the communication channel at time j is represented by σ (j) − (S (j−L + 1),.
, S (j)). For example, assuming that L = 2, ± 1 = 2
When sending a transmission sequence that takes a value, the state of the communication path is σ0 =
(-1, -1), σ1 = (-1, + 1), σ2 = (+
(1, -1) and σ3 = (+ 1, +1), and the state σ (j-1) at time j-1 is σ (j) at time j corresponding to the transmission symbol S (j). Transitions to.

FIG. 14 shows this state. In the figure,
d [σ (j)] is the remaining path metric in state σ (j) at time j, and b [σ (j−1) → σ (j)] is state σ
This is a branch metric when transitioning from (j-1) to state σ (j). For example, d [σ (j) = σ0] is
d [σ (j-1) = σ0] + b [σ0 → σ0] and d [σ
(J−1) = σ0] + b [σ2 → σ0], which is smaller. The selected path is called a surviving path, and the unselected path is called a non-surviving path.

A transition path when the transmission symbol S (j) at time j is +1 is represented by a solid line, and a transition path when S (j) is -1 is represented by a broken line. For example, the state σ at time j-1
When (j−1) is σ1 = (− 1, + 1), that is,
When S (j−2) = − 1 and S (j−1) = + 1, S
When (j) is +1, the state σ (j) at time j is σ3 =
The state transits to (+ 1, + 1), and when S (j) is -1, the state σ (j) transits to σ2 = (+ 1, -1).

In this way, the four states at time j-1 transition to the state at time j in accordance with S (j) of ± 1, respectively, and there are a total of eight paths. Next, the remaining path metric is obtained as follows. First, at time j−1, the state σ (j
-1), the remaining path metric d [σ (j-1)] is determined for each of the four states.
The branch metric b [σ (j−1) −1σ (j)] is calculated and added using “Equation 4” (Equation (4.1) in FIG. 12) for all the transition paths that become (j).

[0014]

(Equation 4)

In this case, there are two transition paths for each σ (j). The smallest of the sums is defined as the remaining path metric d [σ (j)] at time j. This can be expressed by “Equation 5” ((Expression 4.2) in FIG. 12).
The selected path is set as a remaining path. This is repeated from the start state j = 0 to the end state j = Nd (Nd is the data length). Table 1 summarizes the meanings of the codes used in the above equations.

[0016]

(Equation 5)

[0017]

[Table 1]

FIG. 115 shows a trellis diagram starting from σ 0 and starting at time j = 7. The remaining paths to each state at time j = 7 are indicated by solid lines, and the non-remaining paths are indicated by broken lines. The state transition corresponding to the non-remaining path at time j = 6 is not displayed. In FIG. 15, the remaining path is uniquely determined up to j = 4. If the transmission path estimation process is successful, the paths that reach any state are combined into one, as far back as the past.

For this reason, if the length that is traced back in the past is represented by Nh as the path history length, the transmission symbol at time j-Nh can be determined. The remaining path selected at this time is a remaining path metric d [σ for each state at time j.
(J)], the remaining path corresponding to the state that is the smallest among them is selected as the maximum likelihood sequence, and the minimum remaining path metric is set as the minimum path metric dmin (j).

For example, in FIG. 15, Nh = 3 and d
Assuming that min (J = 7) = d [σ0], the remaining path indicated by the thick line is selected as the maximum likelihood sequence, and the transmission symbol at time j = 4 can be estimated to be +1.
(J = 4). Returning to FIGS. 11 and 12, the determination signal u (j) obtained in the transmission sequence estimation process is used in the transmission channel estimation process.

As described above, the transmission path estimation process, the replica generation process, and the transmission sequence estimation process are repeated for each symbol, and a determination result is output. The details of such a conventional method are described in detail in, for example, Literature: Minoru Okada, "Waveform Equalization Technique for Digital Mobile Communication-Section 2, Maximum Likelihood Sequence Estimation Filter" (Trikes Publishing, 1994). Are listed.

[0022]

As described above, in the conventional transmission channel estimation method of the adaptive equalizer, in the transmission channel estimation process, the transmission symbol estimated in the transmission sequence estimation process during the data section is determined by the decision signal. And continuously calculates the impulse response. For this reason, when there is a decision error in the transmission sequence estimation process, there is a problem that the impulse response cannot be obtained correctly due to the decision error in the transmission channel estimation process.

That is, this is because errors in transmission paths are erroneously estimated due to past determination errors, and errors in transmission symbol estimation occur due to erroneous transmission path estimations. Is caused by an error propagation event. An object of the present invention is to provide a method for estimating a transmission path of an equalizer that can improve the bit error rate characteristics by removing such error propagation.

