JPH07221804A - Maximum likelihood sequence estimate device - Google Patents

Abstract

PURPOSE:To provide a means that detects a time fluctuation or the like of propagation distortion such as an amplitude and estimates and selects a transmis sion line optimum to both of power consumption and performance based on the detected value thereby reducing the power consumption in the adaptive maximum, likelihood sequence estimate device used for digital mobile communi cation or the like. CONSTITUTION:A propagation distortion detection circuit 1 detects time fluctuation of propagation distortion and its magnitude, and a maximum likelihood sequence estimate device 2 selects an estimated value different from a state estimated by a state depending propagation line characteristic estimate circuit 3a or an estimated value estimated by a propagation line characteristic estimate circuit 3b based on an output of the circuit 1 as the estimate value of the propagation line characteristic in use. Thus, even when the propagation line characteristic is fluctuated at a high speed, the adaptive maximum likelihood sequence estimate in following to the fluctuation is formed.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an adaptive maximum likelihood sequence estimator, and more particularly to improving the performance of the adaptive maximum likelihood sequence estimator incorporated in a communication device such as a digital mobile radio communication terminal.

[0002]

2. Description of the Related Art When digital mobile communication is performed, electromagnetic waves cause reflection, diffraction, and scattering phenomena a plurality of times in a propagation path of electromagnetic waves, which is a communication medium, and so-called multiple waves are generated. Waveform distortion occurs in the waveform of the received wave, and the propagation characteristics are greatly deteriorated.
Further, the characteristics of the propagation path fluctuate very rapidly and irregularly as the mobile station changes (for example, moves at high speed).

Application of an adaptive equalizer has been proposed as one of the measures against the deterioration of the propagation characteristics.

As an adaptive equalizer for mobile communication, for example, an adaptive maximum likelihood sequence estimator has been proposed.

Now, in the TIA digital cellular system of Japan and North America, it is adopted as a modulation system,
In "π / 4 shift QPSK", encoded audio data is converted into 2-bit parallel data by an S / P (Serial to Parallel) converter, and then differential encoding is performed.

The 2-bit code is (0,0), (0,
Since the four states of 1), (1,0), and (1,1) are taken, each of them is associated with a phase change (± π / 4, ± 3π / 4). In addition, a code mapped on the phase plane is called a complex symbol.

That is, the transmission data includes (0,0),
Four states of (0,1), (1,0), and (1,1) are considered, and when the transmission data is observed for a predetermined period, the state transition is a combination of these four states.

The maximum-likelihood sequence estimator accumulates a received signal deteriorated through various propagation paths for a predetermined time, and out of all possible state transitions that a transmission code can take, the most likely transmitted state transition is calculated. A means for estimating a high code sequence from a received signal and decoding the transmitted data, for example, CPU, RO
It can be realized by M, RAM, various CMOS and the like.

In such a maximum likelihood sequence estimator, in order to perform efficient estimation, a predetermined process is performed by using a "Viterbi algorithm" which is a publicly known and publicly known technique.

The maximum likelihood sequence estimator using the Viterbi algorithm does not calculate the likelihood (a measure indicating the degree of similarity to the transmission sequence) for all possible paths (that is, state transitions). , At a certain time, the likelihood of a path leading to each state is calculated, and only the most probable (maximum likelihood) path leading to that state is considered and executed from the next time. As a result, it is not necessary to calculate the likelihood for all the routes, and the amount of calculation can be greatly reduced.

Considering the influence of the delayed wave up to one symbol time (the propagation path memory length L of the Viterbi algorithm is "1"), the processing using the Viterbi algorithm will be described with reference to the explanatory view of FIG. In addition, in order to facilitate understanding, the explanation will be divided into steps.

[Step 1] First, at time (i-1), the path metric of state (0, 0) (meaning the certainty when a transition is made according to the path shown in the figure to reach that state) S1 is performed. , A branch metric of a path transiting from the state (0,0) at the time (i-1) to the state (0,0) at the time i (a scale indicating the degree of similarity between the received signal and the path). ) Is added and this value is set to A1.

"Step 2" Next, in the path metric S2 of the state (0, 1) at the time (i-1), from the state (0, 1) at the time (i-1) to the time i.
The branch metric of the path transiting to the state (0, 0) is added, and this value is set to A2.

"Step 3" Next, in the path metric S3 of the state (1,0) at the time (i-1), from the state (1,0) at the time (i-1) to the time i.
The branch metric of the path transiting to the state (0, 0) is added, and this value is set to A3.

"Step 4" Next, in the path metric S4 of the state (1,1) at the time (i-1), from the state (1,1) at the time (i-1) to the time i.
The branch metric of the path transiting to the state (0, 0) is added, and this value is set to A4.

[Step 5] A1, A obtained in the above
2, A3 and A4 are compared, and the largest one is set as the path metric of the state (0, 0) at time i. A
Of the 1, A2, A3, and A4, only the path corresponding to the largest one is left, and after time (i + 1), only that path may be considered. This remaining path is referred to as a "survival path", and the surviving path is retained as survivor path information in the storage means.

"Step 6" The above steps 1 to 5 are similarly repeated for the remaining states (0,1) (1,0) (1,1) at time i.

[Step 7] In the case of using the Viterbi algorithm, if the propagation path memory length (which is the symbol length affected by the delay wave) is L, the surviving path is past by "5 × L" time. It is known to be aggregated into one path. Therefore, the maximum likelihood path at time i (of the surviving paths of the number of states at time i,
The path having the maximum path metric at time i) is traced back by time “5 × L” to determine and determine the transmission data.

Next, how to obtain the likelihood, which is a measure for determining the transmission data, when the adaptive maximum likelihood sequence estimator using the Viterbi algorithm is applied to the digital mobile communication will be described.

When the propagation path is assumed to be an "unknown system" in which the input is the transmission signal from the transmitter and the output is the reception signal at the receiver, the system is generally referred to as a transversal type digital filter. (The filter is, for example, a tapped delay line and a plurality of multipliers that multiply the tap output signals by weighting the tap output signals,
It is configured to have an adder that adds the output values of a plurality of multipliers).

