JPH09200178A - Spread spectrum communication equipment and spread spectrum communication method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To allow a canceller to be converged at a high processing speed. SOLUTION: A 1st user sends a data series DI. A multiplex circuit MPX1 selects its input from a signal to a training signal generator TR1 based on a timing signal received from a timing terminal. The training signal generator TRI provides an output of symbols of an orthogonal signal series N' for the 1st user. The output symbol series of the multiplex MPX1 is spread by a spread code by a spread circuit SP1 similarly to the case with a conventional spread spectrum system. The circuits for other users are operated similarly to above.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a spread spectrum communication device and a spread spectrum communication method.

More specifically, the present invention is a spread spectrum communication apparatus and a spread spectrum communication device suitable for quickly converging an interference canceller by orthogonalization when performing spread spectrum communication by a direct spread code division multiple access system. It relates to a communication method.

[0003]

2. Description of the Related Art In recent years, various spread spectrum methods have been studied in order to effectively use frequencies in digital mobile communication (MK Simon, JK Omura, RA S.
choltzand BKLevitt, “Spread Spectrum Communi
cation ", Computer Science Press Publishing, 1985).
In particular, a CDMA (Code Division Multiple Access) method using a direct sequence (DS) method is considered to be relatively easy to put into practical use because its configuration is relatively simple.

In this DS-CDMA system, a plurality of users simultaneously use the same carrier frequency. Each user uses different spreading codes, but since these spreading codes are correlated with each other, the components of other users should be mixed in the despreading signal even when despreading with the spreading code of the desired wave. become. Therefore, when the number of other users is large, the level of the interference wave component mixed in the despread signal increases, and the transmission characteristic deteriorates significantly. This deterioration becomes larger as the reception level of other users becomes higher than the reception level of the desired wave.

It is known that the deterioration of the transmission characteristics due to such cross-correlation between spreading codes can be solved by adding an interference canceller to the receiver.

A blind type is known as a type of interference canceller. This is called an orthogonalized matched filter, and its basic configuration is shown in Fig. 1 (K. Fukaw
a, and H. Suzuki, “Orthogonalizing Matched Filter
(OMF) detection for DS-CDMA mobile radio system
s ”, IEEE Globecom 1994, pp.385-389, Nov.1994.).
This interference canceller operates as follows.

In FIG. 1, a reception signal SIG having the in-phase component amplitude and the quadrature component amplitude of the received wave is input from the input terminal Q1. In the sampling means SMP, the received signal SIG
Is sampled at fixed time intervals and the sampling signal SPS is sampled at the terminal Q.
Output to 2. The signal extraction means EXTR receives the sampled signal SPS, performs despreading and linear combination operations, and outputs a combined signal DCS to the terminal Q5. In the demodulation means DET,
The combined signal DCS is demodulated and the determination signal OPS is output to the terminal Q6. The timing control means TIM controls the operation timing of each of the above means.

Here, the signal extraction means EXTR is used for despreading and
It is composed of synthesizing means DES.COM and coefficient control means C-CONT, and these means operate as follows. First, the despreading / combining means DES / COM performs despreading and linear combination processing of the sampled signal using the weighting coefficient, and the converted composite signal DCS and the multiplied signal by which the weighting coefficient W is multiplied. MPS and MPS are output to terminals Q5 and Q3, respectively. Next, the coefficient control means C-CONT causes the multiplied signal M.
PS and the composite signal DCS are input, and the weighting coefficient W obtained by the algorithm that minimizes the average power of the composite signal under the constraint condition of the weighting coefficient W is output to the terminal Q4.

Various algorithms are conceivable for minimizing the average power of the combined signal under the constraint condition of the weighting coefficient, but a simple method is LMS with constraint condition.
Frost's method (Frost, OL, “An algorithm for linea
rly constrained adaptive array processing ”, Proc.
IEEE, vol.60, No.8, PP.926-935, August 1972) are known.

