KR100425373B1 - Least Squares Frequency Estimation, Transmitter Diversity System and 3GPP System - Google Patents

Abstract

The present invention relates to an LS carrier frequency error estimation method, an estimator, and a transceiver of a transmit diversity system and a 3 PF system using the same. More particularly, the least square method is used when estimating a carrier frequency error in a frequency selective fading channel. The present invention relates to an estimator based on the proposed scheme and that does not require any information about the channel in estimating the frequency error, and to a transceiver of a transmission diversity system and a 3GPP CPICH system.

The estimation method of the present invention calculates a delayed differential decoded received signal based on a least squares (LS) error criterion for recovering a carrier frequency when a waveform formed at a transmitter is received at a receiver through a Rayleigh fading channel, a multiplier, and an adder. Characterized by estimating frequency error, the estimator of the present invention is proposed in a baseband system consisting of a Rayleigh fading channel, a multiplier, and an adder.

Description

LSC carrier frequency error estimation method and estimator and transmitter / receiver of 3PPP system using the same {Least Squares Frequency Estimation, Transmitter Diversity System and 3GPP System}

The present invention relates to a LS carrier frequency error estimation method, an estimator, and a transceiver of a transmit diversity system and a 3PPU system using the same, and more particularly, a least square method when estimating a carrier frequency error in a frequency selective fading channel. This paper proposes an estimator based on the proposed scheme and does not require any information about the channel in estimating the frequency error, and it relates to a transceiver of a transmission diversity system and a 3GPP system.

Modern communication systems tend to impose strict requirements on the frequency error that exists between the oscillator of the transmitter and receiver.

For example, in the 3rd Generation Partnership Project (3GPP) standard for 3G wireless communication systems, a difference between a carrier frequency of a terminal and a carrier frequency of a base station is required to be within 0.1 ppm.

A method that can alleviate this requirement is carrier frequency recovery by signal processing at the receiving end.

Various schemes have been proposed for this carrier frequency recovery.

(1) U. Mengali and AN D'Andrea, Synchronization techniques for digital receivers, Plenum Press, New York, 1997.

(2) H. Meyr, M. Moeneclaey, and SA Fechtel, Digital communication receivers, John Wiley Sons, New York, 1998.

(3) SA Tretter, “Estimating the frequency of a noisy sinusoid by linear regression,” IEEE Trans. Inform. Theory , vol. IT-31, pp. 832-835, Nov. 1985.

(4) S. Kay, “A fast and accurate single frequency estimator,” IEEE Trans. Acoust., Speech, Signal Processing , vol. 37, pp. 1987-1990, Dec. 1989.

(5) MP Fitz and WC Lindsey, “Decision-directed burst-mode carrier synchronization techniques,” IEEE Trans. Commun ., Vol. 40, pp. 1644-1653, Oct. 1992.

(6) MP Fitz, “Planar filtered techniques for burst mode carrier synchronization,” in Proc. IEEE Global Telecommun. Conf., Pp. 365-369, 1991, Orlando, FL.

(7) MP Fitz, “Further results in the fast estimation of a single frequency,” IEEE Trans. Commun ., Vol. 42, pp. 862-864, Feb./Mar./Apr. 1994.

(8) M. Luise and R. Reggiannini, “Carrier frequency recovery in all-digital modems for burst-mode transmissions,” IEEE Trans. Commun. vol. 43, pp. 1169-1178, Feb./Mar./Apr. 1995.

(9) U. Mengali and M. Morelli “Data-aided frequency estimation for burst digital transmission,” IEEE Trans. Commun. , vol. 45, pp. 23-25, Jan. 1997.

(10) WY Kuo and MP Fitz, “Frequency offset compensation of pilot symbol assisted modulation in frequency flat fading,” IEEE Trans. Commun. , vol. 45, pp. 1412-1417, Nov. 1997.

(11) M. Morelli, U. Mengali and GM Vitetta, “Further results in carrier frequency estimation for transmissions over flat fading channels,” IEEE Commun. Letters, vol. 2, pp. 327-330, Dec. 1998.

(12) MG Hebley and DP Taylor, “The effect of diversity on a burst-mode carrier-frequency estimator in the frequency-selective multipath channels,” IEEE Trans. Commun. , vol. 46, pp. 553-560, Apr. 1998.

