WO2005048497A2

WO2005048497A2 - Method for estimating time varying channels in ofdm (orthogonal frequency division multiplex) multiple transmit antenna system

Info

Publication number: WO2005048497A2
Application number: PCT/KR2004/002436
Authority: WO
Inventors: Yu-Ro Lee; Dae-Sik Hong; Seong-Rag Kim; In-Kyeong Choi; Choong-Il Yeh; Hyoung-Soo Lim; Seong-Chul Cho; Jong-Ee Oh; Dong-Seung Kwon; Jee-Hwan Ahn; Seung-Ku Hwang
Original assignee: Electronics And Telecommunications Research Institute; Samsung Electronics Co., Ltd.; Kt Corporation; Sk Telecom Co., Ltd.; Ktfreetel Co., Ltd.; Hanaro Telecom, Inc.
Priority date: 2003-11-13
Filing date: 2004-09-22
Publication date: 2005-05-26
Also published as: WO2005048497A3; KR20050046209A

Abstract

Disclosed is a time-varying channel estimation method in an OFDM which comprises: (a) generating a pilot symbol by a comb type method for inserting a pilot symbol into a specific subcarrier for each interval in each OFDM symbol; (b) transmitting the pilot symbol to a receiver from a transmitter through a time-varying channel by using at least two multiple transmit antennas; (c) receiving the pilot signal through the time-varying channel; and (d) estimating a first channel of from among the time-varying channel according to an interpolation method which uses a temporal correlation of a second channel and a third channel.

Description

METHOD FOR ESTIMATING TIME VARYING

CHANNELS IN OFDM (ORTHOGONAL FREQUENCY

DIVISION MULTIPLEX) MULTIPLE TRANSMIT

ANTENNA SYSTEM Technical Field

[1] The present invention relates to a method for estimating a time- varying channel in an OFDM (orthogonal frequency division multiplex) system. More specifically, the present invention relates to a method for estimating a time- varying channel in an OFDM system for using multiple transmit antennas and arranging pilots in a comb manner. Background Art

[2] In the case of transmitting signals through a multiple path channel, ISI

(inter-symbol interference) occurs in the received signals because of the multi-path. In particular, since the symbol period is less than the delay spreading of the channel in the case of high-rate data transmission, the ISI is further increased, and hence, a more complicated receiving method is required in order to compensate for the distortion caused by the ISI and accurately reconstruct transmission signals. The symbol period is to be greater than the delay spreading of the channel in order to reduce the signal distortion phenomenon caused by the ISI. The OFDM method has been proposed as a method for simply compensating for the distortion in the multi-path channel. The OFDM method uses subcarriers having orthogonality to transmit data, differing from the transmission method which uses single carriers. That is, the OFDM method performs serial and parallel conversion on the input data by the number of the subcarriers used for modulation, modulates the respective converted data by using the corresponding subcarriers, and accordingly, increases the symbol period of each subcarrier by the number of subcarriers while maintaining the data rate. Since the OFDM method uses the subcarriers with orthogonality, the OFDM method provides better bandwidth efficiency and a longer symbol period compared to the conventional FDM (frequency division multiplex) method, and hence, it has a stronger characteristic against the ISI than the single subcarrier modulation method.

[3] A modulation process and a demodulation process by a transmitter and a receiver in the OFDM system respectively correspond to an IDFT (inverse discrete Fourier transform) process and a DFT (discrete Fourier transform) process which are effectively realized by using IFFT (inverse fast Fourier transform) and FFT (fast Fourier transform). Also, when a guard interval which is longer than the delay spreading of the channel is inserted for each symbol period, the orthogonality between the subcarriers is maintained to generate no ICI (inter-carrier interference), and no OFDM symbols caused by the multi-path channel are superimposed to thereby completely eliminate the ISI between the adjacent symbols.

[4] Further, the OFDM transmission method has attracted considerable attention in the wireless communication system field since it is appropriate for multi-path fading channels, and has high bandwidth efficiency. IEEE80211a (US) and ETSI HIPERLAN/2 (EU) which are WLAN (wireless local area network) standards for the OFDM support the maximum data rate of 54Mbps, which is applicable to high-speed Internet and intranet such as the DAB (digital audio broadcasting), DVB (digital video broadcasting), and ADSL (Asynchronous digital subscriber line) as well as multimedia application fields. It will be naturally realized in the near future that the system transmission rates will be increased in various application fields. Studies for maximizing capacity by using multiple transmit/receive antennas have recently been progressed.