Next, FIG. 2 shows a configuration example in which the despreading means DESP and the linear synthesizing means LCOM are connected in cascade as the signal extracting means EXTR. The despreading means DESP outputs a plurality of despread signals obtained by despreading the sampled signal with a plurality of despreading codes. Also, a plurality of these despread signals are output as the multiplied signals. The linear synthesizing means LCOM multiplies the plurality of despread signals obtained by the despreading means DESP by weighting factors and outputs a synthesized signal. However, in FIG. 2, the spreading factor is 4 and the number of users using the same frequency is 4 in order to simplify the description.

Next, the signal extraction means EXTR shown in FIG.
Will be described. First, a sampling signal is input from the input terminal Q2. The despreading means DESP is composed of four matched filters MF1 and OCF1 to OCF3. The matching filter MF1 uses a spreading code of the desired wave, and the matching filters OCF1 to OCF3 use spreading codes orthogonal to the spreading code of the desired wave and mutually orthogonal. These matched filters perform correlation calculation between the sampled signal and the spread code,
The despread signal is output as the multiplied signal MPS.

The linear synthesizing means LCOM includes multipliers M1 to M.
4 and an adder ADD, a plurality of despread signals are multiplied by a weighting coefficient to generate a combined signal DCS, which is output from the output terminal Q5. The coefficient control means C-CONT is
A plurality of despread signals and the combined signal are input, and a weighting coefficient W obtained by an algorithm that minimizes the average power of the combined signal under the constraint condition of the weighting coefficient is output.

This algorithm does not require a detection output, that is, decoded data information, and is therefore called a blind type. Also, when controlled by this algorithm,
Output SINR (Signal to Interference and Noise Ra
tio) is also known to be maximum. The matched filter MF1 and OCF1 to OCF3 can be replaced with a correlator, and the same applies to the matched filter described below.

When the optimum value of the four-dimensional weighting coefficient vector at this time is Wo = [Wo ₁ Wo ₂ Wo ₃ Wo ₄ ] ^T ,

[0015]

[Equation 1]

## EQU1 ## Here, α is a scalar value, R is a correlation matrix of a 4 × 4 despread signal, and T is a four-dimensional steering vector. R is the despread signal X (i) = [x
₁ (i) x ₂ (i) x ₃ (i) x ₄ (i)] ^T ,

[0017]

[Equation 2]

It becomes like this. Where i is the time in units of the symbol period T, Wo _j is the optimum value of W _j , x _j (i) is the despread signal at the time i in the j-th matched filter, ^T
Is the transposed matrix, ^H is the complex conjugate transposed matrix, and <> is the set average. This R can be approximated as follows.

[0019]

(Equation 3)

However, N _t is a very large natural number. The steering vector T is in this case

[0021]

(Equation 4)

Do as follows.

The control of the weighting coefficient is performed so that the signal level of the desired wave included in the combined signal is kept constant. Since the spreading codes of the matched filters OCF1 to OCF3 are orthogonal to the spreading code of the desired wave, x ₂ (i) to x ₄ (i) do not include the signal component of the desired wave. Considering this, the constraint condition of the weighted coefficient is

[0024]

(Equation 5)

## EQU2 ## The above α is set so that Wo satisfies this constraint condition.

The signal extraction means EXTR described above is represented as a cascade connection of the despreading means DESP and the linear combining means LCOM. Since each of these means is a linear means, these two processes can be combined into one linear process. At this time, it is known that this one linear processing can be realized by one transversal filter (the same document as the document of the configuration of FIG. 1). Even in this case, since the result of the processing is the same, the orthogonalization processing is defined by the equation (1) using the inverse matrix of the correlation matrix of the equation (2).

As another type of interference canceller, a non-blind type is known (Yoshida,
S., A. Ushirokawa, S. Yanagi and Y. Furuya, “DS / C
DMAadaptive interference canceller on differential
detection in fastfading channel ”, Proc. 44nd Veh
icular Techn. Conf., pp.780-784, June 1994).

FIG. 3 shows the configuration of an interference canceller using a non-blind type orthogonalization filter. This interference canceller has a tap length corresponding to several bit periods, and is composed of a quadrature filter (TVF) operating at a rate m times the chip rate (m = 1, 2, ...) And a delay wave detector DD. To be done. The orthogonalization filter TVF realizes the despreading means and the linear synthesizing means by one, as in the above-mentioned example.