(13) M. Morelli and U. Mengali, “Carrier-frequency estimation for transmissions over selective channels,” IEEE Trans. Commun. , vol. 48, pp. 1580-1589, Sept. 2000.

Among them, data aided methods (3) to (13) using a training sequence are widely used because they show good performance even with a short training signal.

This is mostly done with channel additive white Gaussian noise (3) to (9) or flat fading (10) or (11), or frequency-selective fading. In the case of fading (12) and (13), the carrier frequency error is estimated.

In the case of a flat fading or frequency selective fading channel, it is difficult to use channel information in frequency estimation since the channel is usually estimated after carrier frequency recovery.

When the channel is flat fading, the estimators at (10) and (11) require the Doppler bandwidth of the channel and thus assume (10) or use the estimate (11).

The estimator at (12) needs the statistical characteristics of the frequency selective channel.

Therefore, the carrier frequency error estimators proposed in (10) to (12) deteriorate when the channel information is not accurate.

(13) proposes a frequency estimator on a frequency selective channel based on maximum likelihood.

This estimator does not require channel information and shows good performance on a fixed channel.

However, when the channel changes over time, there is a decrease in performance, and the implementation of this method requires a large amount of computation.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and proposes a new method for carrier frequency recovery based on the least squares (LS) error criterion, so that no information about the channel is required and that a time-varying channel is needed. The purpose of the present invention is to provide a method for estimating the LSC carrier frequency error, an estimator, a transmit diversity system using the same, and a transceiver of a 3PPP system having good performance.

In order to achieve the above object, the present invention calculates a delayed differential decoded received signal based on a least square error criterion for carrier frequency recovery when a waveform made at a transmitter is received at a receiver through a Rayleigh fading channel, a multiplier and an adder. It is intended to provide a method for estimating the LS carrier frequency error, characterized in that for estimating the frequency error.

In order to achieve the above object, in the present invention, d (j) is a transmission M-ary PSK or QAM symbol, h (t) is a baseband waveform shaping function, n (t) is AWGN, and θ is an initial arbitrary phase. If the,

Is input to the Rayleigh fading channel and the output of the Rayleigh fading channel A multiplier for multiplying and outputting a multiplier and an adder for adding and outputting n (t) to the output of the multiplier are configured in a baseband system.

In order to achieve the above object, the present invention, a space-time coder for generating a symbol to be transmitted through a plurality of transmit antennas, a plurality of multipliers for multiplying a channel response to a symbol generated by the space-time coder, multiplied by the multiplier An adder that adds all the added values to the output of the adder. A multiplier to multiply by the output of the multiplier It is intended to provide a transmit diversity system, characterized in that it consists of an adder that adds.

In order to achieve the above object, the present invention provides a CPICH symbol generator for generating a pilot signal of a common pilot channel to be transmitted through two transmit antennas, a multiplier for multiplying the pilot signal with the same channelization code and a scrambling code, 3GPP comprising a multipath channel for transmitting a signal output through a multiplier, a multiplier and an adder through which the transmitted signal is passed, and a RAKE receiver comprising a plurality of fingers receiving a signal passing through the multiplier and the adder To provide a transceiver for the system.

1 is a block diagram of a baseband system according to the present invention.

2A and 2B are graphs of performance comparisons between the present invention and prior art in frequency selective fading channels.

3 is a performance comparison graph of the present invention and the prior art when the Doppler frequency varies from 10 ⁻³ to 10 ⁻² .

4 is a complexity comparison graph of the present invention and the prior art.

5 is a baseband configuration diagram of a transmit diversity system using an estimator according to the present invention.

6 is a system configuration diagram for 3GPP CPICH using an estimator according to the present invention.

FIG. 7 is a finger diagram illustrating a RAKE receiver in FIG. 6.

8 is a symbol type diagram for a common pilot channel of 3GPP.

FIG. 9 is a performance comparison graph in a transmission diversity system, in which FIG. 9A illustrates a speed of a moving body of 3 km / h and FIG. 9B illustrates a speed of a moving body of 120 km / h.