To estimate the characteristics of the propagation path is, so to speak,
Estimating the tap coefficient of the transversal filter, and estimating the impulse response of the transversal filter. This estimation is based on LMS (Least Me
an Square: Least squares method) and other adaptive algorithms.

At this time, the sampled value rs of the received signal
(I) is the modulation vector X of the transmission signal as
It can be modeled by the product of (i) and the impulse response C (i) of the propagation path.

[0023]

[Equation 1]

[0024]

[Equation 2]

[0025]

[Equation 3]

Here, n s (i) is noise.

Hi (k) is the impulse response of the channel and a (i) is the complex symbol.

The adaptive maximum likelihood sequence estimator comprises a received signal,
The branch metric is calculated from the difference signal between the modeled estimated received signal (which can be represented by the product of the transmitted signal and the estimated channel impulse response).

Since the characteristics of the propagation path change every moment,
In parallel with the determination of transmission data, it is necessary to adaptively estimate the characteristics of the propagation path.

Based on the above matters, the operation of the adaptive maximum likelihood sequence estimator will be described with reference to the configuration diagram of a general adaptive maximum likelihood sequence estimator shown in FIG. The description of this operation will also be given according to steps for easy understanding.

First, assuming that a received signal is input, this time is designated as time i. Then, the surviving path leading to each state at time i is selected according to the following steps.

"Step 11" First, a transmission code sequence is generated.

That is, in the state estimator 6, the time (i-
From L), a code sequence corresponding to the state transition to time i is generated. The code sequence is a combination of the 2-bit signals described above.

"Step 12" Modulation signal generation is performed.

That is, in the signal generator 7, step 1
A modulated signal corresponding to the code generated in 1 is generated.

"Step 13" Estimated received signal generation is performed.

That is, the transversal filter 8 performs the convolution integration of the estimated value of the propagation path characteristic stored in advance and the modulation signal generated in step 12, the modulation signal of step 12 is transmitted, and the estimated characteristic is calculated. A signal when the signal is propagated through the propagation path and is received is generated as a replica (estimated received signal).

"Step 14" The subtractor 9 obtains the difference between the received signal and the estimated received signal obtained in step 13, and calculates the error signal.

"Step 15" The value obtained by squaring the error signal and adding a negative sign is the branch metric of the path set in step 11 (a measure showing the degree of similarity between the path and the received signal). And

[Step 16] The branch metric of the path obtained in step 15 is added to the already stored path metric in the state of the path start point.

"Step 17" At time i, the processing from step 11 to step 16 is executed for all the paths connected to each state.

[Step 18] The state estimator 6 compares the likelihoods of several states connected to each state at time i, sets the maximum one as the path metric in each state, and determines the path having the maximum likelihood as The surviving path is used.
It holds the path metric and information about where the path came from.

"Step 19" The processing from step 11 to step 18 is executed for each state at time i.

[Step 20] In the state estimator 6, among the surviving paths having several states, the path having the maximum likelihood at time i is the maximum likelihood path, and the maximum likelihood path is the path with one surviving path. The transmission data is judged and output by going back by a predetermined time (judgment delay time) sufficient to collect the data.

"Step 21" Propagation path estimation circuit 1
At 0, the propagation path characteristic is estimated by the adaptive algorithm using the modulated signal of the determined data and the error signal obtained from the data determined to be the reception signal past by the determination delay time.

In practice, when the LMS algorithm is used as the adaptive algorithm, the tap coefficient is updated according to the following equation.

[0047]

[Equation 4]

[0048]

[Equation 5]

Here, Xm (i) is the estimated signal modulation vector, Cem (i) is the estimated propagation path impulse vector,
em (i) is an estimation error signal, and α is an update coefficient.

The processing from step 11 to step 21 is performed each time a received signal is input.

Since the maximum likelihood sequence estimator uses the Viterbi algorithm, the determination of the determination value is delayed by the determination delay time unique to the Viterbi algorithm. Channel estimation circuit 1
In 0, since this delayed decision code sequence is used, the past propagation path is estimated by this decision delay. Therefore, it cannot follow when the propagation path characteristics fluctuate so fast that the delay of the judgment delay time cannot be ignored.

In order to remove the influence of this decision delay, a method of estimating the propagation path for each state is known. Details of this method are described in, for example, Japanese Patent Application Laid-Open No. 3-165632.

The feature of this method is that the branch metric is calculated by using the propagation path characteristic estimation value estimated in the path start point state, and the propagation path characteristic estimation value is estimated according to the surviving path connected to each state. It is to be. This method will be described.

First, assuming that a received signal is input, this time is designated as time i.

Then, according to the following steps, the time i
Select the surviving path, which leads to each state of.

"Step 31" Transmission code sequence generation is performed.

That is, in the state estimator 6, the time (i-
The code sequence corresponding to the state transition (path) from L) to time i is generated.

"Step 32" Modulation signal generation is performed.

That is, the signal generator 7 executes step 31.
A modulated signal corresponding to the code generated in 1 is generated.

"Step 33" The estimated received signal is generated.

That is, the transversal filter 8 performs convolution integration of the already stored estimation value of the channel characteristic estimated at the path start point and the modulation signal generated in step 32 (this becomes the estimated reception signal). ).

[Step 34] Next, in the subtractor 9,
An error signal is calculated by obtaining the difference between the received signal and the estimated received signal obtained at step 33.

[Step 35] Next, the value obtained by squaring the error signal and adding a negative sign is used as a branch metric of the path set in Step 31 (a scale indicating the degree of similarity between the path and the received signal). Yes).

"Step 36" The branch metric of the path obtained in step 35 is added to the already stored path metric in the state of the path start point.

"Step 37" The processing from step 31 to step 36 is executed for all the paths connected to each state at time i.

"Step 38" The state estimator 6 compares the likelihoods of the paths existing in several states connected to each state at time i, sets the maximum one as the path metric in each state, and has the maximum likelihood. Let the path be the surviving path. The path metric and information on where the path came from are stored.

