Progress In Electromagnetics Research C, Vol.

16, 69–84, 2010

CANCELLATION OF COMPLICATED DRFM RANGE

FALSE TARGETS VIA TEMPORAL PULSE DIVERSITY

G. Lu, S. N. Liao, S. C. Luo, and B. Tang

School of Electronic Engineering
University of Electronic Science and Technology of China
No. 2006, Xiyuan Avenue, West Hi-Tech Zone, Chengdu, China

Abstract—In this paper, a jamming cancelation approach based on

the concept of pulse diversity is proposed to suppress some newer
complicated digital radio frequency memory (DRFM) range false
targets (RFT). Just repeating the intercepted radar electromagnetic
signal, as done in the conventional re-transmitting jammer, is not
effective because only one range false target is produced. In contrast,
the newer DRFM-based RFT generation methods, especially chopping
and interleaving (C&I) and smeared spectrum (SMSP) can yield a
multi-lobe filter output by transforming the internal structure of
the intercepted radar signal. The presented approach to overcome
this challenge is based on the temporal pulse diversity technique,
and it does not require parameter estimation of the jamming signal.
By transmitting pulses with specific transmission pulse block and
the following proper processing, it can cancel out the protruding
spikes of the jammer at the price of an acceptable performance loss.
Particularly, this method is applicable to broad DRFM repeat jammer
in electronic warfare (EW) area.

1. INTRODUCTION

Range false targets (RFT) are employed to jam the victim radar by
transmitting the signals resembling the true target echoes reasonably
but positioned at different ranges. This deception jamming has been
enhanced significantly in recent years with the developments of digital
radio frequency memory (DRFM) techniques [1–3]. Apparently, based
on the DRFM device, the jammer has the ability to generate false
targets with all the qualities typical of the true targets specified. Hence,
Received 14 June 2010, Accepted 6 September 2010, Scheduled 10 September 2010
Corresponding author: G. Lu ([email protected]).
70 Lu et al.

such jamming is liable to be detected as target and makes radar systems

can not search or track properly. Consequently, the real targets can be
protected and the jamming purpose is then achieved.
Furthermore, some newer types of RFTs, which are generated
by transforming and processing the intercepted radar signal more
subtly before transmitting rather than merely re-transmitting, have
come into being. Especially, two effective types of such electronic
attack (EA) — C&I and SMSP [4] — were designed to attack the
LFM pulse compression radar, which made conditions much more
deteriorated. These two EAs were designed specifically to aim at the
radar’s matched-filter, and produced a multi-lobe filter output. In
practice, it is difficult for the radar to detect and recognize target
with a multitude of bunched lobe spikes which are positioned ahead
or behind the true targets and spaced closer than the time duration of
radar signal. Moreover, it is inevitable that more sophisticated RFTs
would be excogitated.
For interferences coming from different directions, the space
domain filtering algorithms can be employed to suppress them [5–
7]. But these algorithms may not suitable for DRFM repeat jammer,
because radar echo and jammer primarily come from the same
direction. For the conventional DRFM repeat jammer, it is capable
of being suppressed by transmitting (quasi-)orthogonal pulses [8, 9],
or by applying penalization measures to reduce the power of jammer
through varying modulated parameters and a three-step matched-
filtering approach [10], or by continuously emitting modified version
of previous waveforms following an orthogonal structure in sequence
or simultaneously [11, 12]. The performances of several sorts of CFAR
processor in strong pulse jamming are investigated in [13].
All of the presented EP approaches, however, do not involve the
complicated jamming types. Consequently, no suppression mechanisms
against such EA have been addressed in literatures, whereas in our
research, the sophisticated RFTs are taken into consideration. The
proposed approach enables the radar to recover echo from jammer
by using redundant codes based on temporal pulse diversity concept.
Moreover, the approach is capable of dealing with certain potential
jamming even if the transformation is unknown. Apparently, this is
a prominent merit different from other methods. The results of the
research demonstrate that this suppression approach is applicable to
a wide range of DRFM repeat jammers. Certainly, these two new
jamming — the C&I and SMSP — can be suppressed successfully.
The rest of the paper is organized as follows. Section 2 analyzes
the jammer in details, especially the C&I and SMSP, and gives the
digital representations of the jamming signal. Section 3 describes
Progress In Electromagnetics Research C, Vol. 16, 2010 71

the jamming cancelation algorithm. Section 4 provides simulations.

