Paper 4
Paper 4
Abstract—We propose a new algorithm for Doppler rate estima- the iterative procedure developed in [8] for joint estimation of
tion in burst-mode phase shift keying (PSK) transmissions. This phase, frequency and frequency-rate of an unmodulated carrier.
issue may arise in mobile radio links when the received signal ex- All these schemes achieve the Cramer–Rao lower bound at high
periences significant time-varying Doppler distortion, as in low-
earth-orbit satellite systems. The algorithm is based on the trans- SNR values, however, their estimates are plagued by large er-
mission of a training sequence and has a feedforward structure that rors (outliers) at SNRs lower than 10–15 dB. This makes the al-
is easy to implement in digital form. Its estimation accuracy is close gorithms unsuitable for many coded transmission systems. The
to the Cramer–Rao bound even at SNR values as low as 0 dB. Com- scheme proposed in [9] performs suboptimum ML-based esti-
parisons with earlier methods are discussed. mation of the Doppler rate. It gives good estimates provided that
Index Terms—Burst-mode transmission, Doppler rate, Doppler the Doppler shift is sufficiently small. Unfortunately, its perfor-
shift. mance is unsatisfactory even at Doppler shifts as low as 1% of
the symbol rate.
I. INTRODUCTION In this paper, we propose a new DA Doppler rate estima-
tion method that has some advantages over previous schemes. It
(2)
Paper approved by L. Vandendorpe, the Editor for Transmission Systems of
the IEEE Communications Society. Manuscript received June 14, 2000; revised where is statistically equivalent to . Thus,
April 15, 2001, and July 15, 2001. may be viewed as samples of an unmodulated carrier
The author is with the Department of Information Engineering, University of
Pisa, 56100 Pisa, Italy (e-mail: [email protected]). affected by Doppler distortion and embedded in white Gaussian
Publisher Item Identifier S 0090-6778(02)05117-6. noise.
0900-6778/02$17.00 © 2002 IEEE
708 IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 50, NO. 5, MAY 2002
In the sequel, the parameters , , and are assumed to be general formula , it can be shown
deterministic but unknown constants. Our goal is to estimate that
based on the observation of the samples . The phase
and the frequency are viewed as nuisance parameters.
In [6], it is shown that the Cramer–Rao bound (CRB) to the (10)
error variance of any unbiased estimator of the Doppler rate provided that and the are sufficiently small.
is given by Clearly, (10) relates to in a linear fashion and the
problem is to estimate a constant, , from the noisy mea-
(3) surements .
Letting , the ML estimator
Note that for large decays as while the CRB of is given by [11]
for Doppler shift decreases as [4].
(11)
III. ESTIMATION ALGORITHM
where is a -dimensional vector of all ones and
We begin by dividing the observed vector is the covariance matrix of . From (10), the
[the superscript indi- -entry of is found to be
cates vector transpose] into adjacent segments, each containing
samples. The samples within each segment are then summed
together to produce quantities
(12)
(4)
otherwise
(13)
(5)
with
where (14)
(7) Comparing the above result with (3), we see that for the
variance of attains the CRB.
can be made in (5). Correspondingly, we have The computational complexity of the proposed estimator can
be assessed as follows. Assume that the coefficients
have been precomputed and stored. Then, computing
(8) from requires real additions. Also, the right-hand
side (RHS) of (13) needs multiplications and ad-
Hence, denoting by the argument of and assuming
ditions. Thus, the total number of operations is .
high SNRs, from (8) we get
IV. SIMULATION RESULTS
(17)
710 IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 50, NO. 5, MAY 2002
V. CONCLUSION
We have proposed a new algorithm for Doppler rate estima-
tion in burst-mode transmissions. The algorithm is based on the
transmission of a training sequence and is suitable for PSK sig-
naling. Its accuracy attains the CRB at intermediate/high SNR
values. The estimation range is large and the threshold can be
reduced by increasing a parameter . Compared with other ex-
isting methods, the algorithm is simpler to implement and more
robust against Doppler shifts.
REFERENCES
[1] I. Ali and N. Al-Dhahir, “Doppler characterization for LEO satellites,”
IEEE Trans. Commun., vol. 46, pp. 309–313, Mar. 1998.
[2] “Special issue on communications in the intelligence transportation
system,” IEEE Commun Mag., vol. 34, Oct. 1996.
[3] A. W. Rihaczek, Principles of High Resolution Radar. New York: Mc-
Graw-Hill, 1969.
[4] M. Morelli and U. Mengali, “Feedforward frequency estimation for
PSK: A tutorial review,” Eur. Trans. Commun. Related Technol., vol. 9,
pp. 103–116, Mar./Apr. 1998.
[5] T. J. Abatzoglou, “Fast maximum likelihood joint estimation of fre-
quency and frequency rate,” IEEE Trans. Aerosp. Electron. Syst., vol.
22, pp. 708–715, Nov. 1996.
[6] P. M. Baggenstoss and S. M. Kay, “On estimating the angle parameters
of an exponential signal at high SNR,” IEEE Trans. Signal Processing,
vol. 39, pp. 1203–1205, May 1991.
Fig. 4. MSEE of the GLR estimator. [7] S. Peleg and B. Friedlander, “The discrete polynomial-phase transform,”
IEEE Trans. Signal Processing, vol. 43, pp. 1901–1914, Aug. 1995.
[8] M. Z. Ikram, K. A. Meraim, and Y. Hua, “Estimating the parameters of
Note that the RHS of (17) requires operations chirp signals: An iterative approach,” IEEE Trans. Signal Processing,
vol. 46, pp. 3436–3440, Dec. 1998.
(real additions plus multiplications). Thus, the computational [9] F. Giannetti, M. Luise, and R. Reggiannini, “Simple carrier frequency
complexity is with the GLR scheme while it is rate-of-change estimators,” IEEE Trans. Commun., vol. 47, pp.
with the proposed estimator. 1310–1314, Sept. 1999.
[10] U. Mengali and A. N. D’Andrea, Synchronization Techniques for Digital
Receivers. New York: Plenum, 1997.
[11] S. M. Kay, “A fast and accurate single frequency estimator,” IEEE Trans.
Acoustic, Speech, Signal Processing, vol. 37, pp. 1987–1990, Dec. 1989.