Abstract

Broadband wireless communication requires fast and reliable synchronization and channel estimation with low overhead and computation complexity. In this work, we address the above issue by two aspects. First, we develop training signal design for not only achieving low complexity reliable receiver processing algorithms, but also improving the theoretical limit. Second, we develop low complexity high performance receiver algorithms for both single user and multiuser system.

We first investigate the optimal periodic preamble design for frequency offset estimation over frequency-selective fading channels by minimizing the average Cramer-Rao bound (CRB). The proposed design includes the optimal sequence structure design and the optimal signal design, and brings new dimensions into the existing theory of the training signal designs for frequency offset estimation.

To track the channel variations as well as the phase shift, data-pilot multiplexing scheme is commonly used in OFDM (orthogonal frequency division multiplex) systems. The two performance metrics which are not jointly incorporated yet in the existing data-pilot multiplexing scheme design are the number of data tones transmitted over one OFDM symbol (data throughput) and the channel response and phase estimation performance. In this work, we propose two multiplexing schemes which significantly improve the estimation performance with low data throughput sacrifice, resulting in a substantial performance improvement over current pilot-data multiplexed schemes.

In the uplink of OFDMA (orthogonal frequency division multiple access) systems, synchronization and channel estimation problems become more challenging since multiuser, multi-dimension parameters rather than single user parameters issue have to be estimated. The existing techniques have one or more of the following unfavorable requirements: high computation complexity, loss of frequency diversity, and performance sensitivity to the number of subscribe stations. To address the above drawbacks, we propose several training designs as well as the corresponding estimators for OFDMA uplink in this thesis, yielding robust low complexity estimators with enhanced performance.