Finally, Section 5 concludes the paper.

2. JAMMING SIGNAL ANALYSIS

The DRFM techniques use a high-speed sampling digital memory

for storing the intercepted radar signal and later recalling [14, 15].
Generally, in a common DRFM-based EA system, the intercepted
signal is first down-converted in frequency, and then sampled and
stored in a digital memory. The digital samples are then modulated in
phase, amplitude or frequency. Under the control of the system, the
modulated signal is later recalled and retransmitted. It follows that
the signal transmitted back by the DRFM-based jammer is a simple
replica of the original signal with some parameters manipulated.
Different from common DRFM-based jamming, the C&I and
SMSP do not simply re-transmit the received signals, but manipulate
the signal complicated before emitting the signals to the victim radar.
The C&I [4] scheme consists of two-step process. First, uniformly
spaced sampling signal segments of the radar pulse are picked out.
So, the selected signal segments have certain different time slots
and frequencies. Second, the EA signal is created by placing
(concatenating) the segments in the adjacent vacant slots. The
components of the EA signal are segments of the received radar
signal to which the filter transfer function is matched. Consequently,
through matched-filtering, the output of this jamming shows a multi-
lobe structure. Furthermore, positions of the grouped range lobes
are controllable, and this makes it possible to position them ahead
or behind the true targets.
Figure 1 shows an example of C&I jamming generated from the
original LFM signal with 20 µs pulse width and 10 MHz bandwidth.
In this example, 5 uniformly spaced sampled segments of the original
LFM signal are placed in the adjacent vacant slots, and each selected
segment is placed for 4 times.
The SMSP [4] scheme is another method used for generating
EA signal which is comprised of certain short time duration sub-
waveform. It makes one sub-waveform repeat (at least) n times, and
the chirp rate of the sub-waveform is n times as the original waveform.
When an SMSP signal inputs into a matched-filter with respect to the
original waveform, the response amplitude looks like a comb structure
(uniformly distributed spikes). Each lobe has a spectrum density
distribution which is difficult to be distinguished from the target echo
in practice.
Figure 2 shows an example of SMSP jamming generated from
72 Lu et al.

(a) (b)

(c)
Figure 1. An example of C&I jamming.

(a) (b)

(c)
Figure 2. An example of SMSP jamming.
Progress In Electromagnetics Research C, Vol. 16, 2010 73

original LFM signal with 20 µs pulse width and 10 MHz bandwidth. In

this example, the sub-waveform is transmitted for 5 times, each with 5
times chirp rate and 1/5 pulse width with regard to the original signal.
Both of the two jamming techniques exploit the properties of the
matched-filter. The replica of at least a portion of the intercepted
radar signal is matched to the matched-filter impulse response of victim
radar. Consequently, a multi-lobe filter response is produced, and
these plurality lobes can occupy exactly over the whole LFM signal
frequency band and time duration. Thus, it is difficult for the radar
to discriminate the true target from these complicated RFTs.
In practice, we can model both the C&I and SMSP techniques
as a two-step procedure: A signal transformation and a cascade
amplification process. Only the first transformation process is
considered here (all the power gain is considered in received signals
in victim radar), thus any complicated DRFM repeat jammer can be
expressed as
y(t) = T x(t) = [T x](t) (1)
where y(t) and x(t) are produced jamming signal and intercepted radar
signal, respectively, and T denotes a bounded linear operator, which
indicates a transforming process applied to the received radar signals.
Specifically, if T is an identical transformation, i.e.,
[T x](t) = x(t), (2)
obviously it is just a conventional repeat jammer.
We assume that the received radar signal x(t) is band-limited for
some fixed B > 0, and satisfies F[x(t)] ∈ L2 [−B, B], where F denotes
the Fourier transform. Thus the transform can be expressed from the
series representation [16]
X
[T x](t) = x(tn )[T φn ](t) (3)
n

for some specific kernel φ. In the Hilbert space L2 [−B, B],

sin π(2wt − n)
φn (t, w) = , n = 0, ±1, ±2, . . . (4)
π(2wt − n)
form a complete orthogonal basis, where w is the frequency variable.
Thus, the digital representation of the transform process is given by
X ³ n ´ · sin π(2B(·) − n) ¸
[T x](t) = x T (t). (5)
n
2B π(2B(·) − n)
Here (·) denotes the independent variable. And, if T is time-invariant,
which is often encountered in practice, the previous expression takes
74 Lu et al.

