JP2560911B2 - Sequence estimation device - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a sequence estimation device that estimates a transmission signal sequence by following a temporal change in characteristics of a transmission path.

(Prior Art) A maximum likelihood sequence estimator (MLSE) is known as an equalization method with the best equalization capability (for example, Reference 1: GDForne.
y, “Maximum Likelihood Sequence Estimation of Digi
tal Sequences in the presence of intersymbol inter
eference, "IEEE Transaction on Information Theory, v
ol.IT-18, no.3, May 1972). The maximum likelihood sequence estimator generally comprises a single channel response estimator, and the channel response is estimated by using this when receiving a known channel.

Further, when the characteristics of the transmission line fluctuate with time, an adaptive maximum likelihood sequence estimator that follows the temporal fluctuation of the characteristics of the transmission line has also been proposed (for example, Reference 2: G.
Ungerboeck, “Adaptive Maximum Likelihood Receiver
for Carrier-Modulated Data Transmission Systems, "
IEEE Transaction on Communications, vol.COM-22, no.
5, May 1974). The adaptive maximum likelihood sequence estimator first obtains a channel response when receiving a known sequence, and subsequently operates an channel estimator using an adaptive algorithm when receiving an information data sequence to sequentially transmit the channel response. It is characterized in that it updates to follow the temporal fluctuations of the transmission line characteristics. However, when the channel characteristics change rapidly, the adaptive operation of the channel estimator cannot catch up, and it is difficult for the adaptive maximum likelihood sequence estimator to perform good sequence estimation.

On the other hand, Japanese Patent Application No. 2-203436 proposes a new type sequence estimation device that can follow a transmission line that changes at high speed. This device is characterized in that the Viterbi algorithm is applied by estimating the corresponding transmission path response for each series, assuming that not only the transmission signal series but also the characteristics of the transmission path are unknown. The transmission path response is estimated by solving the transmission path equation determined by the transmission signal series candidate, the transmission path response, and the received signal for each series. This is equivalent to sequentially finding the optimum solution of the channel response, so
The sequence estimation device of Japanese Patent Application No. 2-203436 can follow high-speed transmission line fluctuations.

(Problems to be Solved by the Invention) However, in the sequence estimation device of Japanese Patent Application No. 2-203436, a transmission line equation determined by three of the transmission signal sequence candidate, the transmission line response, and the reception signal depending on the pattern of the transmission signal sequence candidate. However, there is a drawback that there is a case where the transmission line response becomes indefinite because it cannot be solved by nature. Further, since the sequence estimation device of Japanese Patent Application No. 2-203436 successively obtains the solution, it has a drawback that the channel estimation value fluctuates excessively sensitively due to noise or the like, resulting in erroneous sequence estimation.

Therefore, an object of the present invention is to provide a sequence estimation device capable of following a transmission line that changes at high speed while stably performing transmission line estimation for each sequence.

(Means for Solving the Problem) A first sequence estimation apparatus according to the present invention includes a register that stores a plurality of sample values of a received signal, and a plurality of sample values that are input from the register to generate a plurality of sequences. On the other hand, the transmission line response calculation circuit for estimating the transmission line response at the current time by the direct solution method, and the validity of the transmission line response for the plurality of sequences obtained by the transmission line response calculation circuit are inspected,
If it is valid, the output of the transmission path response calculation circuit is designated as the current time transmission path response, and if it is not valid, the survival sequence connection information is instructed from the plurality of transmission path responses at the previous time as the current time transmission path response. A transmission line characteristic inspection circuit for outputting a transmission line response at a previous time; a transmission line characteristic storage circuit for storing the transmission line response output by the transmission line characteristic inspection circuit for a plurality of times respectively for a plurality of streams; A branch metric calculation circuit for obtaining a virtual reception signal point for each series based on the transmission path response at the current time stored by the path response storage circuit, and for obtaining a distance from the sample value of the reception signal; Viterbi that receives the output of the calculation circuit and determines the received signal by the Viterbi algorithm and outputs the connection information of the survival sequence to the transmission line response inspection circuit Wherein the processor, in that it is composed of.

