JPS63200624A - Adaptive equalizing system for gmsk synchronizing detection - Google Patents

Abstract

PURPOSE:To reduce the effect of jitter of a recovered timing signal by utilizing the in-phase and orthogonal signal of an error when a signal crosses a zero point. CONSTITUTION:An error signal converting section 4 uses timing information of a timing recovery section to decide a sample point of an in-phase or orthogonal signal, and decides whether or not an orthogonal signal in case of the in- phase signal sampling and an in-phase signal in case of the orthogonal signal sampling crosses a zero point from the sampling point to preceding and succeeding T(sec) (T is one bit length) depending on a tap data from a transversal filter section and converts an error signal based on the information. That is, since the orthogonal signal when it crosses the zero point at the in-phase signal sampling or the in-phase signal when the orthogonal signal crosses the zero point at the orthogonal signal sampling, has less nonlinear distortion caused in the process of modulation, the estimated error of the in-phase and orthogonal signal is utilized in the case of zero crossing only so as to separate the distortion due to band limit before FM modulation at transmission from the distortion due to a reception filter.

Description

[Detailed Description of the Invention] (1) Description of the technical field to which the invention pertains
This relates to an adaptive equalization method in SK synchronous detection.

(2) Description of conventional technology There is currently no report on applying an adaptive equalization method to GMSK.

(3) Purpose of the Invention The present invention provides an orthogonal signal when the quadrature signal crosses the zero point when sampling the in-phase signal, or an in-phase signal when the in-phase signal crosses the zero point when sampling the quadrature signal, in the received signal. , less nonlinear distortion occurs in the modulation process,
Focusing on the fact that most of the distortion is linear distortion that occurs during reception, the feature is that it is possible to separate the distortion caused by the band limitation before FMf tuning during transmission and the distortion caused by the reception filter by checking if it crosses the zero point. The purpose is to reduce the residual distortion after equalization,
The purpose of the present invention is to reduce both distortion at points other than sample points, and to reduce the influence of jitter on a reproduced timing signal.

(4) Explanation of the configuration and operation of the invention When applying an adaptive equalizer to GMSK, it is necessary to know the types and mechanisms of distortion that occur, so we will first briefly describe the GMSK Rili system, and then explain the present invention. explain.

The GMSK modulation method is N RZ (Non Retur
nt.

Zero) signal to Gaussian LPF (Low P
a5s Filter) and performs FM modulation. Therefore, MS K (Minimum 5hif
Band-limiting the baseband signal of t Keying),
This can be seen as an attempt to narrow the band. In the case of MSK, the phase change is the data mark (hereinafter referred to as H).
When , the speed is constant. Therefore, the initial phase is 01 data (M, S.

M, M, S, 5.5), the phase change is as shown by the solid line in FIG. 1(a). On the other hand, in the case of GMSK, due to the band-limiting effect of the baseband signal, the phase change becomes gradual at the point where the polarity of data changes.

Therefore, the phase change becomes as shown by the dotted line in FIG. 1(a).

Generally, a GMSK transmission signal is as shown in the following equation.

5(t)=A-cos(ωct+1(t)+φe)
(1) A: Amplitude ω of the transmitted signal, : Carrier angular frequency d (tri instantaneous phase shift φ.: Initial phase coherent detection, cos(ωct+φ.) and −5in(ω
ct" (Its). As a result, the in-phase signal (u, (1)) and quadrature signal (u, (1)) of the baseband signal after detection are as shown in equation (2). Become.

u+(t)=eos(tA(t)) uo(t)=sin(φ(t)) In Fig. 1(b) and (c), if the phase changes as shown in Fig. 1(a) , cos(1(t)), 5in(φ(
t)).

The data is recovered by sampling uc(t) and u,(t) and transforming the logic. here,
In the case of the sample point in FIG. 1, the in-phase signal is t=o,
2T, 4L...The orthogonal signal is t=L 3L 5L・
It is...

GMSK has the advantage that the signal bandwidth is narrower than MSK, making it possible to use frequencies more effectively.
As can be seen from FIGS. (b) and (c), the distortion at the sample points becomes large.