The orthogonalization filter TVF receives the received signal of several bit periods as input, and adaptively forms tap weights (despreading codes of its own station) orthogonal to each spreading code of all other stations. Other station interference is removed, and the local station signal component is extracted. The tap weights of this orthogonalization filter TVF are adaptively updated for each symbol, and a demodulated signal is obtained for each symbol.

The differential detector DD is an orthogonalization filter TVF.
Detection is performed from the signal of only the local station from which the interference of other stations is removed by the output of.

In the interference canceller having the orthogonalization filter TVF alone, it is necessary to control the tap weights of the orthogonalization filter TVF so that the interference cancellation and the carrier phase tracking are simultaneously satisfied. It is sufficient to form tap weights exclusively for interference removal by separating the synchronization function, and the requirement of high speed required for adaptive control for carrier phase tracking is relaxed. That is, tap weight control must be performed in a complex manner corresponding to the complex signal introduced to the input terminal IN, but when phase synchronization is not required,
The tap control of the orthogonalization filter only needs to be performed for real numbers. Thus, since it is not necessary to follow the change of the carrier phase, the high speed required for the adaptive control is relaxed.

However, since the output of the decision circuit DEC is determined to be a certain constant value, the orthogonalization filter TVF adjusts the level so that the input of the decision circuit DEC is also at the same level. In the configuration of FIG. 3, since the orthogonalization filter TVF is adaptively adjusted based on the determination value, it is a non-blind type. However, since the orthogonalization filter TVF minimizes the output error, the SINR of its output is maximized, and the operating principle thereof is the same as that proved in the blind form. That is, it is defined by the equation (1) using the inverse matrix of the correlation matrix R.

[0033]

As described above, the conventional method for canceling interference by orthogonalizing these input signals cannot exhibit its characteristics unless the synthesis coefficient converges. However, the method of establishing the convergence quickly is not yet known. In the non-blind type, training is performed, but the training signal needs to be known in advance on the receiving side.

Therefore, in view of the above points, an object of the present invention is to
It is an object of the present invention to provide a spread spectrum communication device and a spread spectrum communication method for speeding up convergence of a canceller in a communication system using an interference canceller by orthogonalization.

[0035]

In order to achieve the above object, in a spread spectrum communication apparatus according to the present invention, each user synchronously transmits orthogonal signals which are orthogonal to each other as a transmission symbol sequence. And a receiving means including an orthogonalization filter for removing cross-correlation by adaptive signal processing. In addition, other spread spectrum communication devices include a transmission unit that allows each user to synchronously transmit a signal having a sharp autocorrelation peak as a transmission symbol sequence, and a reception unit that includes an orthogonalization filter that removes cross-correlation by adaptive signal processing. And means.

In the spectrum communication method according to the present invention, on the transmitting side, each user synchronously transmits orthogonal signals that are orthogonal to each other as a transmission symbol sequence, and on the receiving side, cross-correlation is performed by adaptive signal processing. An orthogonal filter that removes is used.

Further, in other spread spectrum communication methods, at the transmitting side, each user synchronously transmits a signal having a sharp autocorrelation peak as a transmission symbol sequence, and at each receiving side, cross-correlation is performed by adaptive signal processing. An orthogonal filter that removes is used.

[0038]

BEST MODE FOR CARRYING OUT THE INVENTION In a spread spectrum communication system to which the present invention is applied, the transmitting means of each user is (1) orthogonal signals mutually orthogonal as a transmission symbol sequence, or (2) a transmission symbol sequence. A signal having a sharp autocorrelation peak is synchronously transmitted (see FIGS. 4 to 6), and the receiving means (3) removes cross-correlation by adaptive signal processing in the orthogonalization filter included therein (FIGS. 1 to 1). (See FIG. 3).

In this way, on the transmitting side, each user
Orthogonal signals that are orthogonal to each other as a transmission symbol sequence,
Alternatively, it is different from the related art that a signal having a sharp autocorrelation peak is synchronously transmitted.

[0040]

EXAMPLE First, the operating principle of this example will be described.

