KR20140139896A - Method and apparatus for estimating a transmitted signal in a wireless communication system - Google Patents

Abstract

A method for detecting data received in a wireless communication system includes receiving a plurality of received signal sequences transmitted over a transmission channel, obtaining a channel correlation matrix from the received signal sequence, And estimating a transmission signal sequence prior to being transmitted through the transmission channel from the channel correlation matrix, wherein the channel correlation matrix comprises a first channel correlation matrix and a second channel correlation matrix, And the channel correlation matrix provides a fixed coefficient filtering method for least squares approximation, thereby enabling reduction of hardware overhead.

Description

The present invention relates to a method and apparatus for estimating a transmission signal in a wireless communication system,

The present invention relates to a method and an apparatus for estimating a transmission signal in a wireless communication system, and more particularly, to a method and system for estimating a transmission signal in a wireless communication system by using a Cholesky approximation in an inverse transformation of a block Toeplitz matrix inversion), thereby estimating the transmission signal.

In wireless communication systems, inter symbol interference (ISI) due to multipath channels degrades data detection performance. The Minimum Mean Square Error (MMSE) is a well known method for wireless communication systems and many wireless communication systems employ it as an equalization method. However, MMSE requires a complex inverse operation procedure with a large computational complexity of O (N ³ ). A triangular solution associated with a given square matrix transformation on a triangular matrix can reduce complexity to O (N ² ) by a simple filtering procedure without any performance degradation, and this is generally a linear system solution It is called.

Methods for solving the banded teplitz linear system can be obtained by matrix decomposition of a triangular matrix. The LU decomposition and QR method and the Schur algorithm are known to have small computational complexity and high performance accuracy. However, the triangular matrix has time-varying coefficients for each symbol period, and therefore the existing linear system solutions include a variable digital filtering procedure. This requires a coefficient update algorithm, additional memory and access time for the hardware implementation, which hinders high-speed filter processing.

Figure 1 shows a prior art communication system model for estimating a transmitted signal from a received signal. The communication system model 100 includes a transmission channel 110 and a reception apparatus 120. [ The receiver 120 includes a receive filter 130 and a signal estimator 140. The received signal sequence 111 of the communication system 100 may be represented by a linear model as shown in equation (1) below by the transmitted signal sequence 101, the added noise sequence 102, and the transmitted channel matrix HN .

,

및

는 각각 상기 송신 신호 시퀀스(101), 상기 수신 신호 시퀀스(111) 및 상기 부가 잡음 시퀀스(102)이다. 상기 송신 채널(110)이 시불변(time-invariant) 이고, 상기 송신 시퀀스(101)의 길이 N 이 상기 송신 채널(110)의 채널 특성에 해당하는 임펄스 응답의 길이 L보다 크다는 가정하에, 길이(N+L-1)×N의 상기 송신 채널 행렬 HN은 아래의 식 (2)와 같이 실베스터(Sylvester) 구조의 테플리츠(Toeplitz) 행렬에 의해 표현된다.here,

,

And

Are the transmission signal sequence 101, the reception signal sequence 111 and the additional noise sequence 102, respectively. It is assumed that the transmission channel 110 is time-invariant and the length N of the transmission sequence 101 is greater than the length L of the impulse response corresponding to the channel characteristic of the transmission channel 110, The transmission channel matrix HN of N + L-1) × N is represented by a Toeplitz matrix of a Sylvester structure as shown in the following equation (2).

에 의해 나타내어지고, 여기서

는 복소 공액 행렬 전치를 나타낸다. LS 해는 정합 필터 출력(131)

를 계산한 이후에, 아래의 식 (3)에 의해 획득될 수 있다. The estimated signal sequence 141 by Least Square (LS) estimation

Lt; RTI ID = 0.0 >

Represents the complex conjugate matrix transpose. The LS solution includes a matched filter output 131,

Can be obtained by the following equation (3).

, (3)

, (3)

는 Gram 행렬이라 불리우며, 상기 Gram 행렬은 아래의 식(4)와 같이 표현된다.here,

Is called a Gram matrix, and the Gram matrix is expressed by the following equation (4).