Furthermore, if the signal is passed through a reception filter to remove out-of-band noise during reception, the distortion becomes even greater.

In digital communications, GMSK is essential because this distortion causes judgment errors during data demodulation.

An equalizer is most suitable for reducing this distortion, but in applying it, it is first necessary to elucidate the mechanism by which this distortion occurs, and then to take a configuration suitable for that purpose.

There are two main causes of distortion in GMSK received signals. One is distortion caused by band limiting before FM modulation during transmission, and the other is distortion caused by a reception filter.

First, consider distortion caused by band limiting before FM modulation during transmission.

The baseband signal in that case has the following two major properties.

The first point is a point where the data polarity is switched and the band limiting effect is large, and in a channel that is not a sample point (for example, 5 inch (4 (t)) at t = 21), adjacent points around that point Two sample points (5
in(φ(T)) and 5in(φ(3r))) become larger (for example, at t=2T, the distortion of cos(4(t)), which is the sample point, becomes larger than the distortion of 5in(1(t)). ) has a larger distortion than ).

The second point is a channel that does not become a sample point at a point where the data polarity does not switch and the band limiting effect is small (for example, at t = 3'r, cos(+It (t)))
, two adjacent sample points (cos(1(2r)) and cos(dr(4T)) centering on that point
) ) have opposite signs, and at that point, both channels have a property that there is almost no distortion.

Next, consider distortion caused by the reception filter.

Distortion due to band limitation of the reception filter becomes large when two adjacent sample points have opposite signs in each channel.

Therefore, if adjacent diagonal sample points have different signs,
At the zero cross point between these two points, most of the distortion is due to the reception filter, and if this information is effectively used, the band before FM modulation during transmission can be controlled.

The present invention is constructed using the above properties.

The configuration of the present invention is shown in FIG. In the figure, 1 is a transversal filter section, 2 is a reference signal generation section, 3 is an error signal generation section, 4 is an error signal conversion section, 5 is a tap gain update section, and 6 is a timing regeneration section. Hereinafter, a detailed description will be given with reference to the drawings.

The transversal filter section (1) delays an input signal and weights and synthesizes each delay component, and is configured with, for example, a tapped delay line.

The data sampling timing at this time is obtained from the signal from the timing reproducing section (6).

The reference signal generation unit (2) converts the ideal values of the transversal filter outputs y+(t) and yc+(t) into a reference signal r.
Let l(t> and r, (t)) and generate as follows.

■When sampling an in-phase channel■When sampling a quadrature channelHowever, y(t)=y+(t)+j-yo(t)
(5) r(t)=r+(t)+j-ro
(t> (6) Also, r, (t>,
y+(t) are the in-phase signals of r(t) and y(t), respectively, and ro(t>, yo(t) are the in-phase signals of r(t) and y(t), respectively.
(t), shows orthogonal signals of y(t).

On the other hand, the output e(t) of the error signal estimator (3) is a(t
)=r(t)−y(t) (8
). Here, e(t)=e+ (t)+j−eo(t>
(9) and e+(t
>, e, (t) denote the in-phase and quadrature signals of e(t), respectively.

e, (t) and eQ(t) are directly input to the error signal converter (4).

The error signal converting section (4) determines whether the sample point is an in-phase or quadrature signal based on the timing information from the timing reproducing section, and at the same time determines whether the sample point is an in-phase signal or a quadrature signal when sampling an in-phase signal, or When sampling quadrature signals, it is determined whether the in-phase signal crosses the zero point within T(see) (T is 1 bit length) before and after the sampling point. Then, based on this information, the error signal is converted as follows.

■When sampling the in-phase signal a When the quadrature signal crosses the zero point e(t) = e+(t)+j-so(t)
(to)b When the orthogonal signal does not cross the zero point, e(t) = sr(t) (1
1) ■When sampling a quadrature signal a When the in-phase signal crosses the zero point e(t)=e+(t)+j-eo(t>
(12)b When the in-phase signal does not cross the zero point, e(t)=j'ea(t) (
13) The physical meaning of such processing by the error converter (4) can be considered as follows. However, the explanation will be given only when sampling the in-phase signal.

