Code Acquisition and Symbol Timing Recovery


The 14th IEEE 2003 International Symposium on Persona1,lndoor and Mobile Radio Communication Proceedings

Code Acquisition and Symbol Timing Recovery Method in TDS-OFDM for Broadcast Channels Zhi-Xing Yang, Jun Wang, Chang-Yong Pan, Meng Han and Lin Yaiig Department of Electronic Engineering Tsinghua University Beijing 100084,China Wj99@ AD.str.cicr-This paper describes a novel code acquisition and symbol timing recovery method for Time Domain Synchronous OFDM (TDS-OFDM). It is based on the correlation peak of the PN sequence in the Frame Sync, in this way it is possible to decrease the influence from noise or interference. It is derived in the AWGN channel, but simulations show that it can also perform well in multipath channels. Key words-TDS-OFDM, code acquisition, symbol timing recovery, nirrltipath clrannel, H D TV.

1. 1


I. INTRODUCTION Recently, a new Chinese DTV territorial transfer scheme, Terrestrial Digital MultimediaiTeIevision Broadcasting (DMB-T). has been proposed by Tsinghua University, whose key technology is TDS-OFDM[l]. The transmitter of TDS-OFDM processes data with IDFT in the same way as cyclic prefix OFDM (CP-OFDM) which is used as DVB-T niodulation scheme[2], however, it inserts pseudonoise (PN) sequences as the guard intervals, which also serve as the training symbols. The combination of the guard intervals and the training symbols can not only reduce transmission overhead but also separate the synchronization from data streams, this separation makes the synchronization process do not depend on the exact modulation used for the data and increase the modularity of the entire system. After removing the PN sequences at the receiver, TDS-OFDM is equivalent to the zero-padded OFDM (ZP-OFDM)[3], it is more sensitive to the symbol timing error than CP-OFDM. Hence, more accurate symbol timing synchronization methods are needed for TDS-OFDM. In[4], a time synchronization for packet based OFDM is analyzed using PN sequence preambles. In this paper we present a simple and efficient algorithm for combined code acquisition (CA) and symbol timing recovery (STR) for TDS-OFDM based on continuous transnlission of the PN sequences. This paper is organized as follows. The TDS-OFDM signal frame format and system model are described in section 2. In section 3, we present the proposed combined method, and analyze the tracking error variance of the STR. Section 4 is the discus'sion of the simulation results on the AWGN channel and two kinds of multipath channels. The last section is our conclusions.

In the downlink physical channel of TDS-OFDM, signals are transmitted in frames. The Signal Frame consists of two parts, the Frame Sync and the Frame Body, which is shown in Fig.1. The Frame Sync consists of a pre-amble, a PN sequence with 255 symbols (N=255), and a post-amble. BPSK modulation is used for robust synchronization. The PN sequence is generated based on a set of shifted m-sequence. Different

signal frame has different PN sequence in order to guarantee signal frame addressing. The pre-amble and post-amble are cyclical extensions of the PN sequence. the length ofpre-amble can be defined as 0, 24, and 25, and the post-amble can be defined as 1,25, and 104, dependent on the maximum delay spread of the channel. In the simulation below the pre-able has 25 symbols, and post-amble has 104 symbols. OFDM modulation scheme is used for Frame Body, the IDFT block has 3780 symbols. Fig.2 shows the baseband transmitter and receiver structures of the TDS-OFDM system. The symbol rate 1/ T is same for both Frame Sync and Frame Body, which is 7.56 MSPS. In order to limit the bandwidth of transmitted signal to SMHz, a square root raised cosine (SRRC) filter is used for pulse shaping. Denote p ( t ) the impulse response of the SRRC filter. The transmitted baseband signal~ ( can be expressed t ) as

s ( t )= p ( t ) *[PN(k)+IDFT(s(k))] where


* denotes convolution.

At the receiver, CA and STR are first executed, then the PN sequence is removed after SRRC filtering. Finally the demodulated data s'(k) is obtained by discrete Fourier transform (DFT). In the following section, the CA and STR method are discussed in more details.





Originally, the local PN code generator has no information about the phase of the PN sequence received. Therefore, CA is performed first to capture this phase. Once CA is finished, timing tracking for continuous data streams is performed by a STR feedback loop consisting of a timing error detector, a loop filter, and a digital interpolator. Since the stability of the

0-7803-7822-9/03/$17.0002003 IEEE.


Code Acquisition and Symbol Timing Recovery

Code Acquisition and Symbol Timing Recovery相关文档