The Gram matrix is banded and teplitz, and the size is N × N. The element of the Gram matrix corresponding to the channel correlation coefficient is calculated by the following equation (5).

을 만족한다.By a Cholesky factor method, a lower triangular matrix is generated as shown in Equation (6) below,

.

The solution of the two banded triangles is represented by the following equation (7).

The solution of the equation (7) can be obtained by the forward substitution and the backward substitution of the following equations (8) and (9).

For i> N-1 or j <0, the initial values are defined by q _{i, j} = 0, x (i) = 0, and v (j) = 0. By this substitution method, the coefficients q _{i, j} must be changed for each iteration, which results in a more complex hardware structure. Thus, an assignment method with a fixed coefficient, which can be implemented by linear filters, is required.

The decorrelator of the Teflitz matrix can be performed by the forward substitution method and the back substitution method after obtaining the cholesky decomposition of the correlation matrix and the forward substitution method and the back substitution method can be performed by the variable coefficient filtering method . However, in the variable coefficient filtering method, the filter coefficients need to be changed every symbol period, so that the hardware overhead increases. Therefore, in the present invention, a fixed coefficient filtering method is required in order to reduce the hardware overhead due to the change of the filter coefficient.

According to another aspect of the present invention, there is provided a method of estimating a transmission signal in a wireless communication system, the method comprising: receiving a plurality of received signal sequences transmitted through a transmission channel; And estimating a transmission signal sequence prior to being transmitted through the transmission channel from the received signal sequence and the channel correlation matrix. Here, the channel correlation matrix is least squares decomposed into a first channel correlation matrix and a second channel correlation matrix.

The first channel correlation matrix and the second channel correlation matrix are respectively a Toeplitz lower triangular matrix and a Toeplitz upper triangular matrix

The receiving of the received signal sequence may include: applying a first received signal sequence comprised of L-1 zeros; And receiving N second received signal sequences.

The step of estimating the transmission signal sequence is a step of multiplying each of the previous L-1 reception signals in the reception signal sequence by a filter coefficient (filter coefficient).

The step of estimating the transmission signal sequence is a step of multiplying each of the previous L-1 reception signals in the reception signal sequence by a filter coefficient (filter coefficient).

Wherein the channel correlation matrix is a 2x2 sub-matrix composed of a first matrix to a fourth matrix, the first matrix is a Teflitz sub-matrix of (L-1) x (L-1) N × (L-1) -type triplet matrix.

The value of each element of the first and second matrix is the filter coefficient.

The channel correlation matrix is determined from a correlation matrix and an autocorrelation sequence of the received signal sequence.

The step of receiving the received signal sequence further comprises the step of inputting the received signal sequence to a matched filter having an impulse response matched to the waveform of the received signal sequence.

The step of estimating the transmission signal sequence is a step of estimating the transmission signal sequence by an infinite impulse response (IIR) filter.

The filter coefficient has a fixed value while receiving the received signal sequence.

In order to achieve the above object, an apparatus for estimating a transmission signal in a wireless communication system according to the present invention includes: a reception filter for receiving a plurality of reception signal sequences transmitted through a transmission channel; And a signal estimator for obtaining a channel correlation matrix from the signal sequence and estimating a transmission signal sequence before being transmitted through the transmission channel from the received signal sequence and the channel correlation matrix. Here, the channel correlation matrix is a first channel correlation matrix and a second channel correlation matrix, and is a least square decomposition.

The first channel correlation matrix and the second channel correlation matrix are respectively a Toeplitz lower triangular matrix and a Toeplitz upper triangular matrix.

The receive filter forces a first received signal sequence comprised of L-1 zeros and receives N second received signal sequences.

The signal estimator is an equalizer that multiplies the previous L-1 received signals of the received signal sequence by a filter coefficient.

Wherein the channel correlation matrix is a 2x2 sub-matrix composed of a first matrix to a fourth matrix, the first matrix is a Teflitz sub-matrix of (L-1) x (L-1) N × (L-1) -type triplet matrix.