"Step 39" Propagation path characteristic estimation circuit 1
When 0, the propagation path characteristic estimation value of each state at the present time is obtained by the adaptive algorithm according to the surviving path.

[Step 40] In the state estimator 6, the path having the maximum likelihood at the time i among the surviving paths having the number of states is set as the maximum likelihood path, and the maximum likelihood path is the path having one surviving path. For a predetermined time (judgment delay time) that is sufficient to aggregate the data, the transmission data is judged and output.

"Step 41" For each state at time i, the processing from step 31 to step 40 is executed.

Incidentally, the above steps 31 to 41
The processes up to are performed each time a received signal is input.

By the way, according to this method, propagation path estimation is performed for each state using a code sequence having no delay and corresponding to the selected survivor path.

That is, the maximum likelihood sequence estimator for estimating the propagation path for each state can remove the influence of the decision delay. Therefore, there is an advantage that the tracking characteristic is good even when the time variation of the propagation path is fast. However, compared with a general adaptive maximum likelihood sequence estimator, the channel estimation is performed by the number of states, which has a drawback that the amount of calculation increases and the power consumption increases.

[0073]

A conventional maximum likelihood sequence estimator for estimating a propagation path for each state has sufficient equalization when, for example, the moving speed of a transmitter is fast and the fluctuation of propagation distortion is fast and large. In order to obtain the characteristics, state-by-state estimation was always performed.

However, in reality, when the reception state is good and the propagation distortion is small, or when the fluctuation of the propagation path is small, the propagation path is not estimated for each state, and the propagation path is estimated using a judgment value with a delay. In some cases, the transmission data can be decoded without error even if the above is performed.

Even in such a case, using the method of estimating the propagation path for each state, which requires a large amount of calculation, is nothing but waste of power.

This is a major obstacle in achieving low power consumption. In addition, the call time and the standby time of the digital mobile communication terminal and the like are lengthened, which is a major obstacle in reducing the size and weight of the terminal by reducing the size of the power supply, for example, the battery.

Therefore, in the present invention, by solving the above problems and reducing the power consumption while maintaining the equalization performance, it is possible to extend the talk time and the standby time of a digital mobile communication terminal or the like. It is possible to reduce the size and weight of the terminal device by preventing the power source and reducing the size of the power supply.

[0078]

In order to achieve the above object, the following means can be considered.

At least one of the time variation of the propagation distortion including fading and the magnitude of the propagation distortion is detected, and when at least one is larger than a threshold value set in advance for each, a signal indicating that it is large is output. Based on the propagation distortion detecting means, the received signal that represents one kind of state of the modulated signal and can have a plurality of states, and the propagation path characteristic information that represents the characteristics of the propagation path through which the received signal is propagated, A survivor sequence is estimated, an error signal obtained from a maximum likelihood sequence estimator that uses one of the survivor sequences as the maximum likelihood sequence, a difference between the received signal and the survivor sequence for each state, and the survivor sequence for each state. The modulated signal generated correspondingly is used as an input signal, and the propagation state characteristic information corresponding to the survival sequence for each state is estimated and output. And road characteristics estimator, a delay means for outputting a signal by applying a predetermined delay time to the received signal,
A subtractor that outputs a difference between the output of the delay unit and the estimated received signal sequence estimated by the maximum likelihood sequence estimation circuit based on the signal transmitted in the past by the predetermined delay time, and the output of the subtractor And a modulated signal generated corresponding to the past signal series by the predetermined delay time as input signals, and a propagation path characteristic estimation circuit for estimating and outputting propagation path characteristic information corresponding to the past signal series by the predetermined delay time And a configuration provided with.

Further, there is provided a changeover switch for selectively switching between the output of the transmission path characteristic estimating means for each state and the output of the propagation path characteristic estimating means to be the input of the maximum likelihood sequence estimating means.

The changeover switch is configured to select the output of the transmission path characteristic estimating means for each state when the propagation distortion detecting means outputs the large signal.

[0082]

The propagation distortion detecting means monitors the time variation or magnitude of the propagation distortion for a certain period of time, and compares it with a preset threshold value to detect the propagation distortion amount.

As a result, when the propagation distortion is larger than the threshold value, the changeover switch is connected to the propagation path estimation means for each state, and the maximum likelihood sequence estimation means makes the propagation path characteristic estimation value different for each state. Estimate the maximum likelihood sequence using.

Then, each time a survivor path is selected in each state, the propagation path characteristics are estimated by the propagation path characteristic estimation means for each state.

On the other hand, when the propagation distortion is small, the changeover switch is connected to the propagation path characteristic estimating means. At this time, the maximum likelihood sequence estimation means estimates the maximum likelihood sequence using the same propagation path characteristic estimation value in all states, which is the output of the propagation path characteristic estimation means.

After the transmission signal is judged, the judgment transmission signal having a delay and the reception signal delayed by the same delay time as the judgment time are used to make the propagation path characteristic estimation value by the propagation path characteristic estimating means. To update.

[0087]

Embodiments of the present invention will now be described in detail with reference to the drawings.

FIG. 1 is a block diagram of an example of the adaptive maximum likelihood sequence estimator according to the present invention, and FIG. 3 is a flowchart showing the processing procedure of the present invention.

In FIG. 1, 100 is an input terminal. A received wave received by an antenna or the like is frequency-converted into a baseband band, sampled at a symbol period of a modulated wave, and input through an input terminal. The input signal is stored in, for example, the storage unit 1000 described later.

An output terminal 101 is, for example, (0,0), (0,1), (1,
Any one of four types of data of 0) and (1, 1) is output.

The maximum likelihood sequence estimation circuit 2 is provided between the input terminal 100 and the output terminal 101, receives the received signal as an input, and uses the Viterbi algorithm to determine the transmission data sequence most likely to be transmitted. It is a means for estimating and outputting past transmission data from the output terminal for the determination delay time required for the determination, and can be realized by, for example, a CPU, a ROM, a RAM, various CMOSs and the like.

The constituent elements are the maximum likelihood sequence estimation unit 1 of FIG.
As shown in FIG. 1, the state estimator 6, the signal generator 7, the transversal filter 8, the subtractor 9, and the storage means 10 are shown.
00 is configured.

