International Journal of Computer Applications (0975 – 8887) Volume 91 – No.14, April 2014 A Comparative Study of Channel Estimation Techniques for OFDM in DVB-T2 Ahmed H.Eldieb Mona Z.Saleh Salwa Elramly Electronics and Electronics and Electronics and Communications Eng. Dept., Communications Eng. Dept., Communications Eng. Dept., Faculty of Engineering, Faculty of Engineering, Higher Institute of Engineering, Modern Academy Ain Shams University Ain Shams University Cairo, Egypt Cairo, Egypt Cairo, Egypt ABSTRACT At the OFDM system transmitter side, the binary data are Orthogonal Frequency Division Multiplexing (OFDM) with initially grouped and mapped to signal constellation points, pilot symbol assisted channel estimation is a well-known such as QPSK or 16-QAM. Then, the pilots are inserted either multicarrier technique for high rate transmission over wireless as a comb type, block type or a compromise of both (e.g. frequency selective fading channels. In this paper a scattered pilots). The data sequence is then transformed into comparative study of channel estimation techniques is time domain through IFFT. A guard interval is inserted to presented based on a pilot arrangement used in the second prevent inter-symbol interference. The signal is finally applied generation for Digital Video Broadcasting Terrestrial (DVB- to a DAC for transmission on channel [3]. T2) system in time-varying frequency-selective fading Channeleffect, which is caused by the channel characteristics channels. A modified method based on Domain Transform such as attenuation distortion, delay distortion and noise, is Least Square Estimation (DTLSE) has been proposed which compensated by channel estimation at the receiver side. Chan- reduces the computational complexity by avoiding the use of nel estimation is usually performed by transmitting training the matrix inversion .The estimation matrix size for obtaining pilot symbols on sub-carriers these pilots are known to the Channel Impulse Response (CIR) depends only on the length receiver. The system performance and computational of the channel not on the number of pilot subcarriers nor the complexity are twoimportant requirements for selecting the size of OFDM symbol. Two Dimensional Linear Interpolation optimum technique of the channel estimation [4]. (2-DLI) is discussed which performs better than one Dimensional Linear Interpolation (1-DLI)but suffers from Channel estimation in frequency-domain pilot assistedOFDM long processing delay and higher sensitivity to Doppler Systems can be obtained by employing theleast Squares frequency shifts. The Minimum Mean Squared Error (MMSE) (LS)estimate at pilot sub-channels position. Then, the channel estimation is also considered as a reference method complete Channel Transfer Function (CTF) at allsubcarriers because it has a good performance close to the perfect are calculated from interpolation either in frequency-direction estimation at the cost ofhigh computational complexity and (1-D) only or both time and frequency directions (2-D).The the necessity of prior knowledge ofchannel statistics. The (2-D) performs better than (1-D) interpolation at the cost of proposed method is compared with three other methods higher sensitivity to Doppler frequency shifts and long DTLSE, 2-DLI and MMSE.The proposed method provesto processing delay [5]. give a better performance compared to the other methods The MMSE estimation [6] offers the best Bit Error Rate(BER) which suffer from either a much higher computational load or but has heavy computational load and needs knowledge of less robustness to Doppler shifts. channel statistics. The least square estimation at pilot sub channels can also be employed to estimate time-domain CIR General Terms and then transform it to CTF, which is known as the DTLSE Channel estimation, wireless systems. [7]. Keywords The proposed estimation method needs a single OFDM Digital video broadcasting system for terrestrial second symbol for getting channel impulse response instead of generation (DVB-T2),domain transform least square estimation multiple ones. It also works for any frequency pilot pattern (DTLSE),two dimensional linear interpolation (2- without needing prior knowledge of channel statistics while in DLI),minimum mean square estimation (MMSE). the same time considerably reducing the computational load through avoiding the usage of matrix inversion. It is also 1. INTRODUCTION worth mentioning that the size of the estimation matrix for OFDM provides an effective and low complexity means of obtaining CIR depends only on the length of the channel not reducing intersymbol interference for transmissions over the number of pilot subcarriers or the size of OFDM symbol. frequency-selective fading channels. It has been widely applied in wireless communication systems due to its high bandwidth efficiency, high transmission capability and its The paper is organized as follows; the system description and robustness against multipath delay [1]. OFDM has been principle of channel estimation are explained in section (2); adopted by several wireless systems and standards such as performance analysis of the modified method compared with WLAN IEEE802.11a/n, 4G LTE, WiMAX IEEE802.16d/e, other methods is presented in section (3); computational load Digital Audio Broadcasting (DAB), Terrestrial Digital Video of the considered methods are calculated in section (4); Broadcasting DVB-T and DVB-T2 [2]. thesimulation results are presented in section (5) while section (6) concludes the paper. 