[5] A configuration of a transmission and receiving system of the OFDM system and a channel estimation method will be described with reference to drawings.

[6] FIG. 1 shows a block diagram for a general OFDM transmission and receiving system.

[7] The general OFDM transmission and receiving system comprises a signal mapper

21, a serial/parallel converter 22, an IFFT (inverse fast Fourier transform) unit 23, a pilot inserter 24, a parallel/serial converter 25, a Rayleigh fading channel applier 26, an adder 27, a serial/parallel converter 28, a pilot remover 29, a channel estimator 30, an FFT (fast Fourier transform) unit 31, a channel equalizer 32, a parallel/serial converter 33, and a baseband demodulator 34.

[8] When a binary source is generated, the signal mapper 21 of the transmission side maps the data to be transmitted, that is, the binary source according to a modulation method (e.g., the QAM method) by generally using the QPSK (quadrature phase shift keying), 16-QAM, or 64-QAM method.

[9] The data mapped by the signal mapper 21 are converted into parallel data by the serial/parallel converter 22, the parallel data are inverse-Fourier-transformed by the IFFT unit 23, a pilot symbol is inserted into the inverse-Fourier-transformed data by the pilot inserter 24, the pilot inserted data are converted again into the serial data by the parallel/serial converter 25, and serial data are then output. In this instance, in order to allow the receiver side to easily perform demodulation, the transmission side of the OFDM system inserts the pilot symbol which is known to the receiver side into intervals within the data to be transmitted.

[10] The data converted into the serial format by the parallel/serial converter 25 are passed through the Rayleigh fading channel applier 26, they are added to noise by the adder 27, and the noise-added data are transmitted to the serial/parallel converter 28.

[11] The pilot symbol is removed by the pilot remover 29 from the signals converted into the parallel format by the serial/parallel format 28, and the channel estimator 30 uses the pilot symbol to find a channel estimate from the OFDM signal.

[12] The signal output by the channel estimator 30 is input to the FFT unit 31 to be

Fourier transformed, and the Fourier transformed signal is transmitted to the channel equalizer 32 The channel equalizer 32 compensates for abnormal characteristics of the channel, that is, various types of noise, interference with adjacent channels, and channel distortion caused by the multi-path.

[13] The signal passed through the channel equalizer 32 is converted into the serial format by the parallel/serial converter 33, and the serial signal is output as an output signal through the baseband demodulator 34.

[14] The radio channels in the broadband mobile communication system are frequency- selective and time- varying, which represents that the channels are not the same on the frequency axis and the temporal axis in the OFDM system. In addition, when using the multiple transmit antenna, receiving symbols for the respective subcarriers are superimposed after the symbols transmitted from the multiple transmit antenna are independently faded. Therefore, efficient channel estimation is required before demodulating the OFDM symbol since the channel is frequency-selective and time- varying in the mobile OFDM communication system which uses the multiple transmit antenna. The conventional channel estimation is performed by inserting a pilot symbol into the subcarrier known on the temporal axis and the frequency axis, and using the same.

[15] The method for inserting the pilot symbol is classified as a block type method and a comb type method. The block type method is a method for inserting a pilot symbol into all the subcarriers of a specific OFDM symbol, and repeating the inserting process. Since the pilot symbol is included in all the subcarriers, the block type method is relatively strong for the channel's frequency-selective characteristic, and is suitable for slow fading channels, and since the same method needs no interpolation method in the frequency domain, it is known as a pilot symbol arranging method which is relatively strong for the frequency-selective fading.

[16] The comb type method is a method for inserting a pilot symbol into a specific subcarrier for each constant interval and for each OFDM symbol, and it is known as a pilot symbol arranging method that is strong for fast fading compared to the block type method. However, since the pilot symbol is to be inserted for each interval, gains of the subcarriers are to be estimated by using the pilot symbol for each OFDM symbol, and it is accordingly known to be relatively weak for the frequency-selective fading.