When the quadrature signal crosses the zero point, if there is no distortion due to the reception filter, the quadrature signal becomes 0 at the sample point of the in-phase signal.

On the other hand, if there is a reception filter, eQ(t) includes only the distortion due to the reception filter. On the other hand, Ql(t) includes both distortion during transmission and distortion due to the reception filter.

Therefore, if we use e,(t) to update the tap gain, we get
Both distortions can be separated, improving equalization performance.

In the tap gain updating unit (5>), the tap gain is determined by the gradient algorithm or Ka1ma algorithm for each tap data of the transversal filter unit (1).
n algorithm etc. can be applied.

The timing reproducing section (6) reproduces timing signals required in the transversal filter section (1), the error signal detecting section (2), and the error signal converting section (4). Here, the period of the timing signal necessary for the transversal filter section (1) is determined by the frequency characteristics necessary for the transversal filter. On the other hand, the cycle of the timing signal necessary for the error signal detection section (3) and the error signal conversion section (4) is determined by the tap update interval.

In Fig. 3, the bandwidth (Bb) of the transmitting Gaussian filter is shown as Bb
- T = 0.25 (T is 1 bit length), the bandwidth of the receive filter is T = 0.6 (the shape is Gaussian), the delay interval of the transversal filter is T/3, tap The EYE pattern in the case of Equation 21 is shown. (a) is an EYE pattern when the present invention is not applied, and (b) is an EYE pattern when the equalization method of the present invention is used. Also, the left side is an in-phase signal, and the right side is a quadrature signal.

The distortion rate at the sample point is -12,5d in case (a)
In the case of B, (b), it is -32.5 dB, and the degree of distortion improvement according to the present invention is 20 dB.

From the above, it is clear that the equalization method of the present invention is effective.

In realizing the present invention, each part can be constructed using analog circuits, but it is also possible to construct some or all of these parts using digital signal processing.

In addition, MSK (Minimum 5hift Keyi)
ng) corresponds to GMSK without the transmission filter, so the present invention is also applicable to MSK.

(5) Effects of the Invention According to the present invention, distortion after demodulation can be significantly reduced compared to a system that does not apply an equalization method, and furthermore, it is possible to further narrow the band of the reception filter. In communications, especially land mobile communications, frequencies can be used more effectively.

[Brief explanation of the drawing]

Figure 1 shows MSK and GM! 3 shows the waveforms of each part in the transmission and reception process, FIG. 2 is a block diagram of the present invention, and 3rd shows the PIE patterns when the present invention is not applied and when it is applied. In Fig. 2, 1 is a trans/<-sulfur filtration part,
2 is a reference signal generation section, 3 is an error signal generation section, 4 is an error signal conversion section, 5 is an evening/knob gain update section, and 6 is a timing reproduction section. Patent Applicant: Ministry of Posts and Telecommunications Radio Research Institute Director, Engraving of drawings T (seC) Figure 3 (ime a) Genital organs on drawings ime b) Procedural amendment form) June 27, 1988

Claims (1)

[Claims] GMSK (Gaussian filtered Mi
(nimum ShiftKeying) synchronous detection method, a first means for delaying and synthesizing received signals, a second means for generating a reference signal, and a third means for generating a difference signal between the first means and the second means. and from the in-phase and quadrature signals of the error signal generated by the third means, determine whether or not the quadrature signal crosses the zero point when determining the in-phase signal, and whether the in-phase signal crosses the zero point when determining the orthogonal signal;
If the error has crossed the zero point, the in-phase and quadrature signals of the error are used, and if the zero point has not been crossed, the in-phase signal is used when determining the in-phase signal, and the quadrature signal is generated when determining the quadrature signal. and a fifth means for controlling the signal synthesis weight of the first means based on the fourth means; A GM characterized in that distortion contained in a received signal is removed by comprising a sixth means for generating a timing signal to be used.
Adaptive equalization method for SK synchronous detection.