Documente Academic
Documente Profesional
Documente Cultură
Pushpa Kotipalli
ECE Dept., Shri Vishnu Engg. College for Women Bhimavaram, West Godavari District Andhra Pradesh, India - 534202 e-mail: pushpak@svecw.edu.in
Abstract Orthogonal frequency-division multiplexing (OFDM) is a well known multi-carrier digital modulation scheme in which a transmitted data is split into multiple streams that are modulated at different frequencies. The OFDM system is very sensitive to frame and frequency synchronization errors. In this paper, we propose a preamble for frame and frequency synchronization in OFDM systems. The preamble is generated by loading the BPSK data on alternate subcarriers in a symmetric manner. By using the proposed preamble, a frame synchronization method for OFDM systems is described. And, it is shown that the proposed frame synchronization method yields sharp peak at the frame boundary and is independent of carrier frequency offset. Also, we proposed preamble based fractional as well as integer frequency offset estimation techniques. These techniques do not require the prior knowledge of PN sequence or CP length used at the transmitter. Simulations are conducted for AWGN and frequency selective channels to demonstrate the performance of the proposed methods. Simulations show that the proposed frame and frequency synchroniation techniques yeild good results for SNR > 2dB. Keywords-OFDM; time synchronization; preamble and frame synchronization; Frequency synchronization; FFO; IFO.
by the transmitter dynamically [5]. The Schmidls synchronization method [6] uses a preamble containing the two identical halves. This method gives simple and robust estimates for symbol timing and carrier frequency offset. However, the timing metric of Schmidls method has a plateau, which causes large variance of the timing estimate. To reduce the uncertainty due to the timing metric, Minn proposed a method as modification to Schmidls [7]. The result of Minns preamble gives sharper timing metric and smaller variance than Schmidls. In this paper, we propose a preamble (in section II) which can be used to estimate not only frame boundary and fractional frequency offset (FFO), but also integer frequency offset (IFO). In section III, we propose methods for estimating frame boundary, FFO and IFO. Section IV shows simulation results and conclusions are placed in section V. II. OFDM SYSTEM SIGNAL MODEL AND MOTIVATION
A. OFDM System Signal Model We consider the following signal model for the OFDM system. Let the samples of the base-band equivalent OFDM signal, including cyclic prefix (CP), be expressed as
I.
INTRODUCTION
Orthogonal frequency division multiplexing systems allow a bandwidth-efficient transmission over strongly frequency selective wireless channels at a moderate implementation effort [1]. It is commonly known that OFDM systems are very sensitive to synchronization errors, which cause inter-channel interference (ICI) and inter-symbol interference (ISI) when the fast Fourier transform (FFT) window timing is not provided within the ISI-free part of the guard interval [2]. Also, frequency offset that is caused by Doppler shift or misalignment between frequency oscillators exists causing ICI which leads the system performance deteriorating drastically [3]. The existing frame and frequency synchronization techniques for packet OFDM can be broadly grouped into cyclic prefix (CP) based synchronization techniques and preamble based synchronization techniques. The CP based techniques [4] exploit the cyclic prefix information for the purpose of time and frequency synchronization. The main drawback of the CP based techniques is with respect to the unknown nature of the CP length as it may be determined
x( n) =
where N is the total number of carriers, X(k) is the kth subsymbol, and j = 1 , and L indicates the length of CP. The CP is appended at the beginning of OFDM symbol to eliminate ISI. Let the time domain samples corresponding to the preamble and data symbols be denoted by x p ( n) and xd ( n) , respectively. Then, the transmitted signal x(n) can be written as
1 N