[17] The comb type method has higher retransmission rates than the block type method assuming that the two methods have the same ratio of the data to the pilot symbols. Therefore, the comb type method is more appropriate for fast fading channels.

[18] In addition, the OFDM is a method for transmitting plural carriers, and it increases the transmission period of signals in proportion to the number of carriers. In this case, the ISI caused by the frequency-selective channel generated at the time of high data transmission is eliminated, and the channels are approximated with frequency-non- selective channels and processed. Referring to FIG. 1, the signal processing according to the channels is realized by using the IFFT at the transmitter side and the FFT at the receiver side.

[19] FIG. 2 shows a brief schematic diagram for a mobile OFDM system which uses general multiple transmit antennas, and FIG. 3 shows a general comb type pilot symbol arrangement diagram.

[20] As shown in FIG. 2, in the above-noted mobile OFDM system which uses general multiple transmit antennas, pilot symbols la and lb at a first transmitter Tx#l and a second transmitter Tx#2 are respectively passed through a first IFFT unit 3a and a second IFFT unit 3b to be inverse Fourier transformed, and the transformed symbols are transmitted through multiple transmit antennas 5a and 5b at the transmitter. Also, signals received through a multiple transmit antenna 7 at the receiver are Fourier transformed by an FFT 9, and in this instance, a pilot symbol 11 at the receiver is removed through the pilot remover, and a channel estimate is found from the OFDM signals by a channel estimator by using the pilot symbol. A detailed method for estimating the channel will be described later.

[21] Referring to FIG. 3, the general comb type pilot symbols are established to be inserted into specific subcarriers for each predetermined interval for each OFDM symbol. [22] The channel estimation method using the block type and comb type pilot patterns includes methods for applying the LS (least square) and the MMSE (minimum mean- square error). The LS applied method requires no prior information on the channel and noise and needs low complexity.

[23] The OFDM signals are transmitted for each symbol, and the ISI is generated while the OFDM signals are passed through a wireless multi-path channel. In order to prevent the ISI, a guard interval is inserted between the symbols. That is, the inter- signal interference is prevented by providing a guard interval which is longer than the maximum delay time of the channel. The last portion of the symbol duplicated and inserted into the guard interval is called a cyclic prefix, and the breakage of orthogonality caused by a signal delay is prevented by allowing the cyclic prefix.

[24] In the above^nentioned OFDM method, the equalizer in each subchannel has a single tap format since the subchannel is approximated to be a frequency non-selective fading channel, and channel estimation for estimating each subchannel by the receiver is required since the coefficient of the equalizer is an inverse value of the estimated subchannel. The channel estimator can be induced by applying the LS or MMSE standard in the frequency domain. The LS channel estimator has a simple configuration, but it is sensitive to noise since it uses no statistic characteristics of the channel, and the MMSE channel estimator substantially increases calculation compared to the LS estimator since the MMSE channel estimator uses the channel's second moment such as autocorrelation and cross correlation, but provides excellent performance in the case of the fading channel with a severe null since the MMSE channel estimator minimizes the total estimation error considering noise.

[25] The channel estimation of the OFDM method is classified, according to categories of data, as a PSA (pilot-symbol-aided) channel estimation method and a DD (decision-directed) channel estimation method. The PSA channel estimation method is suitable for high-speed fading channels, and the pilot symbols are arranged in consideration of coherence bandwidths of channels, coherence time, and bandwidth efficiency reduction according to usage of pilot tones. The DD channel estimation method is appropriate for low-speed fading channels with great fixed or temporal correlation since the method estimates the channel of a next symbol period by using detection data.