The value of each element of the first and second matrix is the filter coefficient.

The channel correlation matrix is determined from the correlation matrix and the autocorrelation sequence of the received signal sequence.

The receive filter is a matched filter having an impulse response matched to the waveform of the received signal sequence.

The signal estimator estimates the transmission signal sequence by an infinite impulse response (IIR) filter.

The filter coefficient has a fixed value while receiving the received signal sequence.

In order to achieve the above object, an apparatus for estimating a transmission signal in a wireless communication system according to the present invention is a user equipment (UE).

In order to achieve the above object, an apparatus for estimating a transmission signal in a wireless communication system according to the present invention is a base station (BS).

According to an aspect of the present invention, there is provided a method for estimating a transmission signal in a wireless communication system from a plurality of received signal sequences transmitted through a transmission channel according to the present invention, the method comprising: (L-1) first received signal sequences, and multiplying the stored first received signal sequence by a filter coefficient to estimate the transmitted signal. Here, the filter coefficient has a fixed value while the plurality of received signal sequences are received.

The filter coefficients are updated after the plurality of received signal sequences are received.

  The plurality of received signal sequences are subjected to L-1 zeros before the transmission signal is transmitted through the transmission channel.

The estimated transmission signal is calculated by multiplying the inverse transformation of the channel correlation matrix by the plurality of received signal sequences.

The channel correlation matrix is least squares decomposed into a first channel correlation matrix and a second channel correlation matrix.

The first channel correlation matrix and the second channel correlation matrix are respectively a Toeplitz lower triangular matrix and a Toeplitz upper triangular matrix.

Wherein the channel correlation matrix is a 2x2 sub-matrix composed of a first matrix to a fourth matrix, the first matrix is a Teflitz sub-matrix of (L-1) x (L-1) N × (L-1) -type triplet matrix.

The value of each element of the first and second matrix is the filter coefficient.

The channel correlation matrix is determined from a correlation matrix and an autocorrelation sequence of the received signal sequence.

The transfer function of the channel correlation matrix is expressed as a product of a transfer function A (z) of the first channel correlation matrix and a transfer function B (z) of the second channel correlation matrix.

In order to achieve the above object, an apparatus for estimating a transmission signal in a wireless communication system according to the present invention is a user equipment (UE).

In order to achieve the above object, an apparatus for estimating a transmission signal in a wireless communication system according to the present invention is a base station (BS).

In order to achieve the above object, an apparatus for estimating a transmission signal in a wireless communication system from a plurality of reception signal sequences transmitted through a transmission channel according to the present invention and matched and filtered includes: A memory for storing (L-1) first received signal sequences, and a processor for multiplying each of the stored first received signal sequences with a filter coefficient, and then adding the resultant signals to estimate the transmitted signals. The filter coefficient has a fixed value while the plurality of received signal sequences are received.

After the plurality of received signal sequences are received, the processor updates the filter coefficients.

The plurality of received signal sequences are subjected to L-1 zeros before the transmission signal is transmitted through the transmission channel.

The processor estimates the transmission signal by multiplying the inverse transformation of the channel correlation matrix by the plurality of received signal sequences.

The channel correlation matrix is least squares decomposed into a first channel correlation matrix and a second channel correlation matrix.

The first channel correlation matrix and the second channel correlation matrix are respectively a Toeplitz lower triangular matrix and a Toeplitz upper triangular matrix.

Wherein the channel correlation matrix is a 2x2 sub-matrix composed of a first matrix to a fourth matrix, the first matrix is a Teflitz sub-matrix of (L-1) x (L-1) N × (L-1) -type triplet matrix.

The value of each element of the first and second matrix is the filter coefficient.

The channel correlation matrix is determined from a correlation matrix and an autocorrelation sequence of the received signal sequence.

The transfer function of the channel correlation matrix is expressed as a product of a transfer function A (z) of the first channel correlation matrix and a transfer function B (z) of the second channel correlation matrix.

In order to achieve the above object, an apparatus for estimating a transmission signal in a wireless communication system according to the present invention is a user equipment (UE).