In the method using the orthogonalization filter, the coefficient W is the input correlation matrix R as already described in the equation (1).
Form of product of inverse matrix R ⁻¹ of R ⁻¹ and steering vector V R ⁻¹
It can be represented by V. Since the steering vector is constant, the inverse matrix of R, that is, R ⁻¹ changes according to the input. Since R is unknown in the initial state of operation, it is necessary to estimate R from the input signal. This measured R is denoted by Q.

Although R is an original convergence value, Q is obtained by measurement. Therefore, Q gradually converges on R. Further, Q fluctuates in the vicinity of R.
The details will be described below.

Let the code vectors be C ₁ and C ₂ . The output despread with the code C ₁ is v ₁ . At this time, v ₁ (i) at time i is

[0044]

(Equation 6)

It becomes Here, T represents transposition. A ₁ and A ₂ are the received signal amplitude of each code, d ₁ (i) and d
₂ (i) is the user data series of each code. Similarly C
_The output v ₂ (i) despread by ₂ is

[0046]

(Equation 7)

Is as follows. clearly,

[0048]

(Equation 8)

Is as follows.

[0050]

[Equation 9]

Is as follows. The (1,2) element of the correlation matrix R is r
_{If 1,2} ,

[0052]

(Equation 10)

Is as follows. However, <> represents the average,

[0054]

[Equation 11]

[0055] As can be seen from equation (11), r
_{The 1,2} component is proportional to ρ.

On the other hand, since the measured value Q is the average of N sample values, its (1,2) element q _1,2 (i) is

[0057]

(Equation 12)

Is as follows. this is,

[0059]

(Equation 13)

Is as follows. However,

[0061]

[Equation 14]

Is as follows. If these two data series d
_{If 1} (i) and d ₂ (i) are random sequences,

[0063]

(Equation 15)

Is obtained. However, the data series is 0
If they are orthogonal to each other in the N-1 section,

[0065]

(Equation 16)

It can be At this time, equation (1)
From 5), q _1,2 (i) = r _1,2 , and it is possible to converge to the Nth symbol.

In the above description, the case of two users has been described, but if the spread is N times, up to N users should be able to communicate originally. Therefore, it is necessary for each user to use different sequences with N orthogonal codes. for that purpose,
An N′-dimensional orthogonal code composed of N ′ ≧ N symbols is used.

Now, a method of introducing means for speeding up the convergence on the transmitting side according to such a principle will be described below.

FIG. 4 shows the structure of the transmitting means according to an embodiment of the present invention. This figure is an example of a transmitting means of a base station in a mobile communication system. First user is data series D
Send 1. At the timing of input from the timing terminal, the multiplexing circuit MPX1 switches its input to the training signal generator TR1. The training signal generator TR1 outputs an orthogonal signal sequence N'symbol for the first user. The output symbol sequence of the multiplexing circuit MPX1 is the spreading circuit S in the same manner as in the normal spread spectrum system.
It is spread by the spreading code at P1. Other users are operating similarly.

In the case of base station transmission, it is easy to make the timings the same, but in order to synchronize the training signals of the mobile stations, which are dispersed in different places, a separate means for synchronization is required. To do. Actually, since the mobile device operates in synchronization with the signal from the base station, the training signal can be transmitted in synchronization by adjusting the timing in accordance with this synchronization.

The above proof is that the propagation of the transmission path is a single path, but the actual propagation of the transmission path is multipath propagation, and the training signal is orthogonal to the signal having the delay time difference. Need to be. Therefore, it is necessary for the multipath propagation path to have a single peak with a sharp autocorrelation characteristic, that is, be orthogonal to its own signals with different delay times.

By introducing the training signal in this way, the convergence process can be sped up. Although the training signal is inserted on the transmission side, the blind type interference canceller does not need to know the contents of the signal sequence and the position of the insertion timing on the reception signal in advance on the reception side.

The conditions of the training signal are as follows.