[26] A general channel estimation method using the LS method in the mobile OFDM system which uses the comb type based multiple transmit antennas will now be described. For ease of description, the case of using two transmit antennas will be described, and it is obvious to a person skilled in the art to use further transmit antennas. [27] A frequency-axis signal Y received at the p subcarrier with a pilot symbol at the th . . . n.P n time is given in Math Figure 1. [28] MathFigure 1

[29] where X is a pilot symbol on the frequency axis transmitted at the m antenna n,p and at the n time's p subcarrier, H is a corresponding channel value on the n,p frequency axis, W is an AWGN (additive white Gaussian noise) with a mean of 0 and a variance of σ ,

is given as [X X ], n τ0 is [H H ] , and {•} is a transpose. n,p n,p

[30] The number of pilot symbols provided within an OFDM symbol is given as P, and it is the minimum multiple of 2 greater than L (the maximum length of the channel). The P pilot symbols are inserted with the same intervals between the subcarriers within the symbol, and their positions are established to be given as

[31] From Math Figure 1, a necessary and sufficient condition for estimating the LS channel is

. To satisfy this, the pilot symbol of the (n+1) symbol is required, but the desired

is increased to 4 because of the corresponding channel. As a result, the subsequent assumption is provided in order to fix a desired

[32] MathFigure 2

[33] where h ^m is a channel of the n time corresponding to the m transmit antenna. n

Therefore, Math Figure 1 is given as Math Figure 3.

[34] MathFigure 3

[35] where

is [Y Y ] n,p n+l,p

is [H H ],

is [W W ] , and n+l,p

[36] Therefore, the estimation of the LS of is performed according to Math Figure 4. [37] MathFigure 4

[38] where

is a pseudo inverse matrix. Math Figure 5 is given as follows from Math Figures 3 and 4. [39] MathFigure 5

[40] Since

is unbiased, the MSEof channel estimation is given in Math Figure 6. [41] MathFigure 6

[42] where

ό{-) is expectation, and ^ft⁾ is trace. The pilots of other transmit antennas are to be orthogonal in order to obtain the MMSE of the LS channel estimation. Accordingly, a pilot symbol is exemplified as Math Figure 7.

[43] MathFigure 7

[44] However, channel variation within two OFDM symbol period lowers the performance according to the channel estimation, because of the assumption of Math Figure 2 That is, sufficient accuracy on the channel estimation is not guaranteed when the Doppler frequency is high.

[45] The transaction in "Optimal training design for MIMO OFDM systems in mobile wireless channels" by I. Barhumi has been proposed on IEEE Trans, in signal processing, pp. 1615 to 1624, Vol.51, book 6, June 2003.

[46] I. Barhumi proposed the transaction on a pilot design for channel estimation in order to minimize the MSE of channel estimation in the OFDM system which uses the multiple transmit antennas with a pilot pattern of the block type or the comb type. For this, the required number of pilot subcarriers for channel estimation is given as P (which is the length of the transmit antenna channel). In order to minimize the MSE of channel estimation, two conditions need to be satisfied according to the equations provided by the transaction.

[47] The first condition is to arrange the pilot symbols in an OFDM symbol with the same intervals. The second one is to allow the pilot signal of a random reference transmit antenna and the pilot signal of another transmit antenna to be orthogonal, and allow the signal delayed by the channel of the antenna to be orthogonal with the other antenna signal. That is, the signal obtained by circularly transitioning the other antenna on the temporal axis is to be orthogonal with the reference antenna signal.

[48] The MSE of channel estimation is minimized when the two conditions are satisfied. However, the method has been designed with the assumption that the channel is constant during the number of pilot subcarriers needed for channel estimation, that is, P. In detail, the number of total subcarriers is to be greater than P in the case of the system with the block type pilot pattern, and the accuracy of channel estimation is degraded when the channel is time- varying in the case of the data symbol provided after the pilot symbol. Also, an equalizer with high complexity using various algorithms is needed in order to track the channels. Further, it is assumed that the channel is not varied during the symbols which correspond to the number of transmit antennas, when the minimum pilot subcarriers (the length of the channel) are used for one symbol in order to reduce the load caused by the pilot in the case of the system with the comb type pilot pattern. Therefore, the accuracy of channel estimation is degraded when the channel is time varying.

[49]

Disclosure of Invention Technical Problem

[50] It is an advantage of the present invention to provide a more accurate channel estimation method by using a temporal correlation of a time-varying channel in an OFDM system for using multiple transmit antennas and arranging pilot symbols in a comb manner.

