2278 IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 18, NO. 11, NOVEMBER 2000 Timing Recovery for OFDM Transmission Baoguo Yang, Member, IEEE, Khaled Ben Letaief, Roger S. Cheng, Member, IEEE, and Zhigang Cao, Senior Member, IEEE Abstract—Orthogonal frequency division multiplexing (OFDM) is an effective modulation technique for high-rate and high-speed transmission over frequency selective fading channels. However, OFDM systems can be extremely sensitive and vulnerable to synchronization errors. In this paper, we present a scheme for performing timing recovery that includes symbol synchronization and sampling clock synchronization in OFDM systems. The scheme is based on pilot subcarriers. In the scheme, we use a path time delay estimation method to improve the accuracy of the correlation-based symbol synchronization methods, and use a delay-locked loop (DLL) to do the sampling clock synchronization. It is shown that by using this scheme, the mean square values of the symbol timing estimation error can be decreased by several orders of magnitude compared to the common correlation methods in both the AWGN and multipath fading channels. In addition, the scheme can track the symbol timing drift caused by the sampling clock frequency offsets. Index Terms—OFDM, sampling clock synchronization, symbol synchronization, timing recovery, wireless communication sys- tems. I. INTRODUCTION T HE NEXT-GENERATION wireless personal communi- cation systems are expected to provide ubiquitous, high- quality, and high-rate mobile multimedia transmission. How- ever, to achieve this objective, various technical challenges must be overcome [1]. For example, the deployment of broad-band wireless access systems would require a transmission technique which can mitigate the detrimental effects of frequency selec- tive fading [2]. In recent years, there has been a lot of interest in applying orthogonal frequency division multiplexing (OFDM) in wireless systems because of its various advantages in less- ening the severe effects of frequency selective fading [3], [4]. However, OFDM systems are vulnerable to synchronization er- rors [4]–[11]. For example, carrier frequency offsets, which are caused by the inherent instabilities of the transmitter and re- ceiver carrier frequency oscillators, can lead to severe system degradation due to intercarrier interference (ICI) [5]. Symbol timing synchronization must also be achieved in order to avoid intersymbol interference (ISI) [4], [8]–[11]. In OFDM transmission systems, the synchronization tasks include carrier frequency synchronization and timing recovery Manuscript received June 7, 1999; revised February 8, 2000. This work was supported in part by the Hong Kong Telecom Institute of Information Tech- nology and Research Grant Council. B. Yang and Z. Cao are with the State Key Laboratory on Microwave & Dig- ital Communications, Department of Electronic Engineering, Tsinghua Univer- sity, Beijing 100084, China. K. B. Letaief and R. S. Cheng are with the Center for Wireless Informa- tion Technology, Department of Electrical & Electronic Engineering, The Hong Kong University of Science & Technology, Clearwater Bay, Kowloon, Hong Kong (e-mail: eekhaled@ee.ust.hk). Publisher Item Identifier S 0733-8716(00)09205-2. that is further divided into symbol 1 synchronization and sampling clock synchronization. The purpose of symbol syn- chronization is to find the correct position of the fast Fourier transform (FFT) window. Symbol synchronization may be done at the receiver with the aid of the dedicated training symbols. The cyclic property of the guard interval preceding the OFDM symbol can be also evaluated for symbol synchronization, thus reducing the need for training symbols [9]. In multipath fading channels, however, the guard interval is corrupted by ISI and the periodic property is destroyed. Consequently, correct symbol synchronization cannot be guaranteed in the case of ISI [4], [9]–[11]. If the symbol timing error is outside of the ISI free range in the guard interval, the inaccurate symbol timing can cause ISI that destroys the orthogonality of the subcarriers and degrades the performance of OFDM systems [4]. In addition, the performance of the channel estimation via interpolation, commonly used for coherent OFDM systems, can be essentially degraded by the symbol timing errors [8], [12]. Hence, more accurate symbol timing synchronization methods are needed to satisfy the synchronization requirement in coherent OFDM systems. In contrast to the symbol synchronization case, the purpose of sampling clock synchronization is to align the receiver sam- pling clock frequency to that of the transmitter. The sampling clock frequency error can cause ICI [4], [6], [7]. Moreover, the sampling clock frequency error can result in a drift in the symbol timing and can further worsen the symbol synchronization prob- lems. For instance, if the sampling clock specification is 10 parts per million (ppm) and the sampling frequency is 5 MHz, then the symbol timing has a drift of about 50 samples per second. Thus, sampling clock synchronization is also an important issue which needs to be addressed in OFDM systems. In this paper, we propose a timing recovery scheme based on pilot subcarriers for OFDM systems. As pilot subcarriers are used in most coherent OFDM systems for synchronization and channel estimation purposes, our scheme can be implemented without additional overhead for these systems. In this scheme, we use the correlation method based on the guard interval to do the coarse symbol synchronization. A path time delay es- timation method is then employed to further improve the ac- curacy of the coarse symbol synchronization. Finally we use a delay-locked loop (DLL) to do the sampling clock synchro- nization and to maintain the symbol timing. We derive the DLL technique from the joint maximum-likelihood (ML) estimation of the symbol timing and carrier phase in the AWGN channel. We apply the derived algorithm to both AWGN and various mul- tipath fading channels and study their performance via simu- lation. Even though the DLL is not optimal for the multipath 1 A symbol is defined to be a block of FFT in this paper. 0733–8716/00$10.00 © 2000 IEEE