Studies On Novel Anti Jamming Technique of Unmanned Aerial Vehicle Data Link

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Chinese

Journal of
Chinese Journal of Aeronautics 21(2008) 141-148
Aeronautics
www.elsevier.com/locate/cja

Studies on Novel Anti-jamming Technique of Unmanned


Aerial Vehicle Data Link
Huang Wenzhuna,*, Wang Yongshenga, Ye Xiangyangb
a
School of Electronics and Information, Northwestern Polytechnical University, Xi’an 710072, China
b
Telecommunication Engineering Institute, Air Force Engineering University, Xi’an 710077, China

Received 14 August 2007; accepted 26 December 2007

Abstract

Based on the M-ary spread spectrum (M-ary-SS), direct sequence spread spectrum (DS-SS), and orthogonal frequency division
multiplex (OFDM), a novel anti-jamming scheme, named orthogonal code time division multi-subchannels spread spectrum modulation
(OC-TDMSCSSM), is proposed to enhance the anti-jamming ability of the unmanned aerial vehicle (UAV) data link. The anti-jamming
system with its mathematical model is presented first, and then the signal formats of transmitter and receiver are derived. The receiver’s
bit error rate (BER) is demonstrated and anti-jamming performance analysis is carried out in an additive white Gaussian noise (AWGN)
channel. Theoretical research and simulation results show the anti-jamming performance of the proposed scheme better than that of the
hybrid direct sequence frequency hopping spread spectrum (DS/FH SS) system. The jamming margin of the OC-TDMSCSSM system is
5 dB higher than that of DS/FH SS system under the condition of Rician channel and full-band jamming, and 6 dB higher under the
condition of Rician channel environment and partial-band jamming.

Keywords: UAV data link; OC-TDMSCSSM; systemview simulation; orthogonal code; anti-jamming

1 Introduction* spectrum (M-ary-SS), seem to be weak in anti-jam-


ming performance, and unable to satisfy the
Unmanned aerial vehicle (UAV) has become
requirements of UAV data link in unfavorable and
one component of the prosperous modern aviation
complex electromagnetic environments[2-3]. Recent
weapon system for its outstanding tactical and tech-
researches introduce the application and perform-
nical performance. However, in implementing tasks
ance analysis on frequency hopping spread spec-
such as reconnaissance, surveillance, airborne warn-
trum (FH-SS), multi-carrier code division multiple
ing, communication relay and electronic counter-
access (MC/CDMA)[4-5], and hybrid direct sequence
measures, it is often subjected to interference. As an
frequency hopping spread spectrum (DS/FH SS)
important part of UAV system, the UAV data link
system[6]. They show that the hybrid DS/FH SS sys-
demands the highest reliability in data transmission,
tem has an anti-jamming performance better than
and the most powerful ability of anti-jamming and
DS-SS, M-ary SS, FH-SS and MC/CD-MA. Thus,
anti-multipath[1]. As an existing UAV data link, the
to enhance the anti-jamming ability of the UAV data
applied simple anti-jamming systems, direct se-
link in various complex environments, the advan-
quence spread spectrum (DS-SS) or M-ary spread
tages of the traditional spread spectrum system can
*Corresponding author. Tel.: +86-13892852375.
be utilized and optimized in the development of a
E-mail address: wenzhunw@sohu.com new anti-jamming technology to improve the elec-
Foundation item: Aeronautical Science Foundation of China (2007ZC53030)
· 142 · Huang Wenzhun et al. / Chinese Journal of Aeronautics 21(2008) 141-148

tronic counter measures of UAV data link. and each symbol selects respective orthogonal
This paper proposes a novel anti-jamming spread spectrum subchannels to transmit according
technique, termed orthogonal code time division to the mapping coding principle. Then orthogonal
multi-subchannels spread spectrum modulation spread spectrum subchannels are transmitted timely
(OC-TDMSCSSM), in which the M-ary is associ- through a radio frequency (RF) part.
ated with the DS-SS and orthogonal frequency divi-
sion multiplex (OFDM). Different from the M-ary
system, the former spread spectrum is added before
serial-parallel converter, and in mapping coding, the
draft of orthogonal spread spectrum subchannel is
chosen instead of pseudorandom (PN) code. Differ-
ent from the DS/FH SS system, though the former
spread spectrum employs the same theory as DS-SS,
in the DS/FH SS system, one hoping frequency Fig.1 OC-TDMSCSSM transmitter.

transmits multiple bits information, or multiple


The data source period is denoted by Ts, the
hoping frequencies transmit one bit information,
amplitude of qth data source by Aq, and the power
while in the new scheme associated with M-ary SS,
of data source denoted by Pq equals Aq2 / 2 . In the
multiple orthogonal spread spectrum subchannels
orthogonal code set, the code length of each or-
represents one bit information. Different from the
thogonal sequence equals N, and the period is Tc.
OFDM system, a new segment of former spread
For given M modulators, their carriers are in an or-
spectrum is added and orthogonal spread spectrum
thogonal frequency set, f O (O 1, 2, ˜˜˜, M ) , then Tc
subchannels are adopted instead of the orthogonal
= Ts/MN.
subchannels, thereby improving the spread spectrum
Assume that the chip waveform of the spread
gain and correlated characteristic between subchan-
spectrum code is a unit-amplitude rectangular pulse
nels[7]. The simulation results confirm that the out-
with chip duration Tc, and then p(t) is
put signals of OC-TDMSCSSM system like the
Gaussian white noise signals lead to an anti-jam- ­1, 0 d t d Tc
p(t ) ® (1)
ming performance better than the DS/FH SS system. ¯0, Other