Each component is, for example, a CPU, an R
It can be realized by OM, RAM, various CMOS and the like.

The state estimator 6 generates a code sequence corresponding to the state transition (path) of the Viterbi algorithm and outputs it to the signal generator 7. Further, the state estimator 6 calculates a branch metric from the signal input from the subtracter 9, estimates the maximum likelihood sequence using the Viterbi algorithm based on the calculation result, and calculates the maximum likelihood sequence in the past by the determination delay time. It is a means for determining the transmission data and outputting it to the output terminal 101.

The signal generator 7 receives the code sequence generated by the state estimator 6, generates a modulated signal when the input code sequence is assumed to be transmitted, and outputs the generated modulated signal to a transversal filter. 8, means for outputting to the transmission path characteristic estimation circuit 3a and the state-specific propagation path characteristic estimation circuit 3b.

The transversal filter 8 performs convolutional integration of the output signal from the signal generator 7 and the propagation path characteristic estimated value to generate an estimated received signal. In this embodiment,
The propagation channel characteristic estimation value is given by the output of the propagation channel characteristic estimating circuit 3b or the state-specific propagation channel characteristic estimating circuit 3a, and is switched to either 3b or 3a by the changeover switch 5. Specifically, it switches according to the output signal of the propagation distortion detection circuit 1. For example, in the propagation distortion detection circuit 1,
When the magnitude of the propagation distortion is equal to or larger than a predetermined threshold value, it may be configured to switch to the 3a side.

A detailed description of the configuration example of the transversal filter 8 is omitted because it is a known technique. For example, a delay line with taps and a plurality of tap coefficient multiplying tap weight signals are multiplied. It is configured to include a multiplier and an adder that adds output values of a plurality of multipliers.

The subtractor 9 is a means for calculating the difference between the received signal input via the input terminal 100 and the output signal of the transversal filter 8, and is, for example, various CMOS.
Etc. can be realized.

The storage unit 1000 is a unit for storing all data necessary for the processing according to the present invention, such as a sampled input signal, a state number described later, information about a sequence, etc., and is realized by, for example, a RAM or the like. It

The propagation distortion detection circuit 1 is a means for detecting the time variation of the propagation distortion or the magnitude of the propagation distortion. When the magnitude of the propagation distortion is equal to or greater than a predetermined threshold value, the changeover switch is selected. 5 is a means for outputting a signal for switching, and can be realized by, for example, a CPU, a ROM, a RAM, various CMOSs, and the like.

Further, the propagation path characteristic estimation circuit 3a for each state
Is an output signal of the maximum likelihood sequence estimation circuit 2, which is a modulated signal corresponding to the surviving path and an error signal determined by the received signal at the present time and the surviving path as input signals. State-specific propagation path characteristics,
Means for estimating using an adaptive algorithm such as LMS algorithm, for example, CPU, ROM, RAM,
It can be realized by various CMOSs.

The propagation path characteristic estimating circuit 3b uses the modulated signal corresponding to the transmission data determined by the maximum likelihood sequence estimating circuit 2 and the output signal of the subtractor, and further applies an adaptive algorithm such as an LMS algorithm. It is a means for estimating propagation path characteristics by using, for example, CPU, ROM, RA
It can be realized by M, various CMOSs and the like.

The delay circuit 4 is a means for delaying the input received signal by a predetermined time necessary for collecting the surviving paths into one path and outputting the delayed signal. For example, it can be realized by various CMOSs. .

The changeover switch 5 is the propagation distortion detection circuit 1
When the propagation distortion changes with time or the propagation distortion is large, the state-by-state propagation path characteristic estimation circuit 3a and the maximum likelihood sequence estimation circuit 2 are connected. On the contrary, when the propagation distortion is small, Is a means for switching the connection state so as to connect the maximum likelihood sequence estimation circuit 2 and the propagation path characteristic estimation circuit 3b, and is constituted by, for example, an analog switch and a means for applying a voltage to the switch.

The subtracter 12 outputs the output signal of the delay circuit 4 and
It is a means for inputting the output signal of the transversal filter 8 corresponding to the determined transmission data and obtaining the difference between them, which can be realized by various CMOSs, for example.

Next, referring to the flow chart of FIG.
The operation of each unit shown in FIG. 1 will be described.

In step 102, the time variation or magnitude of the propagation distortion is detected by the propagation distortion detecting circuit 1 and compared with a preset threshold value. The propagation distortion time variation or magnitude may be detected, for example, from the magnitude of the difference signal between the received signal and the known signal, or the magnitude of the difference signal between the received signal and the estimated transmission signal.

As a result, if the propagation distortion is large, the process branches to step 104 and the following process is executed. Of course, at least one of the time variation of the propagation distortion and the magnitude of the propagation distortion may be the detection target.

First, in step 104, the received signal is input from the input terminal 100 and stored in the storage means 1000, for example. In addition, this time is set to time t.

Next, in step 106, the storage means 10
The value of the state number n already stored in 00 is set to "0" and cleared.

Next, in step 108, "1" is added to the state number n, and the process for the state of state number 1 is first performed.

Hereinafter, for convenience, the state of the state number n will be referred to as "state n".

In step 110, in the state at the start point of the path connected to the "state 1" at time t, the propagation path characteristic estimated value already estimated by the propagation path characteristic estimation circuit 3a is used to make the "state 1 at time t". The survivor path leading to “” is selected in the maximum likelihood sequence estimation circuit 2.

Now, a specific processing example in step 110 will be described taking the case where the propagation path memory length L is "1" as an example.

At this time, the states are (0, 0),
There are four types of (0,1), (1,0) and (1,1),
In each order, "State 1", "State 2",
These are “state 3” and “state 4”.

The maximum likelihood sequence estimation circuit 2, for example, as shown in FIG. 2, has a state estimator 6, a signal generator 7, a transversal filter 8, a subtractor 9, and a storage means 1000. I shall.

First, the likelihood of a path transiting from "state 1" at time (t-1) to "state 1" at time t is calculated. It should be noted that, for ease of understanding, a series of processes will be described by dividing them into steps.