A second sequence estimation device according to the present invention is a register for storing a plurality of sample values of a received signal, and a plurality of the sample values are input from the register to transmit a current time to each of a plurality of sequences. A channel response calculation circuit for estimating a channel response by a direct solution method, and the legitimacy of the channel response for a plurality of sequences obtained by the channel response calculation circuit is inspected. The output of the transmission line response calculation circuit is output as a transmission line response of the previous time indicated by the connection information of the surviving sequence from a plurality of transmission line responses of the previous time as the transmission line response of the current time if not valid. Supplied from a characteristic inspection circuit, a transmission path characteristic storage circuit that stores the transmission path response output from the transmission path characteristic inspection circuit for a plurality of times for a plurality of times, and the transmission path characteristic storage circuit And a transmission line response conversion value given by the transmission line response conversion circuit, which obtains a transmission line response conversion value according to a predetermined conversion rule from the transmission line responses for the present and past plural times for each sequence A branch metric calculation circuit that obtains a virtual reception signal point for each sequence based on the above and obtains a distance from the sample value of the reception signal, and a reception signal is determined by a Viterbi algorithm by receiving the output of the branch metric calculation circuit. And a Viterbi processor that outputs the connection information of the surviving sequence to the transmission line response inspection circuit.

(Operation) In the following, consider a transmission line in which responses of a plurality of delayed waves exist with respect to the main wave as shown in FIG. The transmission line impulse response is a vector _t ^T = (h _t ⁰ , h _t ¹ , ..., H _t ^L ) (the superscript T means transposition), and the transmission signal sequence is a vector _t ^T.
= (S _t , s _t-1 , ..., S _tL ), and let v _t be additive noise that is independent of the transmitted signal and includes noise in the observation process. In this case, the receiver input r _t at time t, given by the sum of the convolution and noise between the vector h _t and the vector _t as shown in equation (1).

r _t = _t ^T · _t + v _t (1) This state is shown in FIG. Hereinafter, Expression (1) will be referred to as a transmission line equation at time t.

Next, a method of least-squares estimating the transmission path impulse response vector _t from N received signals from time t−N + 1 (where the transmission signal time interval is 1) to time t will be described. First, for that purpose, from time t-N + 1 to time t, N
Transmission signal sequence vectors s _τ (t−N + 1 ≦ τ ≦ t)
Then, the transmission signal matrix S _t is defined as follows.

Further, the received signal vector _{t 1} and the noise vector _{t 1} are defined below. _t ^T = (r _t , r _t-1 ,, ..., r _{t-N + 1} ) (3) _t ^T = (v _t , v _t-1 ,, ..., v _{t-N + 1} ) (4) From the above , The transmission line equation over N time points can be written by the equation (5). _t = S _t ^T · _t + _t (5) At this time, the transmission line impulse response vector _{t, ls} by the least-squares estimation is _{t, ls} = (S _t ^T · S _t ) ⁻¹ · S _t ^T · _t ( Obtained in 6). (For example, Reference 3: Nakamizo "Signal Analysis and System Identification", Corona, 1988). In particular, the number of received signals (N) used for impulse response estimation is the number of transmission path responses (L +
When it is equal to 1), the transmission signal matrix S _t becomes a square matrix, and therefore the transmission signal response estimation value by least square estimation is obtained by simply multiplying the reception signal by the inverse matrix of the transmission signal matrix S _t . _{t, ls} = S _t ⁻¹ · _t (7) The sequence estimation device of Japanese Patent Application No. 2-203436 shown in FIG. 3 has all transmission signal matrices S _t , that is, all possible combinations of transmission signals (S _t , S _t-1 , ..., S _{tL-N + 1} ) _, the solution of the channel response estimation value vector _{t, ls} is obtained. Then, each transmission line response estimation value is used in the likelihood (branch metric) calculation obtained by Expression (8) for the transition from the current time to the next time.

| r _{t + 1} − _{t + 1} ^T · _{t, ls} | ² (8) where _{t + 1} ^T = (S _{t + 1} , S _t , ..., S _{t-L + 1} ) and all of this value The Viterbi algorithm is used to find the transmission signal sequence over all times that minimizes the value (path metric) determined by the sum over time. Here, the state of the trellis diagram for operating the Viterbi algorithm is the combination of the transmission signals appearing in the components of the transmission signal matrix S _t (s _t , s _t−1 , ..., s
_{tL-N + 1} ). In this sense, the combination of transmission signals (s _t , s _t-1 , s _{tL-N + 1} ) will be simply referred to as a state hereinafter. The sequence of the maximum likelihood states obtained by the Viterbi algorithm over the entire time is the transmission signal sequence estimated value.