In order to achieve the above object, an apparatus for estimating a transmission signal in a wireless communication system according to the present invention is a base station (BS).

When the packet size N is sufficiently larger than the channel length L (N > L), the fixed coefficient filtering method according to the present invention has only a very small increase in complexity and a degradation in performance that can be ignored, And a data estimating method and data estimating apparatus that can be implemented by a small memory size.

1 shows a prior art communication for estimating a transmission signal from a received signal
System model.
2 shows a communication system model according to the present invention for estimating a transmission signal from a reception signal.
3 is a detailed configuration diagram of a signal estimator corresponding to the signal estimator of FIG.
4 is a graph showing output signal-to-noise ratio (SNR) degradation according to a channel length L according to the present invention for a plurality of packet sizes (N = 32, 64, 128).
5 is a graph showing the complexity according to the channel length L according to the present invention for a plurality of packet sizes (N = 64, 128).
6 is a graph showing the complexity according to the channel length L for a plurality of packet sizes (N = 64, 128) with and without zero applied.
FIG. 7 is a graph illustrating a bit error ratio (BER) of a GSM system according to a bit energy-to-noise power density ratio Eb / N0 for a case where a zero is applied and a case where a zero is applied. Quadrature Phase Shift Keying) and 16-QAM (16-Quadrature Amplitude Modulation).

2 shows a communication system model according to the present invention for estimating a transmission signal from a reception signal. The communication system model 200 includes a transmission channel 210 and a reception device 220. The receiver 220 includes a receive filter 230 and a signal estimator 240. Here, a fixed coefficient filtering method is proposed to obtain approximate solutions for the forward substitution method and the backward substitution method of equation (8). The received signal sequence 211 is an extended signal obtained by inserting (L-1) zeros in front of the received signal sequence 111 in FIG. 1 and expressed by the following equation (10). The received signal sequence 211 of the communication system model 200 is determined by the transmit signal sequence 201, the additive noise sequence 202 and the transmit channel 210 H _{N + L-1}

은 길이 (L-1)의 제로 열 길이를 나타낸다. 이제, 확장된 채널 행렬의 크기는 (N+2L-2)×(N+L-1)이 되고, 이는 아래의 2×2 블록 행렬 표현에 의해 표현될 수 있다.here,

Represents the zero column length of the length (L-1). Now, the size of the extended channel matrix is (N + 2L-2) x (N + L-1), which can be represented by the following 2x2 block matrix representation.

here,

The sizes of the sub-matrices C and D are (L-1) L-1 and N + L-1 L-1, respectively, Is displayed.

은 확장에 따른 추가 출력이고,

은 도 1에서 제시된 수신 필터(130)의 출력 값(131)이다. 여기서, 도 1 및 도 2의 수신 필터(130, 230)는 정합 필터(matched filter)로 구현될 수 있다. 상기 정합 필터는 상기 송신 신호 시퀀스(101, 201) 및 상기 수신 신호 시퀀스(111, 211)의 신호 파형(s(t))과 정합되는 임펄스 응답(h^*(t))을 가진다. 즉, 상기 정합 필터의 임펄스 응답 h^*(t)=k S(T-t)로 표현되며, 여기서 T는 상기 송신 시퀀스(101, 201) 및 수신 신호 시퀀스(111, 211)의 각 신호의 주기를 나타낸다.here,

Is an additional output due to expansion,

Is the output value 131 of the receive filter 130 shown in FIG. Here, the reception filters 130 and 230 of FIGS. 1 and 2 may be implemented as a matched filter. The matched filter has an impulse response h ^* (t) matched with the signal waveform s (t) of the transmission signal sequence 101, 201 and the reception signal sequence 111, 211. That is, the impulse response h ^* (t) = ks (Tt) of the matched filter is represented by T, where T denotes a period of each signal of the transmission sequence 101 and 201 and the received signal sequence 111 and 211 .