the simpler form

X ³ n ´ · sin 2πB(·) ¸³ n ´
[T x](t) = x T t− . (6)
n
2B 2πB(·) 2B

3. PRINCIPLE OF JAMMING CANCELATION

3.1. Scenario
Pulse diversity is an effective way to suppress DRFM repeat jammer.
It utilizes the fact that radar varies its transmitting pulses in the
slow-time domain, so the jammer is forced to detect and analyze each
updated pulse. It takes a certain period of time to complete a series of
processing before it can transmit a jammer with respect to the updated
pulse. Therefore, we assume the jammer lags one pulse behind the
radar. This scenario is reasonable in view of the fact as follows. Even
if the jammer has the ability to receive and re-transmit the current
pulse at once, it can only use the previous pulse coming from the radar
when it attempts to protect targets which are closer to the radar than
itself. Considering the process time, the actual protection range would
expand.

3.2. Radar and Jamming Signal Model

Suppose the radar transmission pulse in slow-time interval u is Pu (t),
where t denotes the fast-time. Thus, the jammer transmitting pulse
in slow-time interval u can be expressed as T [Pu−1 (t)], where T is an
operator defined in Section 2. Consequently, the received pulse of radar
in slow-time interval u is as follows
X X
s(t, u) = αn (u)pu [t − τn (u)] + βl (u)T {pu−1 [t − τl (u)]} (7)
n l

where, αn (u) and βl (u) denote the reflectivity of the target echoes and
jamming signals, respectively. And, τn (u) and τl (u) denote the time-
delay of the target echoes and jamming signals relative to each pulse
repetition interval (PRI), respectively. It is obvious that the received
radar signals in a given slow-time interval contain both current echoes
and re-transmitting jamming with respect to the previous pulse.
In some applications, coding schemes are designed to provide full
diversity, simple decoding strategy and higher system performance.
Here, a real transmission pulse block across four pulse intervals is
defined in Table 1 [17–19], and the corresponding jamming signals are
provided also.
Progress In Electromagnetics Research C, Vol. 16, 2010 75

Table 1. Transmission pulse block I.

u 0 1 2 3
Radar p1 (t) p2 (t) −p2 (t) p1 (t)
Jammer T [p0 (t)] T [p1 (t)] T [p2 (t)] T [−p2 (t)]

In the initial pulse diversity block, p0 (t) may be any arbitrary

pulse. Nevertheless, it is definite in subsequent blocks, just the last
radar transmission pulse.
When slow-time u = 1 and 3, according to the transmission pulse
block shown in Table 1, the received signals of radar can be expressed
as
X X
s(t, 1) = αn (1)p2 [t − τn (1)] + βl (1)T {p1 [(t − τ l (1)]} (8)
n l
and
X X
s(t, 3) = αn (3)p1 [t − τn (3)] + βl (3)T {−p2 [t − τ l (3)]} (9)
n l
The radar transmission pulses in slow-time intervals 0 and 2
are also used to disturb the repeat jammer by keeping the jammer
analyzing and processing the updated pulses.

3.3. Jamming Signal Cancelation

When radar works in high pulse repetition frequency (PRF), we
assume the echoes reflected from the true targets and the jamming
re-transmitted by the jammer are stationary due to the slight change
over several short periods of PRI, and the amplitude of signals maintain
constant in one transmission pulse block. Consequently, for the
received signal of radar, the reflectivity notations αn (u) and βl (u) can
be reduced to αn and βl , respectively.
For uniform motion target, the round-trip time delay τn (u) and
τl (u) can be modeled uniformly as equation below (although the
jamming is single-trip actually, it can be simulated by the jammer
as round-trip model)
2(R − vt)
τ (t) = (10)
c
where R is the initial range of target, v is the relative radial velocity
between radar and target, and c is the velocity of light. Actually, as we
will see later, even if the assumption of uniform velocity is not fulfilled,
it does not affect the performance of cancellation.
76 Lu et al.

