JPH05292139A - Receiver estimating maximum likelihood series - Google Patents

Abstract

PURPOSE:To improve the reception characteristic with less processing quantity. CONSTITUTION:The receiver is provided with a transversal matching filter 5, a transmission line estimate circuit 9 estimating an impulse response of a transmission line, and a state estimate circuit 6 estimating a transmission symbol series from an output of the matching filter 5 based on the estimated impulse response. Then the transmission line estimate circuit 9 sets a time interval to the estimated impulse response so that a largest sample points in the order of the amplitude from a sample point having the maximum amplitude corresponding to a tap number of the matching filter 5, sets a tap coefficient based on only the sample points in the time interval and estimates a transmission symbol series. The optimum impulse response of the transmission line whose length is NT is estimated by implementing the amplitude comparison of the signals without calculation of a square sum of the amplitude of the signals.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a maximum likelihood sequence estimation receiver used in high speed digital mobile communication.

[0002]

2. Description of the Related Art A conventional maximum likelihood sequence estimation receiver of this type is described in "IEEE Journal on Selected Areas in Communi
cations, Vol.7, No.1, pp.122-129, January 1989 』
It is described in. In high-speed digital mobile communication, transmission characteristics are greatly deteriorated due to frequency selective fading due to multipath propagation. Then, in order to compensate for this deterioration, the use of a maximum likelihood sequence estimation receiver is often studied. The maximum likelihood sequence estimation receiver filters out-of-band noise of the received signal, then digitizes the signal and passes through a matched filter that minimizes the effect of noise,
The maximum likelihood of the transmitted symbol is estimated from the output using the Viterbi algorithm. The tap coefficient of the matched filter estimates the impulse response of the transmission line before or in parallel with the data processing, and is set based on this. The estimated impulse response is also used for maximum likelihood estimation of transmitted symbols. Maximum likelihood sequence estimation receivers generally have better reception characteristics under frequency selective fading as compared to other adaptive equalization receivers that compensate for this degradation. However, the amount of processing rapidly increases as the number of taps of the matched filter increases, and thus it has been conventionally considered difficult to realize. However, recently, due to the development of digital signal processing technology, studies for realization have been actively conducted.

In the maximum likelihood sequence estimation receiver, it is necessary to minimize the number of taps of the matched filter in order to keep the processing amount small. Therefore, when it is necessary to consider multipath delayed until time τ, the sampling period of the signal is set to T
As a result, the number of taps of the matched filter is often N = τ / T + 1. However, in this division, the mantissa part is rounded up to be an integer. On the other hand, since the estimated impulse response of the transmission line is spread over a period longer than the normal time τ, it is necessary to terminate the impulse response in the period NT. The optimum way to take this period NT for the impulse response depends on the multipath situation. If the period NT is always set at the same position, the reception characteristic will be deteriorated. Therefore, when there is no multipath, the peak of the signal is
However, if there are multipaths, it is generally necessary that the peaks of both the main signal and the multipaths fall within the period NT. As shown in the above document, conventionally, the energy within the period NT is set to be maximum.

[0004]

However, in the conventional maximum likelihood sequence estimation receiver described above, the period N
The signal energy in T needs to be calculated for all possible positions of the period NT, and the calculation of energy is the sum of squares of the signal amplitudes, so a large amount of calculation is required to determine this position. There was a problem that was. Therefore, an object of the present invention is to provide a maximum likelihood sequence estimation receiver with a small amount of calculation and good reception characteristics.

[0005]

In order to solve the above-mentioned problems, the present invention provides a transversal type matched filter for minimizing the influence of noise and a transmission line estimation circuit for estimating the impulse response of the transmission line. And a maximum likelihood sequence estimation receiver having a state estimation circuit that estimates a transmission symbol sequence from the output of the matched filter based on the impulse response estimated by the transmission path estimation circuit, the transmission path estimation circuit,
The estimated impulse response should include the most sample points corresponding to the number of taps of the matched filter and having the amplitudes larger than a predetermined threshold value in descending order of amplitude from the sample point having the largest amplitude. Maximum likelihood sequence estimation receiver so as to set a different time interval and set the tap coefficient of the matched filter and estimate the transmission symbol sequence in the state estimation circuit based only on the sample points within this time interval. Is configured.