3 is a detailed configuration diagram of the signal estimating apparatus 300 corresponding to the signal estimator 240 of FIG. The signal estimator 300 estimates a transmission signal 241 in the wireless communication system from a plurality of reception signal sequences 231 transmitted through the transmission channel 210 of FIG. 2 and output by the matched filter 230 . The signal estimation apparatus 300 includes a memory 310 and a processor 320. [ The memory 310 stores the (L-1) first received signal sequences by shifting the plurality of received signal sequences by one bit. The processor 320 includes an arithmetic unit 321 and a control unit 322. The arithmetic unit 321 multiplies the product and then adds the products of (L-1). The filter coefficient has a fixed value while the plurality of received signal sequences are received. The control unit 322 updates the filter coefficient after the plurality of received signal sequences 231 are received. The plurality of received signal sequences are transmitted through the matched filter 230 after L-1 zeros are applied before N transmission signals are transmitted through the transmission channel.

The operation unit 321 estimates the transmission signal 241 by multiplying the inverse transformation of the channel correlation matrix by the plurality of received signal sequences 231. The channel correlation matrix is least squares decomposed into a first channel correlation matrix and a second channel correlation matrix. The first channel correlation matrix and the second channel correlation matrix are respectively a Toeplitz lower triangular matrix and a Toeplitz upper triangular matrix.

According to the proposed new method, the plurality of received signal sequences 231 corresponding to the extended reception filter output are expressed by the following formula (14) by forward substitution and backward substitution .

Here, P _{N + L-1} is a triplet matrix with a triplet and a banded sub-matrix as shown in the following equation (15).

The size of P _{N + L-1} is (N + L-1) × (N + L-1) and matrix elements p _n satisfy equation (16).

To obtain a stable solution for Eq. (14), {p ₀ , p ₁ , ..., p _L-1 } and (R p) are calculated for AR (Auto Regressive) filters performing forward assignment and post- the coefficients p _n satisfying the equation (16) must be the minimum phase because {p * _L-1 , p * _L-2 , ..., p * ₀ } is used.

이며, 계수 p_n은 자기 상관 시퀀스 r_n으로부터 구할 수 있다.The coefficients p _n that satisfy equation (16) relate to the signal estimator 240 performing the forward substitution and reverse substitution methods. The signal estimator 240 compensates for signal distortion due to inter-symbol interference (ISI) by a multipath channel, and thus performs an equalizer function. The coefficients p _n correspond to the coefficients of each stage of the equalizer. The equalizer should be the minimum phase since it can be considered as the numerator coefficients of the IIR filters. The IIR filter is designed to take account of all the influences of all the previous inputs, and is implemented in a structure that feeds back the previous finite outputs for practical implementation thereof. For n = 0, 1, ..., L + 1

, And the coefficient p _n can be obtained from the autocorrelation sequence r _n .

The matched filter output of equation (13) can be obtained by finite impulse response (FIR) filtering of equation (17) below.

Here, y (i) = 0 for i < 0.

,

를 이용하면 식 (14)의 전진 및 후진 대입법은 아래의 차분 방정식에 의해 수행될 수 있다.At this time, LDL ^H decomposition may be applied instead of cholysky decomposition. In this case, instead of P _i

,

, The forward and backward substitution methods of Eq. (14) can be performed by the following differential equations.

The initial value is defined by v (i) = 0 for i <0 and x _new (i) = 0 for i> N-1. Unlike Eqs. (8) and (9), the filter coefficients of Eqs. (18) and (19) are not variable and can be implemented by a linear filter.

After obtaining the matched filter output in equation (13) and the minimum phase filter coefficients p _n in equation (16), the forward substitution and back substitution methods in equation (14) can be performed by IIR filtering with fixed filter coefficients , Which is called a fixed coefficient filtering method for the cholesky-based LS solution. That is, the output of the matched filter is determined by the equation (13), and the coefficient p _n can be obtained from the autocorrelation sequence r _n by the equation (14). The autocorrelation sequence r _n is determined by obtaining a received signal correlation matrix from the received signal sequence 211. Wherein the received signal correlation matrix includes a received signal correlation matrix and an autocorrelation matrix, wherein the received signal correlation matrix and the autocorrelation matrix are E {y _e (n) y _e (nm) ^H } and E {y _e (n) y _e (n) ^H }. Therefore, the coefficient p _n can be obtained using the statistical characteristic of the received signal y _e (n).