(I) The synchronized complex symbol sequence of the training signal forms an orthonormal system:

[0075]

[Equation 17]

(Ii) When there is a multipath delay spread, it must be orthogonal to the self or other user's asynchronous complex symbol sequence (asynchronous time difference: Δ ≠ 0):

[0077]

(Equation 18)

However, in principle, it is impossible in principle to completely satisfy the above conditions for the training sequence. Therefore, in the following, a code close to the above condition is used for the training signal, and the characteristics in OMF are obtained by computer simulation for a specific series.

The target system is as follows: -Modulation method: 10 kb / s BPSK-Spreading factor (process gain): 16-Spreading code: Cross-correlation 0.25 or less (selected by computer) -Synchronous detection (・ No diversity) ・ Training signal: 16,32 symbols ・ Number of paths: 1, 2 ・ Number of users: Single path: 16 (OMF), 2 paths: 7
(OMF-RAKE) -Average Eb / N0: 15 dB, 30 dB Starting from the initial value W (0) = T of the coefficient vector, the transient response of the bit error rate (BER) characteristic was measured. As the training signal, Walsh code and orthogonal Gold code were used.

FIG. 5 shows the characteristics of the bit error rate (BER) using a 16-symbol length training signal. Eb / N
Whether 0 is 15 dB or 30 dB, without a training signal, it takes about 80 symbols to obtain an error rate of about 3 × 10 ^-2 . With the training signal (Walsh),
It is about 32 symbols. This is because the error of the correlation matrix converges at high speed due to the orthonormality of the training signal.

FIG. 6 shows static characteristics of two paths (using a training signal of 32 symbol length). As is clear from this figure, the orthogonal Gold code, which has relatively sharp autocorrelation characteristics and excellent orthogonality, exhibits very good convergence characteristics.

[0082]

As described above, the feature of the present invention is that the coefficient of the canceller can be converged at high speed by transmitting the training signal even when there is almost no fading. Furthermore, (1) the training signal to be transmitted does not need to be known on the receiving side, and (2) the synchronization of the training signal is easy because it is within the same circuit board station on the downlink.

Thus, according to the present invention, since the canceller can be brought into the operating state at high speed, the special effect that the transmission path for transmitting signals in real time can be started up at high speed can be obtained. it can.

Further, in the public mobile communication system, it is necessary to utilize the limited finite frequency band as effectively as possible to accommodate as many users as possible, and the interference canceller by orthogonalization is an important technology. From this point of view, the present invention is particularly effective for a system using such an interference canceller.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a blind interference canceller having an orthogonalization filter.

FIG. 2 is a detailed configuration diagram of a signal extraction means EXTR shown in FIG.

FIG. 3 is a block diagram showing a configuration of a non-blind type interference canceller having an orthogonalization filter.

FIG. 4 is a block diagram showing transmitting means according to an embodiment of the present invention.

FIG. 5 is a diagram showing a convergence characteristic of the present embodiment in a single pass.

FIG. 6 is a diagram showing a convergence characteristic of the present embodiment in two passes.

[Explanation of symbols]

 TR1 to TR4 training signal generator MPX1 to MPX4 multiplexing circuit SP1 to SP4 spreading circuit COMB synthesizing circuit

Claims (4)

[Claims] 1. A transmission means for allowing each user to transmit orthogonal signals which are orthogonal to each other as a transmission symbol sequence in synchronization with each other, and a reception means including an orthogonalization filter for removing cross-correlation by adaptive signal processing. A spread spectrum communication device characterized by being provided. 2. A transmission means for synchronously transmitting a signal having a sharp autocorrelation peak as a transmission symbol sequence to each user, and a reception means including an orthogonalization filter for removing cross-correlation by adaptive signal processing. A spread spectrum communication device characterized by the above. 3. On the transmitting side, each user synchronously transmits orthogonal signals that are orthogonal to each other as a transmission symbol sequence, and on the receiving side, an orthogonalization filter that removes cross-correlation by adaptive signal processing is used. A spread spectrum communication method characterized by being used. 4. On the transmitting side, each user synchronously transmits a signal having a sharp autocorrelation peak as a transmission symbol sequence, and on each receiving side, an orthogonalization filter for removing cross-correlation by adaptive signal processing is used. A spread spectrum communication method characterized by the above.