[51] It is another advantage of the present invention to provide a channel estimation method of a time- varying channel for reducing complexity by estimating a channel according to a linear interpolating method for each subcarrier at which a pilot symbol is provided. Technical Solution

[52] In one aspect of the present invention, a time- varying channel estimation method in an OFDM comprises: (a) generating a pilot symbol by a comb type method for inserting a pilot symbol into a specific subcarrier for each interval in each OFDM symbol; (b) transmitting the pilot symbol to a receiver from a transmitter through a time-varying channel by using at least two multiple transmit antennas; (c) receiving the pilot signal through the time-varying channel; and (d) estimating a first channel from the time-varying channel according to an interpolation method which uses a temporal correlation of a second channel and a third channel.

[53] The step (d) comprises: estimating the time- varying channel through an LS (least square) method.

[54] The interpolation method is a linear interpolation method.

[55] The pilot symbol is designed to satisfy a rank for multiple channel estimation for each symbol period in order to reduce performance degradation caused by time- varying of the channel, when the pilot symbol is designed by using a temporal correlation of the channel.

[56] When a frequency-axis signal Y transmitted from the p subcarrier into which a pilot symbol is inserted at the n time is given as Y=AH+W where Y is [Y Y Y n,p n+l,p T 1 2 1 2

Y ] , the frequency-axis channel value H is [H H H H ], and the n+2,p n+3,p n,p n,p n+3,p n+3,p additive noise W is [W W W W ], the channel h of the (n+l) time and n,p n+lj n+2,p n+3,p n+1 ft . the channel h of the (n+2) time corresponding to the m transmit antenna are n+2 represented in the subsequent equations by using the linear interpolation channel estimation method through the n h and the (n+3) h n+3

[57] m h SS! n + n+3 Stf ^' M n+

H -1 h n+2

[58] and when X is a frequency-axis pilot symbol transmitted from the p subcarrier n,p of the n time of the m antenna, the pilot symbol is designed to satisfy the subsequent equation:

[59]

Brief Description of the Drawings

[60] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate an embodiment of the invention, and, together with the de- scription, serve to explain the principles of the invention; [61] FIG. 1 shows a block diagram for a general OFDM transmission and receiving system; [62] FIG. 2 shows a brief schematic diagram for a general mobile OFDM system which uses multiple transmit antennas; [63] FIG. 3 shows a general comb type pilot symbol arrangement diagram;

[64] FIG. 4 shows a diagram of a time- varying estimating method in an OFDM system which uses multiple transmit antennas and arranges pilot symbols in a comb manner; [65] FIG. 5 shows a diagram for showing the MSE and SNR according to the channel estimation method of a preferred embodiment of the present invention and a general channel estimation method; and [66] FIG. 6 shows a diagram for showing the MSE and the normalized Doppler frequency of f T according to the channel estimation method of a preferred d embodiment of the present invention and a general channel estimation method. Best Mode for Carrying Out the Invention

[67] In the following detailed description, only the preferred embodiment of the invention has been shown and described, simply by way of illustration of the best mode contemplated by the inventor(s) of carrying out the invention. As will be realized, the invention is capable of modification in various obvious respects, all without departing from the invention. Accordingly, the drawings and description are to be regarded as illustrative in nature, and not restrictive. To clarify the present invention, parts which are not described in the specification are omitted, and parts for which similar descriptions are provided have the same reference numerals.

[68] A time- varying channel estimation method in the OFDM system will be described.

[69] In general, it is difficult to apply a general channel estimation method to the time- varying channel since it is needed to assume that symbols which correspond to the number of transmit antennas are required temporally and the channels are not varied during the time in order to estimate the channels in the case of using the comb type pilot arrangement. Also, pilot symbols for a period of two symbols are needed in order to estimate multiple channels of the n symbol period according to the multiple transmit antennas.

[70] Channel estimation for each symbol period is needed in order to reduce performance degradation caused by channel variation. Therefore, temporal correlation of the channel is used in order to satisfy ranks for multiple channel estimation and reduce the Doppler effect, and the time-varying channel estimation method will now be described in detail with reference to FIG. 4. [71] FIG. 4 shows a diagram of a time- varying estimating method in an OFDM system which uses multiple transmit antennas and arranges pilot symbols in a comb manner.

