KR100602518B1 - Method and apparatus for channel estimation for ofdm based communication systems - Google Patents

Abstract

The present invention relates to channel estimation in an OFDM communication system, and further comprising: performing an FFT operation on a received OFDM signal and preliminarily estimating a channel by using an LS symbol on a result of the FFT operation. Performing an IFFT operation on the preliminary channel estimation result to obtain an estimated value of a channel impulse response (CIR); multiplying the estimated value by a predetermined amplitude adjustment coefficient; Comparing a predetermined threshold, improving the accuracy of the CIR estimate through the multiplying and comparing step, and converting the enhanced CIR estimate into a channel estimate by an FFT operation. A channel estimating method and a channel estimator implementing the same are provided.

OFDM, channel estimation, pilot symbol, STO, CIR

Description

Method and apparatus for channel estimation of orthogonal frequency division multiplexing system TECHNICAL FIELD OF APPARATUS FOR CHANNEL ESTIMATION FOR OFDM BASED COMMUNICATION SYSTEMS

1 is a block diagram of a channel estimator in accordance with the present invention.

2 illustrates the amplitude adjustment coefficient for Eb / N ₀ .

3 illustrates a threshold for Eb / N ₀ .

4 to 6 are diagrams for comparing the performance of the channel estimation technique according to the present invention and the conventional channel estimation technique.

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

100: first FFT operator 110: first pilot and data separator

120: LS channel estimator using pilot 130: scaler and zero inserter

140: IFFT operator 150: CIR adaptor

160: second FFT operator 170: second pilot and data separator

180: single tap equalizer

The present invention relates to an OFDM communication system, and more particularly, to a novel OFDM channel estimation technique that minimizes the effects of fading and symbol timing offset.

OFDM is a wideband modulation scheme that divides a frequency bandwidth allocated for a communication session into a plurality of narrowband frequency subbands, each subband including a radio frequency (RF) subcarrier, each subcarrier being included in each of the other subchannels. Is orthogonal to the received RF subcarriers. The orthogonality of the subcarriers allows their individual spectra to overlap without interference with other carriers. By dividing the frequency bandwidth into a plurality of orthogonal subbands, the OFDM scheme enables high data rates and very efficient bandwidth usage.

The OFDM method first converts data to be transmitted into a complex symbol in the form of M-ary quadrature amplitude modulation (M-QAM), and converts a complex symbol sequence, which is a sequence of complex symbols, through a series-parallel conversion. Multi-carrier modulation is performed by converting a plurality of parallel complex symbols and then performing a rectangular pulseshaping on each of the complex complex symbols and sub-carrier modulation. In the multicarrier modulation scheme, frequency intervals between subcarriers are set such that all sub-carrier modulated parallel complex symbols are orthogonal to each other.

When the M-QAM modulated signal is transmitted through the wireless fading channel without using the OFDM scheme, if the delay spread of the channel caused by the multipath delay is greater than the symbol period of the modulated signal, However, inter-symbol interference occurs, so that the correct signal restoration is impossible at the receiving end. Therefore, an equalizer must be used to compensate for random delay spread, but the implementation of the equalizer is not only very complicated but also has a disadvantage in that the transmission performance is degraded due to input noise at the receiving end.

On the other hand, using the OFDM scheme, since the symbol period of each parallel complex symbol can be much longer than the delay spread of the channel, the interference between symbols can be relatively small. In particular, by setting the guard interval longer than delay spread, there is an advantage that the interference between symbols can be completely eliminated. Of course, there is no need to implement an equalizer that compensates for random delay spread due to multipath delay. Accordingly, the OFDM scheme is very effective for data transmission over a wireless fading channel and is currently adopted as a standard transmission scheme for terrestrial digital television and audio broadcasting systems in Europe. In addition, data transmission system through wired channels such as digital subscriber loop (DSL) and powerline communication eliminates degradation of transmission performance due to multipath reflection occurring in line network environment. It is used a lot.