21 International Journal of Computer Applications (0975 – 8887) Volume 91 – No.14, April 2014 2. SYSTEM DESCRIPTION In this method [5], the CFR at the pilot positions are calcul- 푡ℎ 푡ℎ Within a typical OFDM system receiver, after removing the atedat the푘푝 pilotsubcarrier and 푛푃 OFDM symbol as guard interval, demodulation is performed by FFT. When the length of guard interval is longer than the length of channel YnP ,kP WnP ,kP H = = H + 3 impulse response, ISI could be eliminated. In this case cha- np ,kp np ,kp XnP ,kP XnP ,kP nnel can be viewed as a set of parallel subchannels and the received signal isrepresented as The CFR values (H n,k) at the rest of the subcarriers are calculated from the estimation via interpolation which is Y푘 = H푘 X푘 + W푘 푘 = 0, … . , K − 1 (1) applied in both frequency direction (across subcarriers) and time direction (across successive OFDM symbols).At first, where Yk represents the received signal in complex interpolation has to be performed in time direction then baseband, Xk denotes the transmitted signal, Hk and Wk are interpolation is performed in frequency direction to obtain the Channel Frequency Response (CFR) and the additive channel coefficients for the remaining data subcarriers.The white Gaussian noise, respectively, K is the total number of most common interpolation method is the linear Interpolation subcarriers and k is the index of subcarrier within an OFDM (LI) because of its simplicity and low computational symbol [8], [9]. complexity. After interpolation of channel coefficients at all Channel estimation with the aid of pilots in OFDM systems is data subcarriers the complex received signal is equalized as performed by estimating the CFR at the pilot locations. This is 푌 obtained by comparing the received pilot subcarriers with the 푋 = 푛,푘 (4) 푛,푘 퐻 transmitted one which are known to the receiver. Let퐾푃be the 푛,푘 total number of pilots and푘푃denote the index of the pilot subcarrier,the estimation of CFR at a pilot position can be where Y푛,푘 is the complex received signal FFT outputs, H 푛,푘 obtained by the LS method as denotes the estimated and interpolated CFR at all indices and X 푛,푘 is the estimated transmitted signal. This method 푌푘 W푘 isconsidered a 2-DLI which suffers from long processing H = 푃 = H + 푃 (2) 푘푃 푘푃 delay andmore sensitivity to Doppler frequency shifts [5]. X푘푃 X푘푃 3.2 The MMSE Estimation Method whereYk denotes the received pilot subcarrier, Xk is the P P The MMSE estimation [6], provides the best BER transmitted pilot subcarrier, and, H is the estimation of CFR kP performance. It estimates CTF H as at the pilot positions. Then, the CFR values at the rest of the mmse subcarriers are calculated from the estimated ones via −1/2 −1/2 H = R . R . R . H (5) interpolation [10].In DVB-T2 system for pilot pattern (PP1), mmse 푘,푘푝 푘푝 ,푘푝 푘푝 ,푘푝 푘푝 ,ls the scattered pilots used for channel estimation are spread in both time and frequency directions. As shown in Figure 1, whereR푘,푘푃 of size 퐾×퐾푃is the correlation matrix of complete there is one pilot subcarrier out of three subcarriers in the CTF with pilot subchannels, and R푘푃 ,푘푃 of size 퐾푃×퐾푃is the frequency directionand one pilot out of four symbols in time correlation matrix of pilot subchannels with its LS direction. Thus the pilot density for this pilot pattern is estimate H푘 ,ls .This method suffers from heavy computatio- D = 3 and D = 4 plus edge pilots [11]. 푝 x y nal load and needsPrior knowledge of channel statistics. 3.3 DTLSE Estimation Method In this method [7], the least square estimation of pilot subchannels is employed to estimate time-domain channel im- pulse response CIR and then transform it to CTF. The CIR is then given as [7]: h = (FH F + αI )−1. FH . H (6) kp 푘푃 ,푘푃 푘푃 ,푘푃 푘푃 ,푘푃 푘푃 ,푘푃 푘푃 where F푘푃 ,푘푃 is a 퐾푃×퐾푃 Fourier transform coefficientsmat- H rix, (.) denotes Hermitian transpose, I is the 퐾푃×퐾푃identity matrix and α is a regularization parameter [12]. Fig 1: Scattered pilot pattern (PP1) [2]. The whole effective CTF is estimated by applying N-point FFT on the zero padded estimated effective CIR. In this In the following section the performance analysis of a modif- method the CIR depends only on the number of pilot ied method compared with other methods will be introduced. subcarriers, uses one OFDM symbol, works for any frequency pilot pattern and does not require any knowledge of channel 3.