The reference numerals 13a, 13b, 13c, and 13d in FIG. 4 represent symbols into which pilot symbols are inserted for each symbol period in the comb manner, and the black areas indicate that the pilot is inserted into the symbol. [72] Referring to FIG. 4, h and h are represented by the temporal correlation of n+l n+2 h and h when two transmit antennas are used. That is, the channel h of the n n+3 n+l

(n+l) time and the channel h of the (n+2) time corresponding to the m transmit n+2 antenna can be represented by the temporal correlations of the n h and the (n+3) n h according to the linear interpolation. n+3

[73] In the case of using the linear interpolation, the subsequent Math Figure 8 is given

(refer to FIG. 4). [74] MathFigure 8

9tι^ffl + ^m h ^m + Jh ^m

^■ M -1 Yl +

[75] Further accurate channel estimation is possible but the complexity is increased when a more accurate interpolation method such as a high-dimensional polynomial interpolation method is used.

[76] Math Figure 8 can be given as Math Figure 9.

[77] MathFigure 9

Y=ΛH+ W

[78] where Y is [Y Y Y Y ] , the frequency-axis channel value H is [H H n+l,p n+2,p n+3,p ln,p

H H ] , the additive noise W is [W W W W ] , and

2n,p ln+3,p 2n+3,p n,p n+l,p n+2,p n+3,p

[79] p

, and hence, the multiple channels are estimated as follows according to the LS method.

[80] MathFigure 10

[81] As a result,

is given below.

[82]

[83] The above-described has shown the case of using two transmit antennas, but is obvious for a person skilled in the art to estimate the channels in the case of at least two transmit antennas.

[84] A computer simulation for proving the efficiency of the channel estimation method according to the preferred embodiment of the present invention has been performed. Simulation conditions are: L=8, i.i.d (independent identically distributed), and the temporal correlation follows the Jakes' model of

. The number of subcarriers within a symbol in the OFDM system is 64, and the length of the guard interval is 16. P is established to be 8 in order to satisfy the minimum pilot symbol number for channel estimation. Also, the QPSK is used for a modulation method, and the test is based on two transmit antennas. System performance is measured by the MSE of LS channel estimation.

[85] FIG. 5 shows performance of the MSE according to the SNR (signal to noise ratio) for the general channel estimation method and the channel estimation method according to the preferred embodiment of the present invention when the normalized Doppler frequency f T is given as 0 and Q02 d

[86] When the normalized Doppler frequency f T is 0, that is, when the channel is not d varied temporally (i.e., the case of the time-invariable channel), the general channel estimation method generates a gain of about ldB with respect to the SNR, and when the normalized Doppler frequency f T is Q02, the general channel estimation method d generates the error floor phenomenon. The channel estimation method according to the preferred embodiment generates no variation of performance when the channel is

-2 quickly varied. When the MSE is given as 10 , the channel estimation method according to the preferred embodiment achieves a gain of about 2dB compared to the general channel estimation method (refer to a, b, c, and d in FIG.5). [87] FIG. 6 shows the performance of the MSE depending on the frequency of f T with d the given SNR of 24dB according to the conventional block type method, the general comb type channel estimation method, and the comb type channel estimation method according to the preferred embodiment of the present invention. In order to suitably compare the block type method with the comb type methods, the ratio of the pilot symbols and the data is controlled to thus insert a pilot symbol after seven data symbols in the block type method. As shown in FIG. 6, the block type method generates the same performance as that of the comb type method in the time-invariable channel, but as the normalized Doppler frequency of f T is increased, the performance d is steeply reduced, and the actual usage becomes impossible when the Doppler frequency is high. The case of using the general channel estimation method in the comb type method provides better performance than the channel estimation method according to the preferred embodiment when the channel is varied slowly. The reason for providing better performance is that the design of the pilot symbols according to the channel estimation method according to the preferred embodiment fails to satisfy the orthogonal matrix even though the general design of the pilot symbols is performed with the orthogonal matrix in order to satisfy the MMSE. When the pilot symbols are designed which satisfy the orthogonal matrix in the channel estimation method according to the preferred embodiment, the two methods provide the same performance when the channel is temporally varied slowly. When the channel is quickly varied temporally, that is, when the normalized Doppler frequency of f T is d great, the channel estimation method according to the preferred embodiment generates far better performance compared to the general channel estimation method (refer to e, f, and g of FIG. 6). [88] Accordingly, as described above, the transaction by I. Barhumi aims at designing transmission pilots for minimizing the channel estimation MSE in the OFDM system, allows the pilot subcarriers to be arranged with the same intervals within an OFDM symbol and to satisfy transition orthogonality, and applies the LS method when estimating the channel, but when the channel is not varied, or when the channel is time- varying in the optimal channel estimation method, the accuracy of channel estimation is lowered.

