Carrier Synchronization of OFDM Systems over Dispersive Multipath Fading Channels
Date Issued
2005
Date
2005
Author(s)
Liao, Yi-Ching
DOI
zh-TW
Abstract
In recent years, the spectacular growth of video, voice and data communication over the Internet and the equally rapid pervasion of mobile telephony justify the great expectations for mobile multimedia. Research and development are taking place all over the world to define the next generation wireless broadband multimedia communications system that may create the global information village. While present communications systems are primarily designed for one specific application, such as speech on mobile telephony or high-rate data in wireless local area network (WLAN), the next generation wireless broadband multimedia communications system will integrate various functions and applications, besides, it is expected to provide its users with high speed transmissions
over 50 Mbps.
Supporting such large data rates with sufficient robustness to radio channel impairments requires careful choosing of modulation technique. The most promising choice seems to be orthogonal frequency division multiplexing (OFDM) by its immunity to multipath fading and impulse noise [1]. Despite its superiority, the principal weakness of OFDM is its sensitivity to synchronization errors, especially carrier frequency offset and phase noise caused by oscillator instability and/or Doppler shifts. This vulnerability is obviously the main obstacle to the implementation of a practical OFDM communications system. Thus, the task of carrier synchronization is of fundamental importance. To enable fast and robust synchronization, data-aided approaches are in general more attractive than blind approaches. Consequently, the goal of this dissertation is to study and develop effective data-aided carrier synchronization algorithms suitable for operation in dispersive multipath environment.
In this dissertation, we first investigate the statistical property of data-conveying
OFDM signals. Treating the data signal as an Gaussian random process, we show that
cyclic-prefix and pilot signals can be combined to yield a class of frequency offset estimators based on maximum likelihood criterion. The mean square error of the proposed estimators are analyzed to justify simulation. The issue of complexity reduction are also addressed.
Next, we concentrate on the frequency ambiguity resolution over frequency selective slowly Rayleigh fading channels. Approaches in time and frequency domain are studied and proposed respectively. The prior information of the statistic of the channel impulse response is exploited to eliminate the effect of channel selectivity on the estimation performance. The distinguishing features of our proposed methods are that only one preamble is required and the range of carrier frequency offset which can be handled is the overall transmission spectrum. Different from the intuition in signal detection, we find that higher channel selectivity does not degrade the estimation performance of our proposed methods but improve it by making the statistical characteristics of the channel more remarkable which demonstrates the superiority and practicability of our proposed methods.
In addition, the statistical property of Wiener phase noise is investigated and the second order statistics of the ICI introduced by phase noise is derived. This statistics are used to generate a common phase error estimator based on maximum likelihood criterion by using the pilot symbols and tentative decisions. The estimation scheme is shown to outperform conventional approaches as justified by simulations. To proceed, the statistical property of two stationary phase noise model is investigated and the second order statistics of the ICI introduced by phase noise are derived respectively. This statistics are used to generate a common phase error estimator based on maximum likelihood criterion by combining the pilot symbols and tentative decisions. The estimation scheme is shown to provide performance gain to the conventional approaches and is quite robust against possible model mismatch.
Furthermore, the multiple phase noise estimation problem is considered in uplink OFDMA systems. By deriving the unified matrix-form signal models, two multiple phase noise estimation algorithms based on least-square and maximum likelihood criterion are proposed respectively. Numerical results demonstrate that the proposed algorithms has a wide range of applicable phase noise level and practically implemented. Finally, the directions of future researches and conclusions of this dissertation are provided.
Subjects
正交分頻多工系統
載波同步
多路徑衰減通道
OFDM
Carrier Synchronization
Multipath Fading Channels
Type
thesis
File(s)![Thumbnail Image]()
Loading...
Name
ntu-94-F87942045-1.pdf
Size
23.31 KB
Format
Adobe PDF
Checksum
(MD5):0a080ef4708c0eb5447d5846af09c754