Abstract:
Orthogonal Frequency Division Multiplexing (OFDM) has attracted tremendous attention
in the last few years particularly in applications that require high data rates to be
supported under fading channels. OFDM has an inherent resistance to multi path
components and hence offers critical simplifications in its corresponding single carrier
based systems. However, the price of these benefits is paid in the synchronization
operations that introduce considerable degradation in performance if not countered
appropriately. Timing Synchronization requires the precise location of FFT window to be
determined. In addition, the orthogonality of the carriers is frequently disturbed by the
fading channel. If this orthogonality is not restored, it introduces Inter Carrier
Interference (ICI) and performance is severely compromised.
This work analyses some of the algorithms currently being used to achieve timing and
frequency synchronization. These algorithms are extended to high mobility cases where
Doppler shift in frequency is another irritant that brings additional fading and
degradation. The additional fading produced by the Doppler shift results in a residual
error even after a fine frequency estimation scheme. This work proposes a PLL based
recovery method which removes this proportional phase shift iteratively for all samples in
an OFDM symbol. We determine the imaginary component that remains after a QPSK
data point is raised to a power of 4 and this imaginary component is the correction
provided to a loop filter that generates the necessary phase correction for the next sample.
We propose a new system level diagram that removes the fine frequency estimation
procedure in Minn and Bhargava’s estimator and replaces it with our phase recovery
scheme. Both systems have been tested under severely degraded channels and our system
is found to outperform the one that employs fine frequency estimation. The proposed
system, therefore, offers possibilities to support high data rates with a low error
probability in a burst mode and at the same time lowers the computational complexity in
the overall system.