The transmitting end of the data transmission system using the OFDM scheme is a channel encoding means for first converting the data to be transmitted into coded data, and M-QAM and PSK (phase shift) through a mapper. After converting into complex symbols in the form of keying) and DPSK (differential PSK), and converting them into multiple parallel complex symbols through serial and parallel conversion, each parallel complex symbol is square-shaped and subcarrier modulated and then subcarrier modulated. And modulation means for carrier-modulating the sum of the signals, and transmission-end channel matching means, including an amplifier and an antenna, for transmitting the carrier-modulated signal through wireless and wired channels. In contrast to the transmitting end, the receiving end comprises a receiving end channel matching means, a demodulation means, a channel decoding means and the like.

As the channel encoding means, a number of methods including a convolutional encoding, a block encoding, a turbo encoding, or the like, or a suitable combination thereof are used. The square wave shaping and subcarrier modulation means of a plurality of parallel complex symbols among the transmitter modulation means are implemented as an inverse fast Fourier transform (IFFT) signal processing means based on a sampling theorem. Fast Fourier transform (FFT) signal processing means is used.

At the transmitting end of the OFDM system, the encoded data is converted into complex symbols through a mapper, and the complex symbols are adjacent to the complex symbols by a frequency interleaver and a frequency deinterleaver of the receiving end. Fading is independent of each other. Therefore, the coded data reconstructed at the receiving end is prevented from severe performance deterioration due to the loss of the burst form. However, the probability of loss of information due to fading is still very high, which results in a severe degradation of transmission performance compared to data transmission through an unfaded channel.

On the other hand, in an environment such as a terminal moving at a high speed, a problem of occurrence of intersymbol interference due to channel distortion is severely acted on, which causes the equalizer of the receiver to be complicated. Therefore, recent systems prefer the CDMA or OFDM scheme which does not cause the intersymbol interference problem. The CDMA method can distinguish multipaths present in a delay spread, and the OFDM method avoids intersymbol interference by dividing one carrier into several subcarriers so that a symbol interval is longer than a delay spread. However, even when using CDMA or OFDM, a complex equalizer is not required, but channel estimation and channel compensation in the form of a single tap equalizer for compensating for signal distortion by a channel are required.

Many methods have been proposed as channel estimation techniques in OFDM systems. The channel estimation algorithm can be roughly classified into two types, a pilot-symbol-aided channel estimation method and a decision-directed channel estimation method. The pilot estimation channel estimation method is a method of periodically estimating a signal called a pilot in the middle of data and using it to estimate the channel. The decision-oriented channel estimation method uses general data as well as pilot symbols. This method reduces the noise variance of channel estimation values. Both algorithms are useful to obtain good channel estimates by passing through filters using highly correlated adjacent channel values to reduce noise variance of channel estimates.

Representative examples of the specific computational technique include LS (Least Square) or LMMSE (Linear Minimum Mean Square Error) channel estimation technique. Among them, the LMMSE channel estimation technique is an excellent method in terms of mean square error (MSE) or bit error rate (BER) of a pilot subcarrier channel. The LMMSE channel estimation technique assumes that the channel's autocorrelation function and the received SNR are known. To find autocorrelation, Monte-Carlo simulation is used or mathematical models such as exponential / even distribution are used. Comparing their MSE or BER performance, it is better to use the autocorrelation function generated by the simulation. The OFDM transmitter performs LMMSE channel estimation by transmitting OFDM symbols having pilot symbols of equal intervals inserted between preambles and data symbols.

However, in the OFDM receiver, it is necessary to synchronize OFDM symbols for FFT processing, and at this time, a symbol timing offset (STO) may occur due to fading effects. Inter-Carrier Interference (ICI) or Inter-Symbol Interference (ISI) caused by this degrades the performance of the receiver. In other words, due to the effects of STO, the performance of the LMMSE channel estimation technique using the autocorrelation function generated by the simulation deteriorates drastically.

In order to solve the above-mentioned problems of the prior art, the present invention provides a novel and advanced channel estimation technique that is similar to the LMMSE channel estimation technique using the autocorrelation function generated by simulation without being affected by the STO. Has its purpose.

