Master Thesis: Phase Error Tracking and Compensation for Packet-Based MIMO-OFDM Systems

Student(s): Ziegler, Markus

Semester: Fall semester 2008

Time: Apr. 2008 - Oct. 2008

Supervisor(s): Dr. Ulrich Schuster, Dr. Moritz Borgmann

Abstract:
This thesis addresses the problem of tracking and compensating for the Residual Frequency Offset (RFO) and phase noise process in packet-based multiple-input multiple-output (MIMO) orthogonal frequency division multiplexing (OFDM) systems. As a promising candidate for future high data rate wireless communication standard, MIMO-OFDM has recently attracted a large amount of attention in both commercial and academic settings. One disadvantage shared by both single-antenna and MIMO OFDM systems is their sen- sitivity to phase noise and frequency synchronization errors. Compensating for the RFO and phase noise, two types of impairments which cause significant performance degrada- tions in practical MIMO-OFDM systems, requires continuous phase error estimation and compensation. We restrict our attention to pilot-based compensation algorithms, as they tend to be simpler to implement and more suitable for packet-based OFDM systems than decision-directed compensation algorithms. While many different pilot-based phase error correction algorithms have been proposed that address either phase noise or RFO, fewer algorithms have been proposed for pilot- based compensation of both phase noise and RFO. Many of the proposed algorithms either suffer from excessive computational requirements which make them unpractical choices for low-cost MIMO-OFDM implementations or they are ad hoc solutions and lack a solid theoretical underpinning. We introduce a state-space system model that describes the effect of phase errors on the received OFDM signals. We use this system model to express phase error correction as a stochastic control problem and derive a solution using results from stochastic control theory. The resulting phase error correction algorithm consists of a linear regulator applied to an estimate of the phase error computed by an Extended Kalman Filter (EKF). We show that our phase error correction algorithm can be implemented efficiently and that it its computational requirements are lower than those of typical EKF implementations. Simulations demonstrate that our phase error correction algorithm performs well in a variety of conditions and for several different phase noise models. Our algorithm is adaptive and achieves both fast RFO acquisition as well as stable RFO tracking, a combination that is not possible with fixed-gain FIR filters. Finally, the system model we developed can be used as a starting point for the derivation of other error phase error correction algorithms.