Hopping Pilots For Estimation of Frequency-Offset and Multi-Antenna Channels in MIMO OFDM
Hopping Pilots For Estimation of Frequency-Offset and Multi-Antenna Channels in MIMO OFDM
Hopping Pilots For Estimation of Frequency-Offset and Multi-Antenna Channels in MIMO OFDM
Abstract— We design pilot symbol assisted modulation for to accommodate any space-time coded transmission.
carrier frequency offset (CFO) and channel estimation in or- Notation: Upper (lower) bold face letters will indicate matri-
thogonal frequency division multiplexing (OFDM) transmissions ces (column vectors). Superscript (·)H will denote Hermitian,
over multi-input multi-output (MIMO) frequency-selective fading
channels. By separating CFO and channel estimation from sym- (·)T transpose, and · will stand for the nearest integer.
bol detection, the novel training patterns lead to low-complexity The real and imaginary parts are denoted as [·] and [·];
CFO and channel estimators. The performance of our algorithms E[·] will stand for expectation and diag[x] for a diagonal
is investigated analytically, and then compared with an existing matrix with x on its main diagonal. Matrix DN (h) with
approach by simulations. a vector argument will denote an N × N diagonal matrix
with DN (h) = diag[h]. For a vector, · denotes the
I. I NTRODUCTION
Euclidean norm. We will use [A]k,m to denote the (k, m)th
Orthogonal frequency division multiplexing (OFDM) has entry of a matrix A, and [x]m for the mth entry of the column
been adopted by many standards (e.g., IEEE802.11a, vector x; IN will denote the N × N identity matrix; ei the
IEEE802.11g in the US and DAB/DVB, HiperLAN/2 in (i + 1)-st column of IN ; [FN ]m,n = N (1/2) exp(−j2πmn/N )
Europe), because it offers high data-rates and low decoding the N × N fast fourier transform (FFT) matrix. We define
complexity [6]. On the other hand, space-time multiplexing fN (ω) := [1, exp(jω), . . . , exp(j(N − 1)ω)]T .
and multi-antenna transmissions over multi-input multi-output
(MIMO) channels, has been recently proved effective in II. S YSTEM M ODEL
combating fading, and enhancing data rates; see [1], [5] and Let us consider the discrete-time equivalent baseband model
references therein. Therefore, MIMO-OFDM has got much of a block transmission system communicating over MIMO
attention. Implementing MIMO-OFDM however, faces two frequency-selective channels in the presence of CFO, shown in
major challenges: i) with the number of antennas increasing, Figure 1. Every information symbol block, [s(k)]n = s(kNs +
channel estimation becomes more challenging as the number n), is drawn from a finite alphabet. At the transmitter, each
of unknowns to be estimated increases accordingly; and ii) block s(k) is first encoded and/or multiplexed in space and
similar to single-antenna OFDM, MIMO-OFDM exhibits great time, to yield blocks {cµ (k)}N µ=1 of length Nc , which are
t
Coder
cNt Insertion of
uN t
uN t
1
Estimated CFO
null & training
0
x1 y1 z1
S/P R cp FFT Space-Time ŝ −1
Decoder
−2
xNr y Nr zN r
CFO & Channel
S/P R cp FFT Estimator −3
−3 −2 −1 0 1 2 3
True CFO
0 k=0 0
10
−2
0 k =1 10
0 k=2 10
−6
M=L+1
M=K
−8
10 M=2K
−12
10
0 2 4 6 8 10 12 14 16 18 20
inserting training symbols orthogonally per transmission block SNR
R EFERENCES
Channel MMSE
[2] H. Liu and U. Tureli, “A high efficiency carrier estimator for OFDM
communications,” IEEE Commun. Letters, vol. 2, pp. 104-106, Apr. 1998.
[3] M. Morelli and U. Mengali, “Carrier-frequency estimation for transmis-
sions over selective channels,” IEEE Trans. on Commun., vol. 48, no. 9 −2
10
pp. 1580-1589, Sep. 2000. 0 2 4 6 8 10
SNR
12 14 16 18 20
tion: ESPRIT,” IEEE Trans. on Commun., vol. 48, pp. 1459-1461, Sept.
2000.
[8] X. Ma, C. Tepedelenlioglu, G. B. Giannakis and S. Barbarossa, “Non- M=L+1 (w/ comp)
M=K (w/ comp)
M=2K (w/ comp)
data-aided carrier offset estimations for OFDM with null subcarriers: −2
10
M=3K (w/ comp)
M=L+1 (w/o comp)
Identifiability, Algorithms, and Performance,” IEEE Trans. on Commun., M=K (w/o comp)
M=2K (w/o comp)
vol. 19, no. 12, pp. 2504-2515, Dec. 2001. M=3K (w/o comp)
Ideal (perfect channel and CFO)
[9] X. Ma, L. Yang and G. B. Giannakis, “Optimal training for MIMO
frequency-selective fading channels,” Proc. of 36th Asilomar Conf. on −3
10
Signals, Systems, and Computers, Pacific Grove, CA, Nov. 3-6, 2002. 0 2 4 6 8 10
SNR
12 14 16 18 20