1

An Expurgated Union Bound of Space-Time Turbo

Codes in Quasi-Static Rayleigh Fading Channels
L. G. Caldeira1 and C. Pimentel2
1
Department of Electrical Engineering and Telecommunications, Federal Institute of
Education Science and Technology of Paraiba
58015-430, João Pessoa - PB, Brazil, e-mail: guedes@ifpb.edu.br
2
Department of Electronics and Systems, Federal University of Pernambuco
P.O. Box 7800, Recife-PE, 50711-970, Brazil, email: cecilio@ufpe.br

Index Terms

Distance spectrum, error event, frame error rate, Rayleigh fading, turbo codes, weight enumerator.
Abstract

This paper presents a new technique to obtain an expurgated union bound on the frame error rate of space-time
turbo codes (STTuCs) on quasi-static Rayleigh fading channels. We used two algorithms, one to take into account
the effect of drilling in the super state diagram (SSD) and another to calculate the spectrum of distances. Both
algorithms are completely matrix, and manipulate iteratively the entries of the SSD’s matrix which is convenient for
symbolic computation. Then, a new method to identify the dominant quasi-static fading error events is proposed.
Comparisons with simulated results reveal that the expurgated union bound on the frame error rate (obtained using
the limiting before average technique) using these dominant error events is tight.
We consider a communication system that employs two identical space-time trellis codes (STTCs) [1], con-
nected in parallel. The channel is modeled as a quasi-static flat Rayleigh fading channel with additive white Gaussian
noise. The fading coefficients among transmit and receive antennas during each frame are independent complex
Gaussian random variables. Figure 1 shows a transmission scheme of a general STTuC used in this work, which
employs two antennas. In the example, the STTuC consists of two identical encoders, called STTC1 STTC2 in its
recursive form, with QPSK modulation. Both STTCs receive bits of information, the STTC1 receives it directly
and STTC2 a permuted version through a symbol uniform interleaver, denominated π, with length of K symbols.
In fact, π is divided into two interleavers, of length K/2 symbols each, where one interleaves symbols in the even
positions and the other interleaves symbols of the odd positions.
 2

∨ ant. 1
S= [00, 33, 31, 13, 01, 22]
input bits
S1 = [00, 10, 31, 23, 01, 32]
∨ ant. 2
{00, 01, 11, 10, 00, 11}
-
- STTC1 -
@
I
@
? @@
I
π
S2 = [30, 33, 02, 13, 00, 22]

-
{11, 11, 00, 11, 00, 10}
- STTC2 -

Fig. 1. STTuC scheme with two STTCs, QPSK and 2 transmit antennas, STTC outputs S1 and S2 , and the punctured output sequence S.

The performance analysis of STTCs is the union bound on the first error event probability, namely Pfe (e),
which is closely related to the frame error rate (FER) and the bit error rate (BER) [2], [3]. Assuming a maximum
likelihood receiver with ideal channel state information, the modified conditional first error event probability is
given by [3]: v 
u X
X u γt nR
Pfe|Ωn1 R (e) ' hi Qt Ωj Ai Ω?j , (1)
2 j=1
S