The target moves for double PRI from slow-time interval 1 to

interval 3. So, the corresponding time delay difference is
4v
∆τ = , (11)
cfPRI
where fPRI denotes PRF. Because a high-repetition frequency is
assumed, and v is much smaller than the light velocity c, the magnitude
of ∆τ is generally very small in the level of nanosecond. For instance,
when v = 1 km/s and fPRI = 10 kHz, we have ∆τ = 1.33 ns. In
this example, the PRF is not extremely high. If the PRF increases
much higher, which is frequently encountered in practice, the time-
delay difference will be much smaller.
Because the pulse signal is processed in base band, the signal
sampling rate (uniform sampling is assumed) is then relatively in a
lower level, generally varying within several to tens of million samples
per second (MSPS) after decimating, namely the sampling interval is
greater than ten nanoseconds at least.
Accordingly, the time span with respect to the adjacent sampling
points is greater than the time-delay difference by roughly an order of
magnitude at least. So, the slight difference across two pulse intervals
derived from target moving is no more than one sampling point in
most practice cases. It is confirmed that the error introduced by the
time delay difference is likely less than the error introduced in the
sampling process. Hence, this error can be negligible completely on
such conditions. We may notice that uniform velocity is not a necessary
condition to deduce this conclusion. As a result, whether the uniform
velocity is assumed or not, the conclusion is valid only if the error is
less than one sampling point.
In some exceptional cases, the time-delay difference of each
interesting arrival pulse can not be simply neglected. Under this
condition, it can be compensated by easily multiplying the frequency
response of the matched-filter by a delay factor or aligning the leading
edge of received pulses over specific intervals if the time-delay difference
can be estimated or obtained from other sensors, or errors will be
introduced.
Accordingly, the time-delay τn (u) and τl (u) can be simply
expressed as τn and τl , respectively. Where, τn and τl denote the
time-delay after compensating or simply neglecting the difference.
Although the Doppler shift of the target (both the true and the false)
is not apparently addressed in above text, the effect aroused by it
can be completely negligible in view of the high PRF and the actual
acceleration of target. Thus the received signals in slow-time interval
Progress In Electromagnetics Research C, Vol. 16, 2010 77

1 and 3 are given by

X X
s(t, 1) = αn p 2 (t − τn ) + βl T [p1 (t − τ l )] (12)
n l
and X X
s(t, 3) = αn p1 (t − τn ) − βl T [p2 (t − τ l )]. (13)
n l
To achieve the purpose of jamming cancellation, the received
signals in slow-time interval 1 and 3 pass through matched-filter
separately with respect to the true targets and then add together,
namely
sn (t) = p∗2 (−t) ⊗ s(t, 1) + p∗1 (−t) ⊗ s(t, 3)
( )
X X
= p∗2 (−t) ⊗ αn p2 (t − τn ) + βl T [p1 (t − τl )]
n l
( )
X X
+p∗1 (−t) ⊗ αn p1 (t − τn ) − βl T [p2 (t − τl )]
n l
X X
= αn psf p2 (t − τn ) + p∗2 (−t) ⊗ βl T [p1 (t − τl )]
n l
X X
+ αn psf p1 (t − τn ) − p∗1 (−t)⊗ βl T [p2 (t − τl )] (14)
n l
where p u ∗ (−t) denotes the impulse response of matched-filter focused
on the true targets in slow-time interval u, the asterisk denotes
conjugate, ⊗ denotes the convolution operation, and psf(t) denotes
point spread function (psf). It is clear that the terms 1st and 3rd
which contain psf(·) on the right side of (14) embody information of
target, while the terms 2nd and 4th relate to the jammer. Since the
digital representation of operator T isPknown, we can substitute (5)
or (6) into the expression p∗2 (−t) ⊗ βl T [p1 (t − τl )] and p∗1 (−t) ⊗
P l
βl T [p2 (t − τl )], respectively. Generally, p∗2 (−t) ⊗ T [p1 (t − τl )] 6=
l
p∗1 (−t) ⊗ T [p2 (t − τl )], i.e., the jammer-related terms can not be
canceled. But, in a special case where p1 (t) = p2 (t), the cancelation
can be performed. Therefore, if p1 (t) is selected identical with p2 (t),
(14) becomes
X
sn (t) = αn [psf p1 (t − τn )+psf p2 (t − τn )] (15)
n
Evidently, the jammer terms have been canceled. On the other
hand, if we focus only on the suppression of false target, a feasible
transmission pulse block without full diversity can be given in Table 2.
78 Lu et al.