By the way, in the sequence estimation device of this Japanese Patent Application No. 2-203436, the matrix (S _t ^T S _t ) ^-1 and the matrix S are obtained when the transmission path response vectors _{t and ls} are obtained by the formula (6) or (7), respectively. _You are calculating _t ^-1 . Therefore, for the matrix S _t ^T S _t or the state where the matrix S _t becomes singular, the transmission line response cannot be obtained as it is. There is a disadvantage that the branch metric calculation in which the transmission path response is indefinite cannot be performed and the Viterbi algorithm cannot be operated. In contrast, generally channel variations is the property that the smaller negligible in the time interval of the signal appearing on the components of the transmission signal matrix _{S t (L + N-1} ). Using this property, the matrix S _t ^T S _t
Alternatively, the channel response estimate vector _{t, ls} for the state in which the matrix S _t is singular is substituted by the channel response estimate vector vector _{t−1, ls} in the previous time state of the surviving sequence transitioning to the state. You can That is, in the first sequence estimation device according to the present invention, the channel response estimation value regarding the state in which the surviving sequence was taken at the previous time is used for the state in which the matrix S _t ^T S _t or the matrix S _t is singular. Then, for a state in which the matrix S _t ^T S _t or the matrix S _t is nonsingular, it is set to use the channel response obtained by multiplying the received signal by the inverse matrix of each matrix. As a result, the channel response estimation value can always be stably supplied to the branch metric calculation circuit, and if the conventional Viterbi algorithm is operated based on this, it can stably follow the channel that fluctuates at high speed. It is possible to realize a sequence estimation device that can do this.

Further, the sequence estimation device of Japanese Patent Application No. 2-203436 only sequentially obtains the channel response estimated value at each time, and does not use the correlation between the estimated values at each time. Therefore, the channel response estimation value for each state may change significantly at each time due to noise, and the sequence estimation may be erroneous. However, if noise is ignored, the transmission line response originally has the property of continuously changing according to the Doppler frequency. That is, although it is time-varying, the change in the transmission path response is generally gentler than the symbol transmission interval, and there is a high correlation between the transmission path responses at each time. Therefore, the second sequence estimation apparatus according to the present invention newly introduces a conversion operation that utilizes a high correlation between channel responses. That is, for each state, a new estimated value at the current time is obtained by this conversion operation from the estimated value of the transmission path response at the past time and the estimated value at the current time obtained by solving the transmission path equation. This allows
It is possible to prevent the phenomenon that the estimated value of the channel response greatly changes at each time, and to obtain the estimated value of the channel response that is stable and reliable in the true state. Further, by smoothing the channel response estimation value, noise in the channel estimation process can be effectively removed. as a result,
A sequence estimation device capable of stably following a transmission line that changes at high speed can be realized.

(Example) Next, this invention is demonstrated with reference to drawings.