"Step 51" First, the state estimator 6
Then, the code sequence (0,
0) and the code sequence (0, 0) corresponding to "state 1" at time (t-1).

[Step 52] Next, the signal generator 7 generates a modulated signal corresponding to the code determined in step 51.

"Step 53" Next, the transversal filter 8 propagates each state in "state 1" at time (t-1), which is the state of the path start point already stored in the storage means 1000 or the like. Road characteristic estimation circuit 3a
Convolution integration of the propagation path characteristic estimation value estimated in step 5 and the modulation signal generated in step 52 is performed to generate an estimated received signal.

[Step 54] Next, the subtracter 9 obtains the difference between the received signal at time t and the estimated received signal generated in step 53 to calculate the error signal.

[Step 55] Next, the value obtained by squaring the error signal calculated in step 54 and adding a negative sign is
The branch metric of the path transited from the “state 1” at time (t−1) to the “state 1” at time t.

[Step 56] Next, the branch metric of the path obtained in step 55 is added to the path metric in the state of the path start point (currently, "state 1" at time (t-1)), and the time is calculated. The likelihood of the path that transits to "state 1" at time t from "state 1" of (t-1) is calculated.

Next, the likelihood of the path transiting from "state 2" at time (t-1) to "state 1" at time t is calculated.

[Step 510] First, the state estimator 6
Then, the code sequence (0,
0) and the code sequence (0, 1) corresponding to "state 2" at time (t-1) are generated.

[Step 520] Next, the signal generator 7
Then, a modulated signal corresponding to the code determined in step 510 is generated.

[Step 530] Next, in the transversal filter 8, the propagation path characteristics estimated in the “state 2” at time (t−1), which is the state of the path start point, already stored in the storage means 1000. Convolutional integration of the estimated value and the modulation signal generated in step 520 is performed to generate an estimated reception signal.

[Step 540] Next, in the subtractor 9,
The difference between the received signal at time t and the estimated received signal generated in step 530 is calculated to calculate the error signal.

"Step 550" Next, step 54
The value obtained by squaring the error signal calculated at 0 and adding a negative sign is used as the branch metric of the path that transits from "state 2" at time (t-1) to "state 1" at time t. .

[Step 560] The branch metric of the path obtained in step 550 is added to the path metric at the state of the path start point (currently, "state 2" at time (t-1)), and "at time t" The likelihood of the path when it transits to "state 1" is calculated.

Similarly, for the remaining states "state 3" and "state 4" at time (t-1), the same processing as in steps 51 to 56 described above is performed, and time (t- 1) The likelihood of the path when transiting from "state 3" to "state 1" at time t, and the case when transiting from "state 4" at time t-1 to "state 1" at time t Calculate the likelihood of the path.

The likelihoods of paths that transit to the "state 1" at time t, which exist in several states, and are found by the above processing are compared, and the maximum likelihood is determined to be "state 1" at time t.
And the path with the maximum likelihood is the surviving path at time t. Information about where the path metric and the surviving path came from is held in the storage unit 1000, for example.

Now, let us return to the description following the flowchart of FIG.

After selecting the surviving path for "state 1" at time t, at step 112, the "state 1" at time t, which is the output of the maximum likelihood sequence estimation circuit 2 in the propagation path characteristic estimation circuit 3a for each state. In accordance with the survivor path connected to, the estimated value of the propagation path characteristic in the state of the surviving path start point that has already been stored is updated using the adaptive algorithm, and the propagation path characteristic for the "state 1" at time t is estimated.

When the LMS algorithm is used as the adaptive algorithm, it corresponds to the current received signal and the error signal determined by the survivor path (the error signal corresponding to the survivor path calculated by the processing in step 540) and the survivor path. The modulated signal is used to update the estimation value of the propagation path characteristic in the state of the surviving path start point, and the propagation path characteristic estimation value in the state of the surviving path end point at time t is obtained.

In step 114, it is determined whether or not the processes from step 108 to step 112 have been completed for all the states at time t.

Here, in general, the number of states is the Mth power of L. Here, M is the number of states that the transmission code can have,
In this embodiment, it is "4". L is the channel memory length,
In this embodiment, it is "1".

When it is judged "No" in step 114, the process branches to step 108, "1" is added to the state number, and the process of "state 2" at time t is performed.

At step 110, the maximum likelihood sequence estimation circuit 2
In this case, this time, using the estimated value of the channel characteristic estimated by the channel characteristic estimation circuit 3a for each state at the state at the start point of the path connected to the "state 2" at the time t, the "state 2 at the time t" is used. Select the survival path that leads to.

Similar to the case of selecting the surviving path connected to the "state 1" at the time t, the likelihood of the path transiting from the "state 1" at the time (t-1) to the "state 2" at the time t , The likelihood of a path transiting from “state 2” at time (t−1) to “state 2” at time t, and “state 3” at time (t−1) to “state 2” at time t. The likelihood of the path that transits to and the likelihood of the path that transits from “state 4” at time (t−1) to “state 2” at time t are calculated and compared, and the maximum likelihood is calculated. , And the path metric of “state 2” at time t, and the path having the maximum likelihood is the surviving path at time t. The path metric and information about where the path came from are stored in the storage unit 1000, for example.

After selecting the surviving path for the "state 2" at time t, at step 112, the state-specific propagation path characteristic estimation circuit 3a outputs the "state at time t" output from the maximum likelihood sequence estimation circuit 2. The estimated value of the propagation path characteristic in the state of the survival path starting point, which is already stored in the storage unit 1000 or the like in accordance with the survival path connected to "2", is updated by the adaptive algorithm, and the propagation path characteristic for "state 2" at time t is updated. To estimate.

In step 114, it is determined whether the processing from step 108 to step 112 has been completed for all the states at time t.

Since the determination in step 114 is "No", the process branches to step 108 and the state number is set to 1
Is added.

At step 110, the maximum likelihood sequence estimation circuit 2
In this case, this time, using the estimated value of the channel characteristic estimated by the channel characteristic estimation circuit 3a for each state in the state at the start point of the path connected to the “state 3” at the time t, the state t at the time t Select a survival path that leads to.