Equations (17) and (18) can be implemented by an IIR filter represented by the z-transform of equation (20) below.

Here, the input and output sequences are {u _e (0), u _e (1), ..., u _e (N + L-2)} and {v (0), v , v (N + L-2)}.

On the other hand, the equation (19) can be implemented by an IIR filter expressed by the following equation (21).

The input and output are respectively {v (N + L-2 ), v (N + L-3), ..., v (0)} and _{{x new (N-1)} , x new (N- 2), ..., x _new (0)}.

In the following, it will be appreciated that the proposed method using the triplet and triangular matrix not only reduces the memory size required for calculation, but also has little deterioration in performance compared to the conventional method.

Let's analyze the output noise signal to calculate the approximation error. In equation (15), the triplet matrix P _{N + L-1} can be represented by the following block matrix form.

here,

The sizes of sub-matrices E and F are (L-1) × (L-1) and N × (L-1), respectively. The structure is the same as C and D in equation (12). Equation (14) is solved by the following equations (24) and (25), called block advance and block advance substitution.

And

(25)

(25)

및

로 표현될 수 있으므로, 제안된 방법의 출력 신호는 아래의 식 (26)으로 표현된다.Here, equations (24) and (25)

And

The output signal of the proposed method is expressed by the following equation (26).

here,

And

.

및

이 성립함은 분명하고,

및

이다. 따라서, 추정된 송신 신호 시퀀스(241)는 아래의 식(29)와 같이 표현된다.

And

This establishment is clear,

And

to be. Therefore, the estimated transmission signal sequence 241 is expressed by the following equation (29).

(29)

(29)

Here, since the estimated transmitted signal sequence 241 does not include inter-symbol interference (ISI), it can also be considered as zero-forcing.

The singular value decomposition (SVD) H _N of the (N + L-1) × N matrix is a factorization in the form of the following equation (30).

,

및

에 의해 표현된다.A channel unitary matrix of size (N + L-1) × N, a noise space unit matrix of size (N + L-1) × (L-1)

,

And

Lt; / RTI &gt;

이기 때문에, 식(29)는 아래의 식(30)과 같이 표현된다.here,

, Equation (29) is expressed by the following equation (30).

라고 하면, 제안된 등화기의 출력은 아래의 식 (31)에 의해 표현된다.here,

, The output of the proposed equalizer is expressed by the following equation (31).

은 원래의 LS 등화기의 출력이다. 새로운 방법은 또한 제로-추가이지만 식 (3)의 초기 LS 해법과 비교하여 약간 큰 출력 잡음 전력을 가진다. BH_N이라는 사실을 이용함으로써, 출력 잡음 전력은 아래의 식(32)에 의해 계산될 수 있다.here,

Is the output of the original LS equalizer. The new method is also a zero-add but has slightly higher output noise power compared to the initial LS solution of Eq. (3). By using the fact that BH _N , the output noise power can be calculated by the following equation (32).

은 잡음 분산이고, 우측의 첫번째 및 두번째 term은 각각 초기LS 해 및 제안된 방법의 출력 전력이다.here,

Is the noise variance and the first and second term on the right are the initial LS solution and the output power of the proposed method, respectively.

In the following, the performance evaluation result according to the new method is presented.

A. Complexity Analysis

The computational complexity for obtaining equations (6) and (15) is the same. However, the memory sizes for equations (6) and (15) are N and NL, respectively. Here, NL is a relatively large value as compared with L. The complexity of Eqs. (9) and (19) is always the same. For the proposed method, the size of the matched filter and the size of the forward substitution method are increased from N to (N + L-1), but this can be ignored for a system of N >> L. The matched filter and the forward substitution method can be combined and calculated by the IIR filter A (z) in equation (20), and the proposed method has lower complexity than the conventional method.

B. Output SNR degradation

From equation (32), the output SNR degradation for the initial assignment method can be defined by equation (33) below, which corresponds to the appropriate metric for evaluating the performance of the new method.