[0006]

According to the present invention, since the maximum likelihood sequence estimation receiver is constructed as described above, the optimum transmission line of length NT can be obtained by simply comparing the amplitudes of the signals without calculating the sum of squares of the signal amplitudes. Can estimate the impulse response of. Therefore, the above problems can be solved.

[0007]

1 is a block diagram showing an embodiment of a maximum likelihood sequence estimation receiver of the present invention. In FIG. 1, 1 is a received signal input terminal, 2 is a frequency conversion circuit, 3 is a low-pass filter (hereinafter referred to as LPF), and 4 is an analog / digital conversion circuit (hereinafter referred to as A / D conversion circuit). ), 5
Is a matched filter, 6 is a state estimation circuit, 7 is an estimated signal output terminal, 8 is a coefficient setting circuit, 9 is a transmission path estimation circuit, 10 is an estimation circuit, 11 is a position setting circuit, and 12 is a signal generation circuit. A signal in the radio frequency band received from the antenna of the receiver reaches the reception signal input terminal 1. This signal is
The frequency conversion circuit 2 converts the signal into a baseband signal by synchronous detection or the like. At this time, the modulation method is QPS
In the case of a quadrature modulation type system such as K or MSK, two in-phase components and quadrature components are output. In this case, in FIG. 1, after the frequency conversion circuit 2, all signals in phase and in quadrature are handled. The LPF 3 removes noise outside the desired frequency band from the signal from the frequency conversion circuit 2, and the A / D conversion circuit 4 converts the signal into a digital signal. Then, this digital signal is first input to the matched filter 5.

FIG. 2 is a block diagram showing a configuration example of the matched filter 5. 2, 13 is an input terminal, 14 is an adder circuit, 15 is an output terminal, 21, 22 ..., 2N is a delay circuit, 30, 31 ..., 3N is a multiplier circuit, and C0, C.
1, ..., CN are tap coefficients. FIG. 2 shows a so-called transversal type filter. FIG. 2 is a diagram corresponding to the case where the order N is 5. The digital signal input from the input terminal 13 has delay circuits 21, 22, ..., 2 each having a delay time of a period T of N digital signals.
After passing through N, the input terminal 13 and the delay circuits 21, 22, ...
The respective output signals from 2N are multiplied by the multiplication circuits 30, 3
Tap coefficients C0, C1, and
..., CN are multiplied, and all of these output signals are added by the adder circuit 14 and output to the output circuit 15. In the quadrature modulation type, it is usual to represent a signal by a complex number with the in-phase component as the real part and the quadrature component as the imaginary part. In this case, the tap coefficients C0, C1, ..., CN are complex numbers.
The tap coefficients C0, C1, ..., CN need to be set to be a time-reversed complex conjugate of the impulse response of the transmission line in order to operate as a matched filter that minimizes the influence of noise. This is performed in the coefficient setting circuit 8 described later.

In FIG. 1, the output signal from matched filter 5 is subjected to maximum likelihood estimation of a transmission symbol in state estimation circuit 6, and the result is output to estimated signal output terminal 7. Maximum likelihood estimation in the state estimation circuit 6 is usually performed using the Viterbi algorithm. The Viterbi algorithm sequentially selects and determines the most probable transmission symbol by using an evaluation amount called a metric. Then, the impulse response of the transmission line is required to calculate this metric. Since the impulse response of the transmission path is usually unknown, the impulse response of the transmission path is estimated by the transmission path estimation circuit 9 to be described later, and this is used. The digital signal from the A / D converter 4 is also supplied to the transmission path estimation circuit 9. The transmission path estimation circuit 9 estimates the impulse response of the transmission path.

FIG. 3 is a diagram showing a configuration example of a burst signal used in high speed digital mobile communication. This shows an example of the GSM system, which is a European digital mobile communication system. A known sequence called a training sequence on the receiving side is sent near the center of the burst signal. This training sequence is usually defined to have an ideal impulse-like autocorrelation characteristic.

In FIG. 1, the transmission path estimation circuit 9 has a signal generation circuit 12 therein to generate a digital signal by modulating a signal that circulates through the training sequence as a baseband signal corresponding to a modulation method. This digital signal is first correlated with the digital signal from the A / D conversion circuit 4 in the estimation circuit 10. The signal after taking this correlation is the time-reversed impulse response of the transmission path. Therefore, the estimation circuit 10
Inverts this signal further in time. The output signal of the estimation circuit 10 which has been subjected to the correlation operation and the time reversal shows the estimated value of the impulse response of the transmission path.