At step 112, the output from the maximum likelihood sequence estimation circuit 2 in the propagation path characteristic estimation circuit 3a for each state,
According to the surviving path connected to the “state 3” at the time t, the estimated value of the propagation path characteristic at the state of the surviving path already stored in the storage unit 1000 or the like is updated by the adaptive algorithm, and the “state 3” at the time t.
Estimate the channel characteristics for.

At step 114, it is judged whether the processing from step 108 to step 112 has been completed for all the states at time t.

Since the determination in step 114 is "No", the process branches to step 108 and "1" is added to the state number.

At step 110, the maximum likelihood sequence estimation circuit 2
At this time, the estimated value of the propagation path characteristic estimated by the propagation path characteristic estimation circuit 3a for each state at the state at the start point of the path connected to the "state 4" at the time t is used for the time t "state 4". Select a survival path that leads to.

In step 112, the output from the maximum likelihood sequence estimation circuit 2 in the propagation path characteristic estimation circuit 3a for each state,
According to the surviving path connected to the “state 4” at the time t, the estimated value of the propagation path characteristic at the state of the surviving path already stored in the storage unit 1000 or the like is updated by the adaptive algorithm, and the “state 4” at the time t.
Estimate the channel characteristics for.

In step 114, it is determined whether or not the processing from step 108 to step 112 has been completed for all the states at time t. If the determination in step 114 is “Yes”, finally, step 11
6, the maximum likelihood sequence estimation circuit 2 sets the sequence having the largest path metric at the present time (time t) in the maximum likelihood sequence estimation circuit 2 as the maximum likelihood sequence, and sets the maximum likelihood sequence to 1 for the survival path.
The transmission data in the past is output to the output terminal 101 by tracing back in the past by the time until it is collected into one sequence (judgment delay time).

Next, the processing when the propagation distortion is small due to the output of the propagation distortion detecting circuit 1 will be described. That is, this is the process when "No" is determined in step 102.

In step 118, the received signal is input from the input terminal 100 and stored in, for example, the storage means 1000 (for example, the received signal may be digitally sampled and stored). Further, this time is defined as time t.

At step 120, the state number n stored in the storage means 1000 is set to "0" and cleared.

In step 122, "1" is added to the state number n, and first, preparations are made for processing the state number 1.

In step 124, the propagation path characteristic estimation circuit 3b uses the propagation path characteristic estimated value estimated at time (t-1) to find the survivor path connected to the "state 1" at time t as the maximum likelihood series. It is selected by the estimation circuit 2.

Here, the specific processing in step 124 will be described by taking the case where the propagation path memory length is "1" as an example, as in the case where the above-mentioned propagation distortion is large. For ease of understanding, the steps will be explained separately.

First, the likelihood of a path transiting from "state 1" at time (t-1) to "state 1" at time t is calculated.

[Step 61] First, the state estimator 6
Is a code sequence (0,
0) and the code sequence (0, 0) corresponding to “state 1” at time (t−1).

[Step 62] Next, the signal generator 7
Then, a modulated signal corresponding to the code sequence determined in step 61 is generated.

[Step 63] Next, the transversal filter 8 is estimated by the channel characteristic estimation circuit 3b at time (t-1) and the channel characteristic estimated value already stored in the storage unit 1000 is used. , The convolution integration of the modulation signal generated in step 62 is performed to generate an estimated reception signal.

[Step 64] Next, the subtractor 9 obtains the difference between the received signal at time t and the estimated received signal generated in step 63 to calculate the error signal.

[Step 65] Next, the value obtained by squaring the error signal calculated in step 64 and adding a negative sign is
The branch metric of the path transited from the “state 1” at time (t−1) to the “state 1” at time t.

[Step 66] The branch metric of the path obtained in step 65 is added to the path metric in the state of the path start point (currently, "state 1" at time (t-1)), and the time (t- The likelihood of the path that transits to "state 1" at time t from "state 1" in 1) is calculated.

In the same manner, the remaining states "state 2", "state 3" at time (t-1),
Also for “state 4”, the processing from step 61 to step 66 is performed, and the likelihood of the path at the time of transition from “state 2” at time (t−1) to “state 1” at time t, and the time The likelihood of the path when transitioning from "state 3" of (t-1) to "state 1" of time t,
The likelihood of the path at the time of transition from "state 4" at time (t-1) to "state 1" at time t is calculated.

The likelihoods of the paths that are present in the number of states transiting to the “state 1” at the time t found are compared, and the maximum likelihood is set as the path metric of the “state 1” at the time t. The path having the likelihood of is the surviving path at time t. Information on the path metric and where the surviving path came from is, for example, the storage unit 1000.
Keep it in.

In step 126, it is determined whether or not the processing of steps 122 and 124 has been completed for all states at time t.

If the determination in step 126 is "No", the process branches to step 122, "1" is added to the state number, and then the process for "state 2" at time t is performed.

In step 124, the propagation path characteristic estimation circuit 3b uses the propagation path characteristic estimated value estimated at time (t-1) to find the survivor path connected to the "state 2" at time t as the maximum likelihood series. It is selected in the estimation circuit 2.

Similar to the above description, in step 122,
"1" is added to the state number, and in step 124,
Using the propagation path characteristic estimation value estimated at the time (t-1) by the propagation path characteristic estimating circuit 3b, the survivor paths connected to the "state 3" and the "state 4" at the remaining time t are converted into the maximum likelihood series. Each is selected by the estimation circuit 2.

Continuing with the above processing, step 126
Then, for all states at time t, step 122
Then, it is determined whether or not the process of step 124 is completed.

If the determination result is "Yes", then in step 128, the maximum likelihood sequence estimation circuit 2 sets the sequence having the maximum path metric at the present time (time t) among the surviving paths as the maximum likelihood sequence. It traces back in the past by the time (judgment delay time) until the maximum likelihood series is aggregated into one series of surviving paths, and outputs the transmission data in the past by the judgment delay to the output terminal 101.