For a given data size N, SNR degradation is calculated for various channel lengths L. The channel coefficients are randomly generated complex numbers, which can be considered the worst case. 4 is a graph showing output signal-to-noise ratio (SNR) degradation according to a channel length L according to the present invention for a plurality of packet sizes (N = 32, 64, 128). For N = 32 and 64, the SNR degradation is 0.35 dB and 0.2 dB, respectively. For systems with N greater than about 100, performance degradation is considered acceptable because it is less than about 0.1 dB for worst-case channel conditions.

5 is a graph showing a complexity according to the channel length L according to the present invention for a plurality of packet sizes (N = 64, 128). For example, assuming L = 0.2 N, it can be seen that for N = 128 and 256, the complexity increases by 2% and 8%, respectively.

On the other hand, FIG. 6 is a graph showing the complexity according to the channel length L for a plurality of packet sizes (N = 64, 128) with and without zero applied. That is, assuming that L = 0.2 N, for example, when zero is applied compared to when no zero is applied, it can be seen that the complexity increases by 2% and 8%, respectively, for N = 128 and 256 . Also, according to one embodiment, the new scheme (with zero applied) exhibits a computational complexity increase of 1.07% compared to 3,066 COPS of the conventional scheme (without zero applied) at 3,099 COPS.

The following is a comparison of BER performance for GSM system parameters.

C. BER performance in GSM system

To match the SNR analysis in the previous section, the data detection performance of the proposed method is compared with the existing method. System parameters based on the GSM system are summarized in Table 1.

parameter value The carrier frequency (f _c ) 1 GHz Sampling frequency (f _s ) 1.73 kHz Channel model GSM Hilly terrain Modulation method QPSK, 16-QAM

The BER performances corresponding to the signal to noise ratio (SNR) are shown in FIG. 6, and the channel impulse response is assumed to be known. To simulate performance, a tapped delay line (TDL) channel model of the ITU standard for the GSM system was applied for the radio propagation channel, which is generated by the Jake channel model.

FIG. 7 is a graph illustrating a bit error ratio (BER) of a GSM system according to a bit energy-to-noise power density ratio Eb / N0 for a case where a zero is applied and a case where a zero is applied. Quadrature Phase Shift Keying) and 16-QAM (16-Quadrature Amplitude Modulation). FIG. 7 shows that for QPSK and 16-QAM, the new filtering method has approximately the same performance as the initial assignment with variable coefficients. The SNR degradation calculated by equations (33) and (34) is only 0.05 dB, which is an acceptable range. Therefore, the proposed method enables simpler hardware structure than existing methods without degradation of performance.

Claims (26)