[89] According to the preferred embodiment of the present invention, a channel estimation method and a pilot design in consideration of the time- varying channel in the OFDM system which uses the multiple transmit antennas with a comb type pilot pattern are provided. That is, the pilot subcarriers are arranged with the same intervals in one OFDM symbol, the pilot symbols are designed to satisfy the orthogonality, and the LS method using a simple interpolation method is applied by considering the time- varying channel at the time of channel estimation. That is, it is needed to satisfy the orthogonality when designing the pilot symbols, and the channels of the LS method can be estimated together with a simple interpolation method by considering the time- varying channel at the time of channel estimation.

[90] According to the preferred embodiment, there is very much less performance difference between the time-varying channel and the time-invariant channel, and hence, further accurate channel estimation is allowed compared to the conventional channel estimation method.

[91] While this invention has been described in connection with what is presently considered to be the most practical and preferred embodiment, it is to be understood that the invention is not limited to the disclosed embodiments, but, on the contrary, is intended to cover various modifications an equivalent arrangements included within the spirit and scope of the appended claims.

[92] As described, the channel estimation method according to the preferred embodiment of the present invention uses the temporal correlation of the channels and estimates the channels in consideration of the Doppler effect, and accordingly, maintains the accuracy of channel estimation without degradation of performance differing from the general channel estimation methods in the time- varying channels.

[93] Also, since the channel estimation is independently performed for each simple linear interpolation and subcarrier, the complexity is reduced, and the performance of the OFDM system is improved by designing the pilots appropriate for the channel estimation method.

[94]

[95]

Claims

[1] A time- varying channel estimation method in an OFDM (orthogonal frequency division multiplex) system, comprising:

(a) generating a pilot symbol by a comb type method for inserting a pilot symbol into a specific subcarrier for each interval in each OFDM symbol;

(b) transmitting the pilot symbol to a receiver from a transmitter through a time- varying channel by using at least two multiple transmit antennas;

(c) receiving the pilot signal through the time- varying channel; and

(d) estimating a first channel of from among the time- varying channel according to an interpolation method which uses a temporal correlation of a second channel and a third channel.

[2] The time- varying channel estimation method of claim 1, wherein (d) comprises: estimating the time-varying channel through an LS (least square) method.

[3] The time- varying channel estimation method of claim 1, wherein the interpolation method is a linear interpolation method.

[4] The time- varying channel estimation method of claim 1, wherein the pilot symbol is designed to satisfy a rank for multiple channel estimation for each symbol period in order to reduce performance degradation caused by time- varying of the channel, when the pilot symbol is designed by using a temporal correlation of the channel.

[5] The time- varying channel estimation method of claim 3, wherein when a frequency-axis signal Y transmitted from the p subcarrier into which a pilot symbol is inserted at the n time is given as Y=AH+W where Y is [Y Y Y n,p n+l,p T 1 2 1 2

Y ] , the frequency-axis channel value H is [H H H H ], n+2,p n+3,p n,p n,p n+3,p n+3,p and the additive noise W is [W W W W ], the channel h of the n,p n+l,p n+2,p n+3,p n+l

(n+l) time and the channel h of the (n+2) time corresponding to the m n+2 transmit antenna are represented in the subsequent equations by using the linear interpolation channel estimation method through the n h and the (n+3) h n n+3

and when X is a frequency-axis pilot symbol transmitted from the p n,p subcarrier of the n time of the m antenna, the pilot symbol is designed to satisfy the subsequent equation:

x n,p X: n,p 0 0

__"!_

1 xt n+2,p - ^lx n+2,p -X n¹+2,p _T n+2,p

0 0 X n+2,p n+3,p