<Description of the symbols for the main parts of the drawings>

10: Rayleigh fading channels 20a, 20b,... Multiplier

30a, 30b,... Adder 40: Receive Filter

50: space-time encoder 60: CPICH symbol generator

70a, 70b,... : Channelization codes 80a, 80b,... : Scrambling Code

90: RAKE receiver 100a, 100b,... Finger

110a, 110b: multipath channel 120: m-order adder

Hereinafter, the present invention will be described with reference to the accompanying drawings.

The baseband system model covered by the present invention is shown in FIG.

Where d (j) represents the transmit M-ary Phase-Shift Keying (PSK) or Quadrature Amplitude Modulation (QAM) symbol, h (t) is the baseband waveform shaping function, and n (t) is the additive white Gaussian noise ( AWGN), θ is the initial random phase and Denotes a carrier frequency error.

This model Is input to and output from the Rayleigh fading channel 10 and the Rayleigh fading channel 10 output. Multiplier 20a for multiplying and outputting, adder 30a for adding and outputting n (t) to the output of the multiplier 20, and reception filter 40 for removing noise included in the output of the adder 30a. It consists of

In this model all waveform shaping is done at the transmit end, and the receive filter 40 is a simple noise canceling filter.

In this case, the output of the reception filter 40 at t = kT is expressed by Equation 1 below.

In Equation 1, g _k (l) represents the response by the impulse at the time k to l hours ago, and represents an equivalent channel impulse response in the baseband.

G _k (l) represents both h (t) and Rayleigh fading channel 10 of FIG. 1 in the discrete time domain, and the length is L + 1.

η (k) is the variance Assume that is white normal noise.

Training signal {d (k) | k = 1,... , K} and g _k (l) is a fixed channel during the training signal interval, i.e. g _k (l) = g (l), in the present invention, the training signal and the received signal {r (k ) | K = 1,… , K} given channel {g (l) | l = 1,... Without knowing Estimate

From the received signal {r (k)} It is difficult to estimate directly.

because The problem of estimating is This is because it is related to estimating the time-varying phase of.

Thus m-lag differentially decoded signal to shift to a constant phase In other words, Is calculated as in Equation 2.

Where n _m (k) is Represents all noise terms associated with, Denotes a fixed phase shift.

And d _m (k) and g are both (L + 1) ^two- dimensional vectors such as Equations 3 and 4.

In Equation 3, {d (k)} is a training signal.

Can only be represented by these training signals K is in the range of (5).

Having a range of Equation 5 May be represented by a determinant such as Equation 6.

In Equation 6,

,

D _m is Matrix of dimensions,

Is defined.

The carrier frequency error estimator is estimated using Equation 6, and is derived in two steps.

First, the LS estimate of p _m , In the next step To Estimate from

Is selected to p _m which minimizes the sum of error squares, such as

if If this is nonsingular (or if the column vectors of the D _m matrix are linearly independent of each other), an LS estimate that minimizes J ( p _m ) Is obtained as shown in Equation (8).

This solution holds only in the case of determined or overdetermined.

That is, the row order of D _m must be greater than or equal to the column order.

The range of m that satisfies the above condition is expressed by (9).

To get An appropriate training signal is needed to make the non-specific matrix.

next Is It can be estimated by looking at the elements of.

Lay, Because and With g and If it is an estimate of, It can be expressed as

From Equation 4, Looking at the first element of the equation (10) becomes:

As in Equation 10 Some of the elements of represent only the absolute value of the channel. It is the key to derive the estimate of.

In other words, Is Equal to the angle of Is obtained as in Equation (11).

Looking at the second element of is represented by equation (12).

Since channel information is not known in Equation 12, Cannot be estimated from this equation.

By the above equation (12) To estimate If you select the elements of, there are L + 1 elements as shown in equation (13).

here Has a range of.

And Is an estimate Is expressed by Equation 14.

Equation 14 is for a given m value Means an estimate of, Therefore, the estimated range is expressed by Equation 15.

For all possible m values in the same manner as If we estimate and take the average of these values, we get an estimate like Eq.

Where N is the maximum value of m.

Equation 16 is hereinafter referred to as an LS estimator (hereinafter referred to as 'LSE').

Maximum value of frequency error in real system Since is known, the upper bound value of the N value from equation (15) is given by equation (17).