FIG. 1 shows an embodiment of the first sequence estimation apparatus according to the present invention. Receiver input r _t in the supplied time t to an input terminal 101 is sent to the branch metric calculating circuit 106 while being stored in the register 102. From time t−N + 1 to time t, the N received signals stored in the register 102 are input to the transmission path response calculation circuit 103. The transmission line response calculation circuit 103 calculates the vector _{t, ls} for each state according to the equations (6) to (7), and the transmission line response inspection circuit
Output to 104. In the channel response calculation in the equations (6) to (7), since the matrix (S _t ^T S _t ) ^-1 and the matrix S _t ^-1 are both determined only by the transmission signal, all states, that is, (s _t , s _t-1, ..., previously calculated for the combination of s _{tL-N + 1)} all transmission signals may take the may be stored the results. For the matrix S _t ^T S _t or the combination of transmitted signals with which the matrix S _t is singular, the transmission line response calculation circuit
103 outputs a predetermined value (for example, 0) and informs the transmission path response inspection circuit 104 that the vectors _{t and ls} are indefinite. Here, the transmission line response estimation value obtained when the state is not indefinite is called a valid estimation value. The transmission line response inspection circuit 104 inspects whether or not the vectors _{t and ls} are indefinite for all states, and when they are not indefinite, the values of the vectors _{t and ls} are directly stored in the transmission line response storage circuit 105. Output. If it is indefinite, first, the state taken by the survivor path connecting to that state at the previous time is checked from the surviving sequence connection information supplied from the Viterbi processor 107. Next, the transmission line response estimation value vector _{t−1, ls} at the previous time corresponding to the state taken at the previous time is read from the transmission line response storage circuit 105, and the value is set as the transmission line response estimation vector _{t, ls} at the current time. , To the transmission path storage circuit 105 again. Therefore, the transmission path storage circuit 105 always stores valid transmission path response estimated values for all states.
The transmission path storage circuit 105 outputs these transmission path response estimated values to the branch metric calculation circuit 106. The branch metric calculation circuit 106 individually calculates the branch metric determined by the equation (8) for all possible states based on the transmission path response estimated value. The branch metric calculation circuit 106 outputs the branch metric obtained by calculation for each state to the Viterbi processor 107. The Viterbi processor 107 outputs, to the output terminal 108, the transmission signal estimation value at a certain specific time in the sequence in which the sum of all the times of the metric of the equation (8) is the smallest. The operation of the Viterbi processor 107 is similar to that of the sequence estimation device of Documents 1 and 2 or Japanese Patent Application No. 2-203436, and thus the details will be omitted.

FIG. 2 shows an embodiment of the second sequence estimation apparatus according to the present invention. Transmission line response storage circuit 10 of the sequence estimation device of FIG.
5 (corresponding to the transmission path storage circuit 205 in FIG. 2) and the branch metric calculation circuit 106 (corresponding to the branch metric calculation circuit 207 in FIG. 2), the transmission path response conversion circuit 206
Has been inserted. Transmission line response conversion circuit 206
Is the Viterbi processor 2 similarly to the transmission line response inspection circuit 204.
While the connection information of the surviving series is supplied from 08,
From the transmission path memory circuit 205, not only the transmission path response driving value vector _{t, ls} at the present time but also the transmission path response estimation value vector _{t−1, ls} , vector _{t−2, ls} , ... . Then, the transmission path response conversion circuit 206, based on the connection information of the survival series, the transmission path response estimation value vector _{t, ls} for the current state for all states at the current time, and the surviving transition to the state. Vectors of estimated channel response values for past multiple times regarding the path
A transmission path response estimated value vector _t for the state at the new current time is obtained by a predetermined conversion operation from _{t−1, ls} and vectors _{t−2, ls} . For example, in the case of obtaining a new estimated value of the current time from the estimated values of the channel response at the current time and the past two times by simple averaging, 1) The estimated channel response value vector _{t, ls at} the current time of the state
From the transmission path response storage circuit 205, 2) check the state of the surviving path transiting to the state one hour before, two hours before, from the surviving sequence connection information supplied from the Viterbi processor 208, 3) 2 ), The transmission line response estimation value vector _{t−1, ls} at that time in the past with respect to the state of one hour before and two hours before,
The vector _{t-2, ls} is obtained from the transmission line response storage circuit 205. 4) The transmission line response estimation value vector _t for the state at the new current time is _t = ( _{t, ls} + _{t-1, ls} + _{t-2 , ls} ) / 3 (9), and so on. By this conversion operation, the channel response estimation value is smoothed, and noise in the channel estimation process is effectively removed. In addition, the conversion operation gives the transmission path response estimated value an inertia having a high correlation with the past estimated value, so that the transmission path response used in the branch metric calculation circuit has continuity. That is, it is possible to prevent the phenomenon that the estimated value of the transmission line response greatly changes at each time, and it is possible to obtain the estimated value of the transmission line response that is stable and reliable in the true state. In this embodiment,
Although the simple averaging is used as the conversion operation in the transmission path response conversion circuit 206, other linear operations such as weighted averaging, a threshold is provided, and a plurality of transmission path response estimation values are selectively used to take an average. Non-linear statistical manipulation of may be performed.