FIG. 4 is a diagram showing an example of the impulse response of the transmission line. (A) shows an example when there is no multipath, and (b) shows an example when there is multipath. The scale on the time axis shows the sample points of the digital signal, and the numbers show the sample times. Especially when there are multiple passes,
The impulse response of the transmission line has a value that cannot be regarded as zero for a period longer than the period NT that can be covered by the matched filter. Therefore, it is important to consider which part of the impulse response of the transmission path is considered by the matched filter 5 and the state estimation 6. This position setting is performed by the position setting circuit 1 in FIG.
One does. When there is no multipath and the length of the impulse response of the transmission path is relatively short, the signal peak may be near the center of the period NT, as indicated by the thick double-headed solid arrow in FIG. .. However, as shown by the thick double-headed arrow in FIG. 4B, when there are multipaths that are greatly delayed, it is necessary to allow both the peaks of the main wave and the subwave having large amplitude to enter the period NT. The position setting circuit 11 in FIG. 1 uses the period N so as to satisfy the above.
Set T.

FIG. 5 is a flow chart showing an example of the position setting algorithm in the position setting circuit 11. In the algorithm, first, the sample time is rearranged in descending order of amplitude with respect to the estimated sampling point of the impulse response of the transmission path from the estimation circuit 10 in the period to be considered. Various sort algorithms can be used for this. Next, a large sample is taken in order, and the position of this period NT is set so that the largest number of samples enter the period NT. At this time, considering a too small amplitude, as shown by a thick double-sided dotted arrow in FIG.
A small amplitude is overestimated too much and deviates from the position of the optimum period NT. Therefore, the amplitude smaller than the predetermined threshold is ignored. In FIG. 5, this is done by changing the maximum number. In addition, in FIG. 5, the period NT is represented by [lower limit, upper limit].

In FIG. 1, the position setting circuit 11 supplies to the coefficient setting circuit 8 and the state estimation circuit 6 the estimated sampling points of the impulse responses of the N transmission lines within the period NT thus set. The coefficient setting circuit 8 takes the complex conjugate of the estimated sample points of the impulse responses of the N transmission lines, sets C0 on the upper limit side and CN on the lower limit side, and sets the tap coefficient of the matched filter 5 to the opposite. Set. Further, the state estimation circuit 6 uses the estimated sample points of the impulse responses of the N transmission lines for calculating the metric for the Viterbi algorithm.

[0015]

As described above in detail, according to the present invention, the optimum impulse response of the transmission line having the length NT can be obtained only by comparing the amplitudes of the signals without calculating the sum of squares of the signal amplitudes. It is possible to provide a maximum likelihood sequence estimation receiver that can be estimated, has a small amount of calculation, and has good reception characteristics.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a maximum likelihood sequence estimation receiver of the present invention.

FIG. 2 is a block diagram showing a configuration example of a matched filter.

FIG. 3 is a diagram showing a configuration example of a burst signal.

FIG. 4 is a diagram showing an example of impulse response of a transmission line.

FIG. 5 is a flowchart showing an example of a position setting algorithm of the present invention.

[Explanation of symbols]

 1 reception signal input terminal 2 frequency conversion circuit 3 low pass filter 4 analog / digital conversion circuit 5 matching filter 6 state estimation circuit 7 estimation signal output terminal 8 coefficient setting circuit 9 transmission path estimation circuit 10 estimation circuit 11 position setting circuit 12 signal Generation circuit 13 Input terminal 14 Addition circuit 15 Output terminal 21, 22, ..., 2N delay circuit 30, 31, ..., 3N multiplication circuit C0, C1, ..., CN Tap coefficient

Claims (1)

[Claims] 1. A transversal-type matched filter that minimizes the influence of noise, a transmission path estimation circuit that estimates the impulse response of the transmission path, and the matching based on the impulse response estimated by this transmission path estimation circuit. In a maximum likelihood sequence estimation receiver having a state estimation circuit that estimates a transmission symbol sequence from the output of the filter, the transmission path estimation circuit, the estimated impulse response,
A time interval is set that corresponds to the number of taps of the matched filter and includes the largest number of sample points having an amplitude equal to or larger than a predetermined threshold value in descending order of amplitude from the sample point having the largest amplitude. A maximum likelihood sequence estimation receiver characterized in that the tap coefficient of the matched filter is set and the transmission symbol sequence is estimated in the state estimation circuit based on only the sample points within this time interval.