226 IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 54, NO. 2, FEBRUARY 2006
Mathematical Analysis of the Impact of Timing Synchronization
Errors on the Performance of an OFDM System
Yasamin Mostofi, Member, IEEE, and Donald C. Cox, Fellow, IEEE
Abstract—This letter addresses the effect of timing synchro-
nization errors that are introduced by an erroneous detection
of the start of an orthogonal frequency-division multiplexing
(OFDM) symbol.
1
Such errors degrade the performance of an
OFDM receiver by introducing intercarrier interference (ICI) and
intersymbol interference (ISI). They can occur due to either an
erroneous initial frame synchronization or a change in the power
delay profile of the channel. In this letter, we provide a mathemat-
ical analysis of the effect of timing errors on the performance of
an OFDM receiver in a frequency-selective fading environment.
2
We find exact formulas for the power of interference terms and
the resulting average signal-to-interference ratio. We further
extend the analysis to the subsample level. Our results show the
nonsymmetric effect of timing errors on the performance of an
OFDM system. Finally, simulation results confirm the analysis.
3
Index Terms—Orthogonal frequency-division multiplexing
(OFDM), signal-to-interference ratio (SIR), timing-synchroniza-
tion errors.
I. INTRODUCTION
O
RTHOGONAL frequency-division multiplexing
(OFDM) divides the given bandwidth into narrow
subchannels. It handles delay spread by sending low data rates
in parallel on these subchannels [1]. By adding a guard interval
to the beginning of each OFDM symbol, the effect of delay
spread (provided that there is perfect synchronization) would
appear as a multiplication in the frequency domain. Adding
the guard interval will also prevent intersymbol interference
(ISI).
4
Timing-synchronization errors, however, degrade the
performance of an OFDM receiver by introducing intercarrier
interference (ICI) and ISI. Several methods have been proposed
for timing synchronization in OFDM receivers [2]–[6]. To
evaluate and improve the performance of these methods, a
comprehensive mathematical analysis of the effect of timing
errors and the underlying interference terms is necessary.
Paper approved by R. De Gaudenzi, the Editor for Synchronization and
CDMA of the IEEE Communications Society. Manuscript received July 30,
2003; revised July 27, 2005. This paper was presented in part at the 37th
Asilomar Conference on Signals, Systems, and Computers, Monterey, CA,
November 2003.
Y. Mostofi is with the California Institute of Technology, Pasadena, CA 91125
USA (e-mail: yasi@cds.caltech.edu).
D. C. Cox is with the Department of Electrical Engineering, Stanford Uni-
versity, Stanford, CA 94305 USA (e-mail: dcox@spark.stanford.edu).
Digital Object Identifier 10.1109/TCOMM.2005.861675
1
Throughout this letter, the term “timing error” would refer to this type of
error.
2
The analysis presented in this letter is for the case that no equalization tech-
nique has been used to mitigate the introduced ICI and ISI.
3
The results of this letter can be easily extended to address the effect of such
errors on DMT modems.
4
Note that intersymbol interference refers to inter-OFDM symbol-interfer-
ence.
Fig. 1. Discrete baseband equivalent model.
Authors in [5] have provided an approximated formula with lim-
ited applications for the interference caused by timing errors. It
is the goal of this letter to provide an exact mathematical anal-
ysis of the effect of timing errors, which can be a base for eval-
uating the performance of different synchronization methods.
Furthermore, while most of the work on timing synchronization
use sampling-period-level modeling, we also show how to ana-
lyze and evaluate the performance on the subsample level.
II. E
FFECT OF TIMING SYNCHRONIZATION ERRORS
(SAMPLING-PERIOD LEVEL)
Consider an OFDM system, shown in Fig. 1, in which the
available bandwidth is divided into
subchannels, and the
guard interval spans
sampling periods. Let represent the
length of an OFDM symbol (including the guard interval). Then
is the sampling period. In this section, we keep the
analysis on the sampling-period level, which translates to the
following assumptions:
1) no oversampling is done in the receiver;
2) timing error is a multiple of
.
We relax these conditions in Section III, where we extend the
analysis to include oversampling.
A. System Model
represents the transmitted data in the th frequency sub-
band and is related to the time-domain sequence
as follows:
for . and repre-
sent the transmitted and received cyclic prefixes (CPs), respec-
tively.
represents the th channel tap with Rayleigh fading
amplitude and uniformly distributed phase, and
is additive
white Gaussian noise (AWGN). Let
represent the length
of the channel delay spread normalized by the sampling period.
Then the channel would have
taps. In the absence of timing
0090-6778/$20.00 © 2006 IEEE