Based on this pulse block, the cancelation result is given as follows.

sn (t) = p∗2 (−t) ⊗ s(t, 1) + p∗2 (−t) ⊗ s(t, 3)
( )
X X
= p∗2 (−t) ⊗ αn p2 (t − τn ) + βl T {p1 [−(t − τl )]}
n l
( )
X X
+p∗2 (−t) ⊗ αn p2 (t − τn ) − βl T {p1 [−(t − τl )]}
n l
X
=2 αn psf p2 (t − τn ) (16)
n
As expected, the RFTs are canceled completely. According
to (14), (15) and (16), the jammer-related terms can be canceled
completely when required conditions are satisfied. No other special
constrains have been imposed upon the derivation. In theory, the
cancelation performance of the proposed method is independent of the
JSR and jamming clearance.

4. SIMULATIONS

In simulations, the LFM signal with bandwidth 10 MHz and pulse

width 20 µs is employed. Besides, the sampling frequency is 40 MSPS,
namely the time interval is 25 ns. A target echo is received in the
presence of a C&I or SMSP jammer with jamming-to-signal ratio (JSR)
equal to 15 dB. In the process of pulse compression, the matched-
filtering measure is applied with a rectangle window.

4.1. Using Pulse Block I

Here, we choose the transmission pulse block I defined in Table 1.
As described in Section 3.3, however, the pulse p1 (t) and p2 (t) must
be identical here. The proposed approach is employed to cancel the
jamming, and the direct matched-filtering method is used also in order
to show different effect.
First, C&I jamming is tested. It can be seen clearly from
Fig. 3(a) that the signal processed by direct matched-filtering still

Table 2. Transmission pulse block II.

u 0 1 2 3
Radar p1 (−t) p2 (t) −p1 (−t) p2 (t)
Jammer T [p0 (t)] T [p1 (−t)] T [p2 (t)] T [−p1 (−t)]
Progress In Electromagnetics Research C, Vol. 16, 2010 79

(a)

(b)

Figure 3. Cancelation of C&I jamming using transmission pulse block

I.

(a)

(b)

Figure 4. Cancelation of SMSP jamming using transmission pulse

block I.
80 Lu et al.

(a)

(b)

Figure 5. Cancelation of C&I jamming using transmission pulse block

II.

remains a cluster of false targets. While, as shown in Fig. 3(b), the

proposed approach can cancel out the jammer and focus the true target
distinctly.
Second, SMSP jamming is applied to jam the victim radar. In
Fig. 4(a), there are comb outputs besides the true target. Conversely,
as shown in Fig. 4(b), the complicated false targets have been
suppressed and the true target is obtained.

4.2. Using Pulse Block II

In this simulation, the transmission pulse block II defined in Table 2
is used. Unlike the previous case, the pulse width parameter is set to
20 µs and 15 µs while other conditions are remained. It follows that we
alter simply the pulse width of the signal in transmission block, i.e.,
p1 (t) 6= p2 (t).
In Figs. 5 and 6, the cancelation effects of C&I and SMSP are
shown respectively to evaluate the algorithm based on a transmission
block without full diversity. It is proved that the jamming signal can
be eliminated successfully and the cancelation performance is almost
identical with using transmission block I.
Progress In Electromagnetics Research C, Vol. 16, 2010 81

(a)

(b)

Figure 6. Cancelation of SMSP jamming using transmission pulse

block II.

(a) (b)

(c) (d)
Figure 7. RMS of recovered target signal by using block I/II.
82 Lu et al.

4.3. RMS of Recovered Target Signal

Here, the ratio of mainlobe to sidelobe (RMS) of recovered target signal
is provided also to evaluate the performance of the presented approach.
In this simulation, the signal-to-noise ratio (SNR) varies from −10 to
15 dB and the JSR varies from 5 to 15 dB. The noises are additive,
Gaussian and white. Monte Carlo simulation is run 250 times.
As shown in Fig. 7, the RMS ranges from 10.8 to 14.5 dB
and gradually converges to 13.4 dB (rectangle window) as the SNR
increases. In addition, this simulation shows that the JSR is
independent of the RMS.

5. CONCLUSIONS

This paper addresses issues about the newer types of complicated

DRFM repeat jamming and presented a method to suppress them.
By exploiting the redundant but essential code pulses, the proposed
approach enables radar to recover the true target signals easily
through jamming cancelation. Different from other methods, it is
efficient to suppress complicated jamming signals besides conventional
ones. The research results of the present work demonstrate that the
proposed approach is applicable to a wide-ranging scope of DRFM
repeat jammer, particularly the C&I and SMSP, even some potential
complicated ones. It provides radar with an active strategy to counter
against the repeat jammer, not just a passive anti-jamming method.