In order to achieve the above technical problem, the present invention proposes a new channel estimation technique modified from a discrete Fourier transform (DFT) based channel estimation technique using an existing autocorrelation function. The nucleus point of the present invention is to perform a channel estimation based on DFT without using an autocorrelation function, and in the present specification, this is called a channel impulse response adaptation technique, and a block for performing a CIR adaptation function is described. Named the CIR Adapter.

EMBODIMENT OF THE INVENTION Hereinafter, the structure of this invention is described in detail with reference to an accompanying drawing.

1 is a block diagram illustrating components for performing a channel estimation technique according to the present invention. The present invention is based on the premise that a transmitter transmits an OFDM symbol in which a pilot symbol is inserted for channel estimation, the OFDM symbol arrives at a receiver through multipath fading, and the channel model of multipath fading is known.

As shown in FIG. 1, the components related to the channel estimation of the present invention include a first FFT operator 100, a first pilot and data decomposer 110, and a pilot symbol-based LS channel estimator. Square-Channel Estimation; Pilot LS-CE (120), Scaled and Zero Composer (130), IFFT Operator (140), CIR Adapter (150), First 2 FFT operator 160, a second pilot & data decomposer 170, and a one-tap equalizer 180.

The first FFT operator 100 performs FFT on the OFDM symbol to demodulate the transmitted information based on the N orthogonal functions used in the orthogonal modulator on the transmitting side.

The first pilot and data separator 110 separates the pilot and data from the carrier symbols output from the FFT for subsequent processing, provided that the transmitter knows where to place the pilot and data.

Using the pilot symbol The LS channel estimator 120 estimates the channel value of the pilot position by the LS technique using the pilot symbols separated by the pilot and the data separator 110.

를 곱해 진폭을 조정한다. 즉, 스케일링한다. 여기서 N는 총 반송파의 개수이며, N_SP는 파일럿 반송파의 개수이다. 또한, 0은 원래 데이터 심볼을 위해 사용되는 위치에 삽입한다. 기능적으로 스케일러 및 제로 삽입기(130)는 앞서 언급한 파일롯 및 데이터 분리기(110)와 반대의 역할을 수행한다. Scaler and zero inserter 130 is a new feature proposed in the present invention, and serves to place the pilot channel estimate and zero from the pilot symbol using LS channel estimator 120 in an appropriate position. In this case, the pilot channel coefficients have the same energy even after the IFFT operation.

Multiply by to adjust the amplitude. That is, to scale. Where N is the total number of carriers and N _SP is the number of pilot carriers. Also, 0 is inserted at the position used for the original data symbol. Functionally, the scaler and zero inserter 130 play the opposite role to the aforementioned pilot and data separator 110.

An IFFT of the IFFT operator 140 may obtain an estimated value of the CIR. If the position of the CIR estimate is greater than the number of cyclic prefixes (CPs), N _CP , the estimate is zero.

The estimated value corresponding to the smaller or the same position may be expressed as a vector as follows.

에 진폭조정계수(scaling coefficient), α를 곱하여 CIR의 진폭을 보다 정확히 추정한다. 진폭조정계수는 다음의 수학식 2로 표현되며 몬테카를로 시뮬레이션을 통해 구한다. The CIR adaptor (CIR) handles two processes. First, the input vector of Equation 1

Multiply the amplitude scaling coefficient by α to more accurately estimate the amplitude of the CIR. The amplitude adjustment coefficient is expressed by the following Equation 2 and obtained through Monte Carlo simulation.

는 이상적인 CIR 값이고

는 CIR 적응기(150)가 처리하기 이전까지의 과정으로 구한 CIR 값이다. p 는 시스템의 샘플시간으로 정규화시킨 다중경로 페이딩 채널모델의 시간지연을 나타낸다.Where L is the number of paths or taps of the channel,

Is the ideal CIR value

Is the CIR value obtained by the process until the CIR adaptor 150 processes. p represents the time delay of the multipath fading channel model normalized to the sample time of the system.