[0024]

According to the present invention, the above-mentioned object is solved by the means described in the claims. That is, the invention of claim 1 provides a transmission path estimation method in an adaptive equalizer that sequentially estimates an impulse response of a transmission path from a received signal and removes waveform distortion due to fading to obtain a transmission symbol.

A training process for calculating an initial value of an impulse response of a transmission line based on a known training signal and a received signal, and after the training process, a data section of the received signal is divided into blocks of an appropriate length. And a transmission path estimation step of sequentially calculating an impulse response of the transmission path for each block,

The block quality calculation step detects the degree of a decision error in each block. If the result of the block quality calculation step indicates that there is no decision error in the block, the impulse response of the final state of the block is converted to the next block. The initial value of the impulse response of the block, and if there is a decision error, the impulse response of the start state of the block is used as the initial value of the impulse response of the next block. is there.

According to a second aspect of the present invention, in the transmission path estimating method for an equalizer according to the first aspect, when there is a determination error in the block quality calculation process, the impulse response of the starting state of the block is calculated as follows. In addition to the initial value of the impulse response of the block, the step size parameter of the transmission channel estimation process is temporarily increased to perform the transmission channel estimation process.

According to a third aspect of the present invention, in the method for estimating a transmission path of an equalizer according to the first aspect of the present invention, an impulse response in a training process is stored, and a block is determined from a result of a block quality calculation process. If there is no error, the impulse response of the end state of the block is used as the initial value of the impulse response of the next block. If there is a decision error, the impulse response of the training process is used as the impulse response of the next block. Is configured to perform a transmission path estimation process as an initial value of.

According to the present invention, when the transmission path is estimated in the adaptive equalizer, the data section of the received signal is divided into blocks of an appropriate length, and the quality of each block is calculated. From the result, if there is no decision error in the block, the impulse response of the end state of the block is used as the initial value of the impulse response of the next block. If there is a decision error, the impulse response of the block is initialized. A transmission path estimation method characterized by performing a transmission path estimation process as a value.

That is, as an initial value of the impulse response of each block, the channel estimation is performed using the result of the impulse response of the block having no decision error before that block. Therefore, it is possible to prevent the decision error generated in the previous block from spreading to the subsequent block,
The problem of error propagation can be solved.

[0031]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIGS. 1 and 2 show flowcharts (flowcharts) of a transmission path estimation method according to a first embodiment of the present invention. This example corresponds to the first aspect of the present invention. FIG. 3 shows a burst format used in the present invention. A major difference of the present invention from the conventional method is that the data section is divided into M blocks as shown in FIG. 3 and a block quality calculation step is provided after the transmission sequence estimation step as shown in FIG. Calculate the quality of

From the result, if the quality of the block is good, the impulse response in the end state of the block is used as the initial value of the impulse response of the next block. If the quality of the block is bad, the start of the block is started. The transmission path estimation process is performed using the impulse response of the state as the initial value of the impulse response of the next block.

Next, a specific example of the block quality calculation process will be described using an adaptive equalizer in which the channel estimation is applied to the maximum likelihood sequence estimation (MLSE) as in the conventional example. FIG. 4 shows a simulation performed on the adaptive equalizer.
5 shows a time change of a minimum path metric in a burst in which error propagation occurs.

At a certain time, the minimum path metric has rapidly increased from the previous value, and a determination error has occurred several symbols after this. This is because an erroneous path was selected in the transmission sequence estimation process due to the influence of noise or the like, which became larger than the value of the minimum path metric up to that point. It is only because it takes time. A specific example of the block quality calculation process utilizing this phenomenon is shown in FIGS. 5 and 6 as flowcharts.

The counter value ic is set to 0 at the start of the block.
And keep it. A certain threshold value De is provided, and when the minimum path metric dmin (j) is larger than De, ic is increased by +
Count up by one. Further, a certain reference value Ie for determining the quality of the block is provided. If the value of ic at the end of the block is equal to or less than the reference value Ie, the quality of the block is determined to be good. If it is larger, the quality of the block is degraded.

If Nb is the block length and the quality of the block n is good, the final result h (Nb, k) of the impulse response of the block n is set to the initial value h (0, k) of the impulse response of the next block n + 1. I do. However, when the quality of the block n is degraded, the initial value h (0, k) of the impulse response of the block n is set as the initial value h (0, k) of the impulse response of the next block n + 1. The longer the block length Nb, the higher the accuracy of the block quality determination, but cannot be made too long if the speed of change in the environment of the transmission path is high.