REFERENCES

1. West, P. D. and B. J. Slocumb, “ECM modeling for assessment of

target tracking algorithms,” IEEE Proc. 29th Southeastern Symp.
System Theory, 500–504, Cookeville, USA, Mar. 1997.
2. DiFilippo, D., G. Geling, and G. Currie, “Simulator for advanced
fighter radar EPM development,” IEE Proceedings Radar Sonar
and Navigation, Vol. 148, No. 3, 139–146, 2001.
3. Pace, P. E., D. J. Fouts, S. Ekestorm, and C. Karow, “Digital false-
target image synthesiser for countering ISAR,” IEE Proceedings
Radar Sonar and Navigation, Vol. 149, No. 5, 248–257, 2002.
4. Sparrow, M. J. and J. Cikalo, “ECM techniques to counter pulse
compression radar,” United States Patent 7081846, Jul. 2006.
5. Mouhamadou, M., P. Armand, P. Vaudon, and M. Rammal, “In-
terference suppression of the linear antenna arrays controlled by
phase with use of SQP algorithm,” Progress In Electromagnetics
Research, Vol. 59, 251–265, 2006.
Progress In Electromagnetics Research C, Vol. 16, 2010 83

6. Mahmoud, K. R., M. El-Adawy, S. M. M. Ibrahem, R. Bansal,

K. R. Mahmoud Visiting, and S. H. Zainud-Deen, “Performance
of circular Yagi-Uda arrays foe beamforming applications
using particle swarm optimization algorithm,” Journal of
Electromagnetic Waves and Applications, Vol. 22, No. 2–3, 353–
364, 2008.
7. Guney, K. and S. Basbug, “Interference suppression of linear
antenna arrays by amplitude-only control using a bacterial
foraging algorithm,” Progress In Electromagnetics Research,
Vol. 79, 475–497, 2008.
8. Garmatyuk, D. S. and R. M. Narayanan, “ECCM capabilities
of an ultrawideband bandlimited random noise imaging radar,”
IEEE Transaction on Aerospace and Electronic Systems, Vol. 38,
No. 4, 1243–1255, 2002.
9. Schuerger, J. and D. Garmatyuk, “Deception jamming modeling
in radar sensor networks,” IEEE Military Communication
Conference, 1–7, San Diego, USA, Nov. 2008.
10. Soumekh, M., “SAR-ECCM using phase-perturbed LFM chirp
signals and DRFM repeat jammer penalization,” IEEE Transac-
tion on Aerospace and Electronic Systems, Vol. 42, No. 1, 191–205,
2006.
11. Akhtar, J., “An ECCM scheme for orthogonal independent range-
focusing of real and false targets,” IEEE Radar Conference, 846–
849, Boston, USA, Apr. 2007.
12. Akhtar, J., “An ECCM signaling approach for deep fading of
jamming reflectors,” IET International Conference on Radar
Systems, 1–5, Edinburgh, UK, Oct. 2007.
13. Garvanov, I. and C. Kabakchiev, “One and two dimensions CFAR
processors in the presence of strong pulse jamming,” Cybernetics
and Information Technologies, No. 1, 58–72, 2002.
14. Roome, S. J., “Digital radio frequency memory,” Electronic
Communication Engineering Journal, Vol. 2, No. 4, 147–153, 1990.
15. Greco, M., F. Gini, and A. Farina, “Radar detection and
classification of jamming signals belonging to a cone class,” IEEE
Transaction on Signal Processing, Vol. 56, No. 5, 1984–1993, 2008.
16. Habib, M. K., “Digital representations of operators on band-
limited random signals,” IEEE Transaction on Information
Theory, Vol. 47, No. 1, 173–177, 2001.
17. Lu, H.-F., P. V. Kumar, and C. Habong, “On orthogonal
designs and space-time codes,” Proceedings of IEEE International
Symposium Information Theory, 418, Lausanne, Switzerland,
84 Lu et al.

2002.
18. Alamouti, S. M., “A simple transmit diversity technique for
wireless communications,” IEEE Journal Selected Areas in
Communications, Vol. 16, No. 8, 1451–1458, 1998.
19. Su, W. F. and X.-G. Xia, “Quasi-orthogonal space-time block
codes with full diversity,” IEEE Global Telecommunications
Conference, 1098–1102, Taipei, China, Nov. 2002.