In the above embodiment, the transmission line response inspection circuit 104 or 204
Checked only whether or not the transmission path response estimation value is indefinite, but in addition to the case where it is indefinite, it is also selected when a transmission path response unfavorable to the branch metric calculation circuit 106 or 207 is supplied from the transmission path response calculation circuit 103. The processing may be performed assuming that the estimated value is not a valid estimated value as in the case of uncertainties, and the transmission path response estimated value at the previous time may be used as an alternative value. Further, in the above embodiments, the description has been given based on the transmission line in which a plurality of delayed waves exist with respect to the main wave, but the transmission line in which there are responses of a plurality of preceding waves to the main wave. Obviously, the sequence estimation device of the present invention is effective even for a transmission line in which a preceding wave and a delayed wave are mixed.

(Effects of the Invention) As described in detail above, the sequence estimation apparatus according to the present invention can always perform stable channel estimation for each sequence even in the presence of noise and follow a channel that changes at high speed. You can

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a first sequence estimation device according to the present invention, FIG. 2 is a block diagram showing an embodiment of a second sequence estimation device according to the present invention, and FIG. FIG. 4 is a block diagram showing a conventional sequence estimation device, FIG. 4 is a diagram for explaining a transmission line response, and FIG. 5 is a diagram for explaining a transmission line model. 101,201,301,501 ...... Input terminal, 102,202,302 ...... Register, 103,203 303 ...... Transmission line response calculation circuit, 104,204 ......
Transmission line response inspection circuit, 105,205 ... Transmission line response storage circuit, 206 ... Transmission line response conversion circuit, 106,207,304 ... Branch metric calculation circuit, 107,208,305 ... Viterbi processor, 108,209,305,506 ... Output terminal, 401 ... Main wave response , 402 ... delayed wave response, 502 ... delay element, 503 ... multiplier, 504, 505 ... adder.

Claims (2)

(57) [Claims] 1. A register for storing a plurality of sample values of a received signal, and a transmission for inputting a plurality of the sample values from the register and estimating a channel response at a current time for each of a plurality of sequences by a direct solution method. Path response calculation circuit, and checks the legitimacy of the channel response for a plurality of sequences obtained by the channel response calculation circuit, and outputs the channel response calculation circuit as the current time channel response if the channel response is correct. If not valid, a transmission line characteristic inspection circuit that outputs the transmission line response at the previous time indicated by the connection information of the survival sequence from the plurality of transmission line responses at the previous time as the transmission line response at the current time, and the transmission line characteristic A transmission line characteristic storage circuit that stores the transmission line response output by the inspection circuit for a plurality of times respectively for a plurality of series, and a transmission line response storage circuit that stores the transmission line response at the current time stored in the transmission line response storage circuit. And a branch metric calculation circuit that obtains a virtual received signal point for each sequence and obtains a distance from the sample value of the received signal, and a received signal is determined by a Viterbi algorithm by receiving the output of the branch metric calculation circuit. A sequence estimation device, comprising: a Viterbi processor that outputs connection information of a survival sequence to the transmission line response inspection circuit. 2. A register for storing a plurality of sample values of a received signal, and a transmission for inputting a plurality of the sample values from the register and estimating a channel response at a current time for each of a plurality of sequences by a direct solution method. Path response calculation circuit, and checks the legitimacy of the channel response for a plurality of sequences obtained by the channel response calculation circuit, and outputs the channel response calculation circuit as the current time channel response if the channel response is correct. If not valid, a transmission line characteristic inspection circuit that outputs the transmission line response at the previous time indicated by the connection information of the survival sequence from the plurality of transmission line responses at the previous time as the transmission line response at the current time, and the transmission line characteristic A transmission path characteristic storage circuit that stores transmission path responses output by the inspection circuit for a plurality of times respectively for a plurality of series, and for each series supplied from the transmission path characteristic storage circuit. A transmission line response conversion circuit that obtains a transmission line response conversion value from the transmission line responses for the current and past plural times according to a predetermined conversion rule, and a transmission line response conversion value given by the transmission line response conversion circuit. A branch metric calculation circuit that obtains a virtual received signal point for the sequence and obtains a distance from the sample value of the received signal, and a received signal is determined by a Viterbi algorithm by receiving the output of the branch metric calculation circuit and the survival sequence A sequence estimation device comprising a Viterbi processor that outputs connection information to the transmission path response inspection circuit.