A method for estimating a transmission signal in a wireless communication system,
The method comprising: receiving a plurality of received signal sequences transmitted over a transmission channel;
Obtaining a channel correlation matrix from the received signal sequence; And
Estimating a transmitted signal sequence prior to being transmitted over the transmission channel from the received signal sequence and the channel correlation matrix,
Wherein the channel correlation matrix is least squares decomposed into a first channel correlation matrix and a second channel correlation matrix. The method according to claim 1,
Wherein the first channel correlation matrix and the second channel correlation matrix each comprise a first channel correlation matrix and a second channel correlation matrix, each of which is a Toeplitz lower triangular matrix and a Toeplitz upper triangular matrix, Way. The method according to claim 1,
Wherein the step of receiving the received signal sequence comprises:
Applying a first received signal sequence comprised of L-1 zeros; And
And receiving N second received signal sequences. &Lt; RTI ID = 0.0 &gt; A &lt; / RTI &gt; An apparatus for estimating a transmission signal in a wireless communication system,
A reception filter for receiving a plurality of received signal sequences transmitted via a transmission channel; And
And a signal estimator for obtaining a channel correlation matrix from the received signal sequence received from the reception filter and estimating a transmission signal sequence before being transmitted through the transmission channel from the received signal sequence and the channel correlation matrix,
Wherein the channel correlation matrix is least squares decomposed into a first channel correlation matrix and a second channel correlation matrix. 5. The method of claim 4,
Wherein the first channel correlation matrix and the second channel correlation matrix each comprise a first channel correlation matrix and a second channel correlation matrix, each of which is a Toeplitz lower triangular matrix and a Toeplitz upper triangular matrix, Device. 5. The method of claim 4,
Wherein the receive filter is configured to force a first received signal sequence comprised of L-1 zeros and receive N second received signal sequences. The method according to claim 6,
Wherein the signal estimator is an equalizer that multiplies each of the previous L-1 received signals of the received signal sequence by a filter coefficient. 8. The method of claim 7,
Wherein the channel correlation matrix is a 2x2 sub-matrix composed of a first matrix to a fourth matrix, the first matrix is a Teflitz sub-matrix of (L-1) x (L-1) N is a triplet matrix of N x (L-1). 9. The method of claim 8,
Wherein a value of each element of the first and second matrix is the filter coefficient. 5. The method of claim 4,
Wherein the channel correlation matrix is determined from a correlation matrix and an autocorrelation sequence of the received signal sequence. 5. The method of claim 4,
Wherein the receive filter is a matched filter having an impulse response matched to the waveform of the received signal sequence. 5. The method of claim 4,
Wherein the signal estimator estimates the transmitted signal sequence by an infinite impulse response (IIR) filter. 8. The method of claim 7,
Wherein the filter coefficient has a fixed value while receiving the received signal sequence. A method for estimating a transmission signal in a wireless communication system from a plurality of received signal sequences transmitted through a transmission channel and matched to each other,
(L-1) first received signal sequences by shifting the plurality of received signal sequences by one bit; And
Estimating the transmission signal by multiplying each of the stored first received signal sequences by a filter coefficient and then adding the result;
Wherein the filter coefficient has a fixed value while the plurality of received signal sequences are received. 15. The method of claim 14,
Wherein the filter coefficients are updated after the plurality of received signal sequences are received. 15. The method of claim 14,
Wherein the plurality of received signal sequences are subjected to L-1 zeros prior to transmission of the transmission signal through the transmission channel. An apparatus for estimating a transmission signal in a wireless communication system from a plurality of received signal sequences transmitted through a transmission channel and output by a matched filter,
A memory for storing (L-1) first received signal sequences by shifting the plurality of received signal sequences by one bit; And
And a processor for multiplying each of the stored first received signal sequences by a filter coefficient and adding the multiplied filter coefficients to estimate the transmitted signal,
Wherein the filter coefficient has a fixed value while the plurality of received signal sequences are received. 18. The method of claim 17,
Wherein the processor updates the filter coefficient after the plurality of received signal sequences are received. 18. The method of claim 17,
Wherein the plurality of received signal sequences are L-1 zeros prior to transmission of the transmission signal through the transmission channel. 18. The method of claim 17,
Wherein the processor estimates the transmitted signal by multiplying the inverse of the channel correlation matrix by the plurality of received signal sequences. 18. The method of claim 17,
Wherein the channel correlation matrix is least squares decomposed into a first channel correlation matrix and a second channel correlation matrix. 22. The method of claim 21,
Wherein the first channel correlation matrix and the second channel correlation matrix each comprise a first channel correlation matrix and a second channel correlation matrix, each of which is a Toeplitz lower triangular matrix and a Toeplitz upper triangular matrix, Device. 21. The method of claim 20,
Wherein the channel correlation matrix is a 2x2 sub-matrix composed of a first matrix to a fourth matrix, the first matrix is a Teflitz sub-matrix of (L-1) x (L-1) N is a triplet matrix of N x (L-1). 24. The method of claim 23,
Wherein a value of each element of the first and second matrix is the filter coefficient. 21. The method of claim 20,
Wherein the channel correlation matrix is determined from a correlation matrix and an autocorrelation sequence of the received signal sequence. 22. The method of claim 21,
Wherein the transfer function of the channel correlation matrix is represented by a product of a transfer function (A (z)) of the first channel correlation matrix and a transfer function (B (z)) of the second channel correlation matrix. / RTI &gt;

Images