Further, by the given value of N, the lower bound of the training signal length K value is derived from equation (9) to equation (18).

Therefore K must increase in proportion to L ² .

In Equation 8, Since only needs {d (k)} and {r (k)}, the LSE can estimate the frequency error without the information of the channel coefficients.

As mentioned above, the implementation of the estimation method is quite simple.

Because in equation (8) This can be calculated in advance by the given training signal.

In addition, the frequency error can be estimated more accurately by increasing the value of N or the non-hyung signal period.

However, increasing the value of N reduces the frequency estimation range, and increasing the value of K increases the transmission overhead, so these values cannot be made as large as desired.

As mentioned above, LSE Requires training signal to ensure non-specificity

Several training signals are given below and their specificity is observed.

Table 1 shows the midamble (14) of the GSM system, the preamble (15) of the IS-136 system, and the Barker code of length 11.

(14) GSM Recommendation 05.05, Radio Transmission and Reception, V.8.1, ETSI / PT, 1999.

(15) Electronic Industries Association, EIA / TIA IS-136: dual-mode mobile station-base station compatibility standard, TIA, Washington DC, 1994.

Corresponding to each training signal The specificity of was observed under the assumption that L = 1 and the results are shown in Table 2.

As shown, in the case of the preamble of the IS-136 system, Although not specific, the GSM midamble or Barker code is unusual for some m values.

In the following computer simulation, when the channel is a Rayleigh fading channel, the estimator L12 according to the present invention is compared with the estimators 12 and 13 described in the prior art.

To facilitate the comparison, use the simulation parameters used in (13).

In particular, the models for the waveform shaping filter, the training signal and the transmission medium are identical.

Assume a binary-PSK signal and use a raised cosine filter with a rolloff factor of 0.5 for waveform shaping.

The training signal is a hexadecimal representation of K = 32 which is 5230F641.

Also frequency error It is suggested as a range.

The time-invariant channel in (13) is expanded to obtain the channel response g _k (l) of equation (1).

If we model the transmission medium as a typical urban channel of a six-path GSM system, g _k (l) is given by:

From here, Wow Are the attenuations and delays of the path, respectively, and t ₀ is the timing phase. Becomes the value of.

L + 1 channel value To be the channel response of.

Standardized delay Has a value of {0,0.054,0.135,0.432,0.62,1.351}.

For a given i Is a complex Gaussian random process with an average of zero, and its Power Spectral Density (PSD) is Is band limited to (f _D : maximum Doppler frequency).

For different paths, Is statistically independent and the variance has a value of {-3,0, -2, -6, -9,10} (decibels).

After obtaining the time-varying channel g _k (l), we implement the estimator as follows.

First, in case of LSE according to the present invention, Equation 16 is calculated and The range is determined.

K = 32, Since Equation 18 It becomes the range of.

Therefore, the estimator LSE according to the present invention assumes the channel length to be 4 or smaller.

In the case of the autocorrelation-based estimator (hereinafter referred to as 'ABE') in (12), the estimation equation is defined by Equation 20.

From here,

ego,

Is an autocorrelation estimate of the channel, for one million samples Obtained by averaging

In the following ML estimator (hereinafter, referred to as 'MLE'), the estimation equation is defined by Equation 21.

In Equation 21 Is expressed by Equation 22.

In Equation 22, B (i, j) is a KxK matrix The (i, j) th element of.

Here, the matrix A is a KxL matrix whose (i, j) th element is d (ij).

For MLE, the parameter in equation 21 Is fixed to 8, and an exhaustive search is needed to estimate the frequency error.

For both ABE and MLE, L has an upper boundary of K-1.

In particular, the value of L is assumed to be 7 to implement this estimator (ABE, MLE) because it includes all valid channel elements.

However, in the case of the estimator according to the present invention, even if the value of L is assumed to be 1, the value of L is set to 1 because it does not significantly influence performance.

2 is a normalized carrier frequency error The mean square error (MSE) is shown when is changed from 0 to 0,05.

As shown in FIG. 2A, When N = 10, the performance of the estimator according to the present invention is in the middle of the performance of the MLE and ABE, as shown in Figure 2b It can be seen that it shows better performance than MLE and ABE.

MLE shows good performance for slowly changing channels, but shows a dramatic decrease in performance as the Doppler frequency changes.