In step 130, the propagation path characteristic estimation circuit 3b gives the propagation path characteristic estimated value estimated by the propagation path characteristic estimation circuit 3b one time before (time (t-1)), in step 12
8 is updated by an adaptive algorithm such as LMS using the judgment value output in 8 (for example, output data when the transmission data output to the output terminal 101 is input to the signal generator 7 or the transversal filter 8). Then, the propagation path characteristic is estimated.

The operation of each circuit will be described below. First, the delay circuit 4 inputs a received signal, delays it by a predetermined delay time necessary for determination, and outputs it. The subtractor 12 obtains the difference between the output of the delay circuit 4 and the estimated reception signal corresponding to the determination value of the transmission data that is the output of the maximum likelihood sequence estimation circuit 2, and determines the reception signal and the transmission data that are past by the determination delay time. Judgment value (for example, the transmission data output to the output terminal 101 is transmitted to the signal generator 7 or the transversal filter 8).
An error signal determined from the output signal when the input signal is input to) is calculated.

In the propagation path characteristic estimating circuit 3b, the subtracter 12
Output signal and the modulation signal corresponding to the value determined as the transmission data from the maximum likelihood sequence estimation circuit 2 as an input signal, and the estimated value of the channel characteristic stored in the storage unit 1000 is, for example, LMS. Etc. by an adaptive algorithm.

According to the embodiments of the present invention described above,
It is sufficient to estimate the propagation path characteristics for each state only when necessary, and when the propagation distortion is small, it is not necessary to perform the calculation for estimating the propagation path characteristics for each state, and the calculation amount is reduced. It becomes possible to achieve low power consumption.

FIG. 4 shows another embodiment according to the present invention.

FIG. 5 shows a TD which is a digital communication transmission system.
MA (Time Division Multiplex Access)
It is explanatory drawing of a system. As an example, 3 channel TDMA
The case of is described.

In the TDMA mobile communication, the base station directs the TDM (Time D) as shown in FIG.
ivision Multiplex) signal is transmitted.

The mobile station receives only its own time slot of the TDM signal. TD on one radio channel
As many mobile stations as MA multiplex can simultaneously communicate with the base station.
The data sequence of each channel has a slot configuration, and each slot has a training sequence composed of known signals. The training sequence is, for example, a signal transmission time provided for setting the tap coefficient of the equalizer.

In the present invention, for example, the time variation, the magnitude, etc. of the propagation distortion in the propagation distortion detection circuit 1 are detected in the training section, and the changeover switch 5 is switched according to the detected propagation distortion. Should be done. The switch switching processing is not performed in the slot section including the training section. However, each time the slot changes, the propagation distortion is newly detected so that the switching process can be performed.

Data is actually transmitted and received in the tracking section following the training section.

The operation will be described below with reference to the flowchart of FIG.

First, in step 202, the time variation, the magnitude, etc. of the propagation distortion are detected in the training section, and comparison processing with a predetermined threshold value is performed.

Specifically, first, at the beginning of the training section, for example, the propagation path characteristic estimation value stored in the storage unit 1000 is reset.

Processing for converging the estimated value of the propagation path characteristic in the training section (setting the tap coefficient)
In parallel with the above, the time variation, the magnitude, etc. of the propagation distortion are detected from the known signal and the received signal in the training section. It goes without saying that such detection may be either time variation or magnitude of propagation distortion. The same applies to all the embodiments of the present invention. When the training section is completed, the changeover switch 5 is switched according to the magnitude of the propagation distortion detected in the training section.

When it is determined in step 202 that the propagation distortion is larger than a predetermined threshold value, the changeover switch 5 is connected to the state-specific propagation path estimation circuit 3a.

Since the following processing is similar to that of the above-mentioned embodiment, it will be briefly described.

First, at step 204, the input terminal 100
To receive the received signal. The received signal may be sampled at predetermined time intervals and stored in the storage unit 1000.

Next, at step 206, the state number n stored in the storage means 1000, for example, is cleared to "0".

Next, at step 208, "1" is added to the state number n.

Next, at step 210, the maximum likelihood sequence estimation circuit 2 selects the surviving path connected to "state n" using the propagation path characteristic estimation value in the state at the start point of the path connected to "state n". To do.

Next, in step 212, the propagation path characteristic estimation circuit 3a for each state updates the propagation path characteristic estimated value of the starting point of the surviving path connected to "state n" by an adaptive algorithm such as LMS, and the current value is calculated. The propagation path characteristic of "state n" is estimated.

Then, the processing from step 208 to step 212 is repeated for all the current states.

Further, when it is determined in step 214 that the processing from step 208 to step 212 has been completed for all the current states, the maximum likelihood sequence estimation circuit 2 determines in step 216 that the survivor path has survived. , The series having the largest path metric at the present time is set as the maximum likelihood path, the maximum likelihood path is traced back by the judgment delay time in the past, and the transmission data in the past by the judgment delay is output terminal 101.
Output to.

The processing from step 204 to step 216 is continued until it is judged at step 218 that the processing of the last data in the slot is completed.

At this time, the changeover process of the changeover switch 5 is not performed during the slot including the training section.

If it is determined in step 218 that the processing of the last data in the slot has been completed, step 2
Branch to 02.

Next, the processing when it is determined that the propagation distortion is small will be described. Since this processing is also the same as the processing in the above-described embodiment, detailed description will be omitted.

First, when it is determined in step 202 that the propagation distortion is small, the changeover switch 5 is connected to the propagation path characteristic estimating circuit. Hereinafter, the same processing as that of the first embodiment will be performed.

Next, at step 220, the input terminal 100
To receive the received signal. The received signal may be sampled at predetermined time intervals and stored in the storage unit 1000.

Next, at step 222, for example, the state number n stored in the storage means 1000 is set to "0".
To clear.

Next, at step 224, the state number n
"1" is added to.

Next, at step 226, the maximum likelihood sequence estimation circuit 2 uses the propagation path characteristic estimated value estimated by the propagation path characteristic circuit for each state to select the surviving path leading to "state n".

Then, the processing from step 220 to step 226 is repeated for all the current states.