FIGS. 7 and 8 show a second embodiment of the present invention.
3 is a flowchart of a transmission path estimation method of the example of FIG. This example corresponds to the second aspect of the present invention. This example is the first described earlier.
The difference from the example of the embodiment is that the step size parameter used in the transmission channel estimation process is made variable. That is, μ0 and μ
1 are prepared and μ1 <μ0 is set, and μ1 is used in a normal transmission path estimation process.

As a result of the block quality calculation process, if the quality of the block is good, the step size parameter of the next block is kept at μ1, and if it is degraded, μ
Let 0 be the initial value of the step size parameter for the next block.

In the latter case, μ0 is returned to μ1 after a predetermined time Ns (step length, Ns <Nb) starting from μ0. As a result, even when the initial value of the impulse response of the block changes and discontinuity occurs in the impulse response in the transmission path estimation process, it is possible to speed up the entrainment by temporarily increasing the step size parameter.

FIGS. 9 and 10 show flowcharts of a transmission path estimation method according to a third embodiment of the present invention. The burst format used in this example is the same as that of the first example shown in FIG. This embodiment is different from the first embodiment in that the impulse response h due to the training process is different.
(0, k) is stored in the memory as h0 (k),
If the block quality is degraded as a result of the block quality calculation process, the point is that h0 (k) is set as the initial value of the impulse response of the next block.

That is, in the training process, an impulse response is calculated using a known signal, so that the impulse response is more accurate than in the data section. The method of this example is particularly effective for a burst having a short data length, as compared with the transmission path estimation method of the first example described above.

[0042]

As described above, according to the transmission channel estimation method of the first aspect of the present invention, error propagation which is a problem in the adaptive equalizer can be eliminated. Since it can be avoided, the bit error rate characteristics can be improved, and the performance of the adaptive equalizer can be improved. Further, by using the invention of claim 2, it is possible to absorb the discontinuity of the impulse response between the blocks. The invention of claim 3 is particularly effective for short bursts.

[Brief description of the drawings]

FIG. 1 is a flowchart (part 1) illustrating a first example of an embodiment of the present invention.

FIG. 2 is a flowchart (part 2) illustrating a first example of an embodiment of the present invention.

FIG. 3 is a diagram showing a burst format used in the present invention.

FIG. 4 is a diagram showing a change in minimum path metric over time.

FIG. 5 is a flowchart (part 1) showing a specific example of a block quality calculation process.

FIG. 6 is a flowchart (part 2) illustrating a specific example of a block quality calculation process;

FIG. 7 is a flowchart (1) showing a second example of the embodiment of the present invention.

FIG. 8 is a flowchart (2) showing a second example of the embodiment of the present invention.

FIG. 9 is a flowchart (part 1) illustrating a third example of the embodiment of the present invention.

FIG. 10 is a flowchart (2) showing a third example of the embodiment of the present invention.

FIG. 11 is a flowchart (part 1) showing a conventional transmission path estimation method.

FIG. 12 is a flowchart (part 2) illustrating a conventional transmission path estimation method.

FIG. 13 is a diagram showing a conventional burst format.

FIG. 14 is a diagram illustrating a method of calculating a residual path metric.

FIG. 15 is a trellis diagram when transmission channel estimation is performed smoothly.

Claims (3)

[Claims] 1. A transmission path estimation method in an adaptive equalizer for sequentially estimating an impulse response of a transmission path from a received signal and removing a waveform distortion due to fading to obtain a transmission symbol, comprising the steps of: A training process for calculating an initial value of an impulse response of the transmission path based on the received signal; and after the training process, a data section of the received signal is divided into blocks of an appropriate length, and each block is sequentially processed. A transmission channel estimation process for calculating the impulse response of the transmission channel, and a block quality calculation process for detecting the degree of the determination error of each block.From the result of the block quality calculation process, if the block has no determination error, The impulse response of the end state of the block is used as the initial value of the impulse response of the next block. Transmission path estimation method and performing a channel estimation process the impulse response of the initial state as the initial value of the impulse response of the next block. 2. When there is a determination error in the block quality calculation process, the impulse response in the starting state of the block is set as the initial value of the impulse response of the next block.
2. The transmission channel estimation process is performed by temporarily increasing a step size parameter in the transmission channel estimation process.
The transmission path estimation method described in the above. 3. The impulse response in the training process is stored, and if there is no decision error in the block from the result of the block quality calculation process, the impulse response of the final state of the block is replaced with the impulse response of the next block. 2. The transmission channel estimation process according to claim 1, wherein an impulse response by a training process is used as an initial value of an impulse response of a next block when an initial value of the response is used, and when there is a determination error, the transmission channel estimation process is performed. Road estimation method.