where e and c are the erroneous and correct sequences, respectively, Ωn1 R = [Ω1 , · · · , ΩnR ], Ωj = [α1,j , · · · , αnT ,j ]
is a row vector of fading coefficients, A(c, e) is the Hermitian signal matrix [1] of simple error events, denoted
by SMEE, γt = Es /N0 is the signal to noise ratio per transmit antenna, and the superscript ? denotes conjugate
transposition. For a simple error event with a length of ` trellis intervals, two transmit antennas and Nc transmitted
P`−1 PNc i
symbols per antenna, the (i, j)th entry of the 2×2 SMEE is t=0 k=1 (ctk −eitk )(cjtk −ejtk )? , where cit1 , · · · , citNc
are the complex symbols transmitted from antenna i at trellis interval t chosen from a constellation with average
energy 1. Since the computation of the SMEEs becomes infeasible for long simple error events and the union
bound may not converge as the number of SMEEs is increased (the bound may be larger than one for a particular
fading realization [3]), a tight approximation to the union bound (called expurgated union bound) is obtained by
considering a few most dominant SMEEs specified by a set S in summation of (1). Let Ai be the ith SMEE in
S with average multiplicity hi . The distance spectrum of order N of a STTC is the enumeration of the ordered
set of N -pairs L = {(hi , Ai )}. Aktas and Fitz [4] proposed an ordering of the set L with the first n SMEEs with
minor determinants after an exhaustive search in the trellis. To order the set L, we use the criterion of statistical
significance of SMEEs after we use a matrix algorithm for calculating the distance spectrum, as proposed in [5].
After applying the matrix algorithms in SSDs of each STTC, we obtain an enumerator function T1 (Ai , W0 )
and T2 (Ai , WE ), where WO and WE correspond to Hamming distances enumerating of odd and even positions
due to STTC1 and STTC2 respectively. To take into account the interleaver effect we must consider that this is of
 3

the type uniform [6] and calculate the STTuC’s enumerating function , namely T ST T uC (Ai ), given by [7]:
X T1 (Ai , WO )T2 (Ai , WE )
T ST T uC (Ai ) ∼
=     , (2)
WO +WE K/2 K/2
   
WO WE

This work was considered an enumeration of the set L for Hamming distances (w0 + wE ) in the summation of (2)
with a maximum value of 5. Figure 2 shows the expurgated FER bound versus the signal to noise ratio per received
antenna in quasi-static Rayleigh fading channel, for the STTuC in [8] with 8 states and 8-PSK modulation. The
expurgated FER bound has good agreement with simulation results (dashed curves).

0
10
Simulated
Bound

−1
10
FER

−2
10

−3
10

−4
10
6 8 10 12 14 16
SNR (dB)

Fig. 2. FER versus SNR of STTuC [8], 8 states, 8PSK, 3 bits/s/Hz, 2 receive antennas, Rayleigh channel.

R EFERENCES

[1] V. Tarokh, N. Seshadri, A. R. Calderbank, “Space-time codes for high data rate wireless communication: Performance criteria and code
construction,” IEEE Trans. Inform. Theory, vol. 44, pp. 744-764, Mar. 1998.
[2] A. Stefanov and T. M. Duman, “Performance bounds for space-time trellis codes,” IEEE Trans. Inform. Theory, vol. 49, pp. 2134-2140,
Sept. 2003.
[3] A. P. des Rosiers and P. H. Siegel, “On performance bounds for space-time codes on fading channels,” IEEE Trans. Commun., vol.
52, pp. 1688-1697, Oct. 2004.
[4] D. K. Aktas and M. P. Fitz, “The distance spectrum of space-time trellis coded modulations in quasi-static Rayleigh fading channels,”
IEEE Trans. Inform. Theory, vol. 49, pp. 3335-3344, Dec. 1993.
[5] L. G. Caldeira and C. Pimentel, “Distance spectrum and expurgated union bound of space-time trellis codes in quasi-static Rician
Fading Channels,” IEEE Trans. Veh. Tech., vol. 58, no. 9, pp. 5264-5269, Nov. 2009.
[6] S. Benedetto and G. Montorsi, “Unveiling turbo-codes: Some results on parallel concatened coding schemes,” IEEE Trans. Inform.
Theory, vol. 42, pp. 409-428, Mar. 1996.
[7] H. Ogiwara, A. Mizutome, and K. Koike, ”Performance evaluation of parallel concatenated trellis-coded modulation,” IEICE Trans.
Fund., vol. E84-A, pp. 2410-2417, Oct.2001.
[8] Y. Hong, J. Yuan, Z. Chen, and B. Vucetic, “Space-time turbo trellis codes for two, three, and four transmit antennas,” IEEE Trans.
Veh. Tech., vol. 53, no. 2, pp. 318-328, Mar. 2004.