Now, in step 228, steps 220 to 226 are performed for all the current states.
If it is determined that the processing up to is completed, step 230
Then, in the maximum likelihood sequence estimation circuit 2, the sequence having the largest path metric at the present time is set as the maximum likelihood path among the surviving paths, the maximum likelihood path is traced back by the judgment delay time, and the transmission data in the past is judged by the judgment delay. , Output terminal 101
Output to.

Next, at step 232, the propagation path characteristic estimation circuit 3b inputs the transmission data to the signal generator 7 and the transversal filter 8 and uses the judgment value of the output signal obtained to obtain the propagation path. Estimate the characteristics.

Next, in step 234, the processing from step 220 to step 232 is continued until it is judged that the processing of the last data in the slot is completed.

At this time, the changeover process of the changeover switch 5 is not performed during the slot including the training section.

After the processing of the last data in the slot is completed, the process branches to step 202.

Similarly, the processing for the data of the next slot for the own station is continued.

In this embodiment, the time variation and the magnitude of the propagation distortion are detected from the known training signal and the received signal in the training section, and when the training section ends, the propagation distortion detected in the training section is detected. The switching process of the changeover switch 5 is performed according to the size, but the maximum changeover process of the changeover switch 5 is not performed in the slot section including the training section, and the maximum likelihood sequence is estimated.

According to the second embodiment of the present invention described above, the calculation amount can be reduced in the slot with small propagation distortion,
That is, only for slots that require a predetermined operation,
It suffices to perform a predetermined calculation, unnecessary calculation processing is omitted, and power consumption can be reduced.

Further, it is possible to simplify the algorithm relating to the detection of the propagation distortion and the switching of the propagation path characteristic estimation circuit within the slot.

The maximum likelihood sequence estimator according to the present invention is
For example, it goes without saying that the performance of the wireless communication device is improved by mounting it on a wireless communication device such as a car phone, and it can be applied to various wireless communication devices not limited to the car phone.

Further, the maximum likelihood sequence estimator itself is
Since it can be composed of OM, RAM, various CMOSs, etc., and can be made into a custom IC if necessary, it can be realized in a small size and at low cost, and it is very easy to mount it on various wireless communication devices.

[0216]

According to the present invention, a plurality of transmission line characteristic estimation methods (that is, By selecting from the transmission path characteristic estimation method for each state or the transmission path characteristic estimation method that is not for each state, an adaptive maximum likelihood sequence estimator with good minimum tracking power consumption and good followability to fast fading can be obtained. Can be provided.

As a result, the call time and standby time of the digital mobile communication terminal or the like can be reduced, the power source such as the battery can be downsized, and the terminal device can be downsized and lightened.

[Brief description of drawings]

FIG. 1 is a configuration diagram of an adaptive maximum likelihood sequence estimator that is an embodiment according to the present invention.

FIG. 2 is a configuration diagram of an adaptive maximum likelihood sequence estimator.

FIG. 3 is a flowchart showing a processing procedure in an embodiment according to the present invention.

FIG. 4 is a flowchart showing a processing procedure in another embodiment according to the present invention.

FIG. 5 is an explanatory diagram of a transmission format.

FIG. 6 is an explanatory diagram of a Viterbi algorithm.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Propagation distortion detection circuit, 2 ... Maximum likelihood sequence estimation circuit, 3a ... State-specific transmission line characteristic estimation circuit, 3b ... Transmission line characteristic estimation circuit, 4 ... Delay circuit, 5 ... Changeover switch, 6 ... State estimator, 7 ... Signal generator, 8 ... Transversal filter, 9 ... Subtractor, 10 ... Transmission path characteristic estimation circuit, 100 ...
Input terminal, 101 ... Output terminal

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Office reference number FI Technical display location H04B 7/005 4229-5K H04L 25/03 Z 9199-5K 25/08 B 9199-5K

Claims (3)

[Claims] 1. A signal indicating that at least one of a time variation of propagation distortion including fading and a magnitude of propagation distortion is detected, and when at least one of the fluctuations is larger than a threshold value set in advance, the signal indicates a large signal. Based on the propagation distortion detecting means for outputting the received signal, which represents one kind of state of the modulated signal, can have a plurality of states, and propagation path characteristic information which represents characteristics of a propagation path through which the received signal is propagated. , A maximum likelihood sequence estimating means for estimating a survival sequence and using one of the survival sequences as a maximum likelihood sequence, an error signal obtained from the difference between the received signal and the survival sequence for each state, and for each state. With the modulation signal generated corresponding to the survival series as the input signal,
Propagation characteristic estimation means for each state for estimating and outputting propagation path characteristic information corresponding to the survival sequence for each state, delay means for giving a predetermined delay time to the received signal and outputting the signal, and output of the delay means A subtractor that outputs a difference from an estimated received signal sequence estimated by the maximum likelihood sequence estimation means based on a signal transmitted in the past by the predetermined delay time, an output of the subtractor, and the predetermined delay time Only the modulated signal generated corresponding to the past signal series is used as the input signal,
A propagation path characteristic estimation circuit for estimating and outputting propagation path characteristic information corresponding to a past signal sequence by the predetermined delay time, and an output of the state-specific transmission path characteristic estimating means and an output of the propagation path characteristic estimating means. And a changeover switch which is an input of the maximum likelihood sequence estimation means, wherein the changeover switch transmits each state when the propagation distortion detection means outputs the signal indicating the largeness. A maximum likelihood sequence estimator characterized by selecting an output of a road characteristic estimating means. 2. The propagation distortion detecting means according to claim 1, wherein propagation including fading occurs in a transmission section of the known signal in a slot having a transmission section of a known signal including a synchronization signal and a transmission section for transmitting data. Detects at least one of the time variation of distortion and the magnitude of propagation distortion,
A maximum likelihood sequence estimator which outputs a signal indicating a large value when at least one of them is larger than a predetermined threshold value. 3. A switching prevention means according to claim 2, further comprising a switching prevention means, wherein when the switching operation of the changeover switch is performed once in the slot, the switching operation is prevented again. Maximum likelihood sequence estimator.