These results have several About the value It is shown in detail in Figure 3 comparing the = 0 MSE values.

As shown in FIG. 3, the LSE with N = 10 is Starts to show better performance than other estimators, As the value increases, the performance difference widens.

These results indicate that the LSE according to the present invention is less sensitive to channel changes than other estimators and is useful for mobile communication environments.

Looking at the computational complexity of the above estimators, Table 3 shows the number of multiplications and additions required to implement each estimator.

Coefficients in Table 3 above to be.

4 is a sum of the multiplication and the addition according to the training signal K, and the calculated value is calculated by calculating Table 3 for N = K / 2 and L = 1,3.

It can be seen that the computational complexity of the LSE falls between ABE and MLE.

An example in which the LSE described above is applied to a transmission diversity system will be described below with reference to FIG. 5 representing a system model.

The transmit diversity system And a space-time encoder 50 is introduced to generate a symbol {s _i (k)} to be transmitted through the i th transmit antenna.

A multiplier (20b, ..., 20Γ) of the channel response to the symbols generated by the space-time coder 50, the adder 30b that adds all the values multiplied by the multiplier and the output of the adder 30b Multiplier (20a) to multiply by the output of the multiplier (20a) It consists of an adder 30a that adds.

Assuming that the symbol {s _i (k)} passes through the flat fading channel, that is, if the channel between the receiving antenna and the i th transmit antenna is the flat fading channel, the received signal r (k) is expressed by Equation 23.

Where g _{i, k} represents the channel response between the i th transmit antenna and the receive antenna.

In FIG. 5, it may be assumed that a signal transmitted from each transmit antenna is received by the receive antenna with the same signal delay.

Because of the fact that the symbol (or chip in CDMA system) duration is measured in microseconds, the transmission delay is measured in nanoseconds between the transmitting antennas in the cellular bands. to be.

Under these assumptions, the estimate of LSE can be written as:

In Equation 24 Is the same as Equation 8, d _m (k) of Equation 3 and g of Equation 4 are defined by Equation 25 and Equation 26.

6 shows a transmitter and a receiver for a common pilot channel (CPICH) of a 3GPP system employing transmit diversity.

The 3GPP system is It has two transmit antennas.

As shown in FIG. 8, the CPICH symbol generator 60 of the common pilot channel emits pilot signals s ₀ (k) and s ₁ (k).

These signals are multiplied by the same channelization code 70a and scrambling code 80a by multipliers 20b, ..., 20e and then transmitted over multipath channels 110a, 110b.

The transmitted signal is received by the RAKE receiver 90 having W + 1 fingers 100a,..., 100w, finger through the multiplier 20a and adder 30a constituting the estimator.

The RAKE receiver 90 receives W + 1 multipath components separately.

Each of the fingers 100a,..., 100w is a multiplier 20ma that multiplies the output of the adder 30a by the scrambling code 80m, and a channelization code (i.e., the output of the multiplier 20ma). A multiplier 20mb multiplying by 70m) and an mth order adder 120 adding m outputs of the multiplier 20mb.

At each finger 100a, ..., 100w A signal with a chip rate of is descrambled and correlated with the channelization code.

The output of the w th finger, r _w (k), can be obtained by extracting this signal.

Due to the autocorrelation characteristic of the code, the signal of each finger can be seen that the transmission signal of each antenna has passed through the flat fading channel, and r _w (k) is represented by Equation 27.

In Equation 27, g _{i, k} (w) is a channel response associated with the w th finger and the i th transmit antenna.

Since Equation 27 may be regarded as a special case of Equation 23, an estimator calculated by Equation 24 may be used to estimate the frequency error of Equation 27.

The performance of the estimator described above is confirmed through computer simulation, but ABE and MLE, which are easily applied in the form of Equation 23 and Equation 27, are used in the 3GPP system and compared with the LSE according to the present invention.

In addition, if the estimator of (10) is one transmit antenna in the flat fading channel in order to check the transmit diversity effect on the frequency error, ) And other estimators are two transmit antennas Respectively).

The simulation is done according to the following 3GPP specifications.

That is, the CPICH symbol rate is 15 ksymbol / sec, the carrier frequency is 2 GHz and the training signal K is 20.