Figure 2 it can be seen that a view illustrating an example of the amplitude adjustment coefficient for E _b / N _0, the amplitude adjustment coefficient is to have substantially the same value regardless of the E _b / N _0. The vector multiplied by the amplitude adjustment coefficient is expressed as follows.

Second, the CIR adaptor 150 compares the vector of Equation 3 with a threshold β, and makes the element value of the vector less than or equal to the threshold value zero. This process removes CIR values caused by Additive White Gaussian Noise (AWGN) or ICI. The threshold value is expressed by the following equation, and is obtained through Monte Carlo simulation.

는 CIR Adapter전까지의 과정으로 구한 CIR 값 중 다중경로 페이딩 채널모델의 시간지연에 해당하지 않는 값으로 이루어진 벡터이다. 즉, AWGN이나 ICI로 발생한 CIR 값들을 나타낸다. 이 벡터의 원소 중 제일 큰 값과 진폭조정계수를 곱한 후 평균을 취하여 문턱값을 구한다. ε는 임의의 값을 갖는 보정계수이다. 도면 3은 E_b/N₀에 대한 문턱값의 일예를 도시되어 있다. 모델 A는 수학식 4에서 ε = 0일 때의 결과이다. 이것을 토대로 모델 B와 모델 C를 근사화 하였다.here,

Is a vector consisting of values that do not correspond to the time delay of the multipath fading channel model among the CIR values obtained by the process before the CIR adapter. That is, CIR values generated by AWGN or ICI are shown. The threshold value is obtained by multiplying the largest value of the elements of this vector by the amplitude adjustment coefficient and taking the average. ε is a correction coefficient having an arbitrary value. 3 shows an example of a threshold for E _b / N ₀ . Model A is the result when ε = 0 in Equation 4. Based on this, models B and C are approximated.

 The output of the CIR adaptor 150 can be expressed as follows.

For the output from the CIR adaptor 150, the second FFT operator 160 converts the CIR values estimated in the time domain into channel estimates in the frequency domain, and converts the data symbols through the second pilot and data separator 170. Distinguish the channel estimate corresponding to the position. The single tap equalizer 180 equalizes the received data symbols with the channel estimates.

Simulation was performed to confirm the effect of the channel estimation technique according to the present invention. The symbols used in the above equations and OFDM system parameters, channel models, CIR adapter parameters, etc. used in the simulation for checking the performance of the channel estimation technique according to the present invention are summarized in Tables 1 to 3, and the simulation results 3 is shown in FIGS.

OFDM System Parameter Value N _SD : Number of data subcarriers 1365 N _SP : number of pilot subcarriers 683 N _CP : Number of cyclic prefixes 512 N : total number of subcarriers 2048 B : bandwidth 20 MHz T _S (= 1 / B ): sampling period 50 ns

HT (Hilly Terrain) channel model Tap number Relative time (us) T _S -normalized sample time Average relative power (dB) Doppler spectrum One 0.0 0  0.0 CLASS 2 0.2 4 -2.0 CLASS 3 0.4 8 -4.0 CLASS 4 0.6 12 -7.0 CLASS 5 15.0 300 -6.0 CLASS 6 17.2 344 -12.0 CLASS

CIR Adapter Parameter Value Amplitude Adjustment Coefficient (α) 1.73 Threshold (β) A 0.0708, 0.0497, 0.0362, 0.0252, 0.0180, 0.0131, 0.0100, 0.0076, 0.0063, 0.0053, 0.0049 B 0.08, 0.07, 0.06, 0.05, 0.04, 0.03, 0.02, 0.01, 0.01, 0.01, 0.01 C 0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0.01 0.01 D 0.01 E 0.08

3 to 6, PSA-CE (non-uniform) and PSA-CE (uniform) represent a frequency domain LMMSE channel estimation technique using a conventional pilot symbol. 'non-uniform' means autocorrelation function obtained through Monte Carlo simulation and 'uniform' means autocorrelation function obtained through mathematical model of uniform distribution. PSA-CE w / CIRA means a channel estimation technique according to the present invention, and A / B / C / D / E means a threshold shown in Table 3.