9 is a view showing the simulation results, comparing the performance of the LSE, ABE and MLE, as shown in Figure 9a the speed of the moving object is 3km / h ( ), The performance of the LSE is in the middle of the performance of the MLE and ABE, but as shown in Figure 9b the speed of the moving object is 120km / h ( ), It shows better performance than MLE and ABE.

If you look at MLE Showed the best performance when Shows the worst performance when.

Comparing the performance of the estimator in (10) with those of the other estimators, when the channel change is slow (3 km / h), the diversity gain is large but the channel changes rapidly (120 km / h). Performance is better than MLE, and similar to ABE.

In the case of FIG. 9B, it can be seen that only LSE can obtain some diversity gain, and these results indicate that the estimator according to the present invention is better than other schemes in the 3GPP mobile communication environment. .

Since the training signal K can be increased as desired in the 3GPP system, it is preferable to increase the K value instead of increasing the N value in order to reduce the MSE of the estimator according to the present invention. You can get it without sacrifice.

As described above, according to the present invention, a new method for carrier frequency recovery based on the least squares (LS) error criterion is proposed, which does not require any information about the channel and performs well in a time-varying channel. Not only is it simple in terms of application, but it can also be applied to 3GPP systems in which a transmit antenna city technique is introduced.

Claims (7)

When the waveform generated at the transmitter is received at the receiver through a Rayleigh fading channel, a multiplier, and an adder, the frequency difference is estimated by calculating a delayed differential decoded received signal based on a least square error standard for carrier frequency recovery. LS carrier frequency error estimation method characterized in that. The method of claim 1, wherein the frequency error estimation, Obtaining a LS estimate by selecting a value that minimizes the sum of error squares; Estimating a carrier frequency error by looking at each element of the LS estimate value. Linear convolution of d (j), the transmit M-ary PSK or QAM symbol, to be transmitted through the transmit antenna at jth time, and h (t), the baseband waveform shaping function. Rayleigh fading channel is input and output, At the output of the Rayleigh fading channel A multiplier for multiplying and outputting An adder for adding and outputting n (t) to the output of the multiplier, LS carrier frequency error estimator, characterized in that the baseband system. Where n (t) is AWGN, Is the frequency error to be estimated and θ is the initial random phase.) 4. The LS carrier frequency error estimator of claim 3, wherein the carrier frequency error is calculated by Equation 28. Where Denotes an estimated value estimated based on the least squares (LS) error criterion of Pm. When the channel length is L + 1 and the number of training signals is K, the range of k that can take the differential decoding signal is Having the above range Is Where Dm is a matrix of training signals and Pm is a vector of channel and carrier frequency errors. When a waveform made at the transmitter is received at the receiver through a Rayleigh fading channel, a multiplier, and an adder, a transmission that estimates a frequency error by calculating a delayed differential decoded received signal based on a least squares (LS) error criterion for carrier frequency recovery. In a diversity system, A space-time encoder for generating a symbol to be transmitted through a plurality of transmit antennas when the channel is in a flat fading environment, A plurality of multipliers for multiplying a channel response by a symbol generated by the space-time encoder; An adder that adds all the values multiplied by the multiplier, On the output of the adder A multiplier to multiply by the output of the multiplier With an adder that adds Configured to transmit diversity system. (here, Is the frequency error to be estimated, theta is the initial random phase, Means AWGN respectively.) 3GPP estimates the frequency error by calculating the delayed differential decoded received signal based on the least squares (LS) error criterion to recover the carrier frequency when the waveform made at the transmitter is received at the receiver through Rayleigh fading channel, multiplier and adder. In the transceiver of the system, A CPICH symbol generator for generating pilot signals of a common pilot channel to be transmitted through two transmit antennas; A multiplier for multiplying the pilot signal by the same channelization code and a scrambling code; A multipath channel for transmitting a signal output through the multiplier, respectively; A multiplier and an adder through which the transmitted signal is passed; A RAKE receiver comprising a plurality of fingers, each receiving a multipath component of a signal passed through the multiplier and the adder separately, Transceiver of the 3GPP system, characterized in that configured. 7. The transceiver of claim 6 wherein the finger descrambles a signal having a constant chip rate and correlates with a channelization code.