4 and 5 show MSE and BER performance, respectively. The performance of PSA-CE (non-uniform) is the best and the performance of PSA-CE (uniform) is the worst. PSA-CE w / CIRA shows graphs differently depending on the threshold used, and B is best. The results show that the proposed scheme is similar to the performance of PSA-CE (non-uniform).

6 is the BER performance when STO (10 samples) is present. In the case of PSA-CE (non-uniform), the performance decreases drastically because STO changes the characteristics of the autocorrelation function obtained by simulation. On the other hand, PSA-CE (uniform) and the proposed method are hardly affected by STO, so there is little change in performance.

Although the configuration of the present invention has been described in detail with reference to the preferred embodiments and the accompanying drawings, the present invention is not limited thereto, and various modifications and changes can be made, of course. For example, if the IFFT operator 140 IFFT outputs the output value of the pilot LS channel estimator 120 without using a scaler and a zero inserter to reduce the amount of computation, or inserts a zero after this value to perform IFFT. The CIR adaptation technique, which is a channel estimation technique of the present invention, can also be used.

In addition, each component illustrated in FIG. 1 may be implemented in separate hardware, but, alternatively, another component may perform the function of one component, and the functions of all the components in a single processor may be software. It can be obvious that it can be implemented.

Therefore, the protection scope of the present invention will be defined by the interpretation of the claims below.

As described above, the present invention can provide a strong channel estimation technique for multipath fading channels and STOs. In particular, the channel estimation method of the present invention uses only the memory for the amplitude adjustment coefficient and the threshold value, and therefore uses less memory of the system than the conventional method using the autocorrelation function, and adds only the CIR adaptor in the existing receiver configuration. This is easy to implement.

Claims (8)

In the channel estimation method of an OFDM communication system, Performing an FFT operation on the received OFDM signal; Estimating a channel preliminarily by using an LS technique with respect to a result value of the FFT operation step; Obtaining an estimated value of a channel impulse response (CIR) by performing an IFFT operation on the preliminary channel estimation result; Multiplying the estimated value by a predetermined amplitude adjustment coefficient; Comparing the result of the multiplication step with a predetermined threshold value; Improving the accuracy of the CIR estimate by converting the result of comparison between the comparison step and a result value less than or equal to a threshold value to 0; Converting the enhanced CIR estimate into a channel estimate by an FFT operation Channel estimation method of the OFDM communication system comprising a. The method according to claim 1, wherein the predetermined amplitude adjustment coefficient is

Where L is the number of paths or taps for the channel,

Is the ideal CIR value

Is the CIR value obtained by the process until the CIR adaptor 150 is processed, and P represents the time delay of the multipath fading channel model normalized to the sample time of the system.) Determined by. The method of claim 1, wherein the predetermined threshold β is

(here,

Is a vector value consisting of values that do not correspond to the time delay of the multipath fading channel among the CIR values obtained in the IFFT operation step, and ε is an arbitrary correction factor.) Determined by. The method of claim 1, wherein after the preliminary channel estimation step: Adjusting the amplitude such that the preliminary channel estimate has the same energy after the IFFT operation. delete In the channel estimation apparatus of an OFDM communication system, First FFT calculation means for performing an FFT operation on the received OFDM signal; Preliminary channel estimation means for preliminarily estimating a channel by an LS scheme using a pilot symbol with respect to a result value of the FFT operation step; IFFT calculation means for performing an IFFT operation on the preliminary channel estimation result to obtain an estimated value of a channel impulse response (CIR); Multiplication means for multiplying the estimated value by a predetermined amplitude adjustment coefficient; Comparison means for comparing the result of the multiplication step with a predetermined threshold value and converting a result value less than or equal to a threshold value to 0 to improve the accuracy of the CIR estimate; Second FFT computation means for converting the enhanced CIR estimate to a channel estimate by an FFT operation Channel estimator of an OFDM communication system comprising a. 7. The channel estimator of claim 6, further comprising amplitude adjusting means for adjusting the amplitude such that the preliminary channel estimate from the preliminary channel estimating means has the same energy even after the IFFT operation. delete

Images