In the orthogonal code set, the spread spectrum


sequence is expressed by
2 System Description
CO ^C (O , i)` , C (O , i )  [1,  1] (2)
2.1 Transmitter model where O 1, 2, ˜˜˜, M , i 1, 2, ˜˜˜, N .
Fig.1 shows the model of transmitter, in which Hence, the output signal of transmitter is
the input binary data sequence at 1/Ts rate is first f M N

encoded and multiplied by a higher rate channel PN


S (t ) ¦ ¦¦ 2 Pq ˜ C (O , i ) ˜
q f O 1 i 1
sequence. The result is a binary data sequence at iTs OTs
p(t   ) ˜ cos 2Sf O t (3)
Mk/Ts rate. Then these data are grouped according MN M
to k bits forming a data symbol with M=2k symbols.
2.2 Receiver model
Each data symbol is mapped with one of M or-
thogonal spread spectrum subchannels which are The model of a receiver is shown in Fig.2. In
composed of M orthogonal spread spectrum se- the parallel correlated synchronization receiver, M
quences modulated by binary phase-shift-keying digital matched filters (DMFs) are adopted. One
(BPSK) in the orthogonal spread spectrum code set, self-correlated peak is an output in each period of
Huang Wenzhun et al. / Chinese Journal of Aeronautics 21(2008) 141-148 · 143 ·

spread spectrum subchannel through the orthogo- where Q(P, r) is a Gaussian stochastic variable with
nality of spread spectrum subchannel and the corre- zero-mean and NTcN0/4 variance.
lation of PN code. Thus, M self-correlated peaks can The false judge probability of the jth orthogo-
be obtained and detected while one bit information nal code and frequency is
is transmitted. After de-mapping M self-correlated Pc ( j ) {1  P [ Z j ! ( Z1 , Z 2 ," Z j 1 , Z j 1 ," ,
peaks to spread spectrum subchannel coding, the Z M ) / f j ]} ˜ P( f j ) (8)
result is sent to the correlator of local DS sequence,
and one bit information can be obtained. Without where P( fj ) is the false judge probability of the jth
considering the radio frequency part, the receipt frequency in orthogonal frequency set as 1/(M–1).
signal in AWGN can be denoted by
P{Z j ! ( Z1 , Z 2 ," Z j 1 , Z j 1 ," , Z M ) / f j }
r (t ) S (t )  n(t )  J (t ) (4) Zj Zj Zj
³ P( Z / f j ) dZ [ P ( Z j ! Z x , x z j )]M 1 (9)
f ³f ³
"
where S(t) is given by Eq.(3), n(t) is an additive f

( M 1)
white Gaussian noise (AWGN) with single-sided
spectral density N0, and J(t) the jamming signal. Then Eq.(8) can be rewritten into
1
Pc ( j ) ˜ {1  [ P ( Z j ! Z x , x z j )]M 1} (10)
M 1

According to the central limit theorem, when


the orthogonal code length N is big enough, the dis-
tribution of Z can be approximated by the Gaussian
process. Because the variance of Zj – Zx is V2=
NTcN0/4, and the mean is NTc(Pq/2)1/2, the following
formula can be obtained:
Fig.2 OC-TDMSCSSM receiver.
 ( y  NTc Pq / 2 )2
f 1 2V 2
P{Z j ! Z x , x z j} ³0 e dy
3 Bit Error Rate (BER) Analyses in 2ʌV
AWGN Channel erf ( Eb / N 0 ) (11)
In the case that the receiver has caught syn-
where erf(x) denotes the error function. After sub-
chronization in AWGN, assume the transmitting
stituting Eq.(11) into Eq.(10), Pc( j ) can be obtained
signal symbol is the Oth in AWGN channel, thus the
signal in receiver can be expressed by 1
Pc ( j ) ˜ {1  [erf ( Eb / N 0 )]M 1} (12)
M 1
r (t ) S (VO , t )  n(t ) (5)
Thus in detecting the Mth orthogonal code and or-
where n(t) is white Gaussian noise with dou-
thogonal frequency, the false judge probability is
ble-sided spectral density N0/2. If coherence recep-
M
tion is adopted, the Pth correlator, denoted by ZP, is 1
Pc ¦ Pc ( j ) Pc ( j ), j 1, 2," , M (13)
the signal symbol of 1/M bit defined as M j 1

N
ZP
NTc
r (t )¦ C ( P , n) ˜ p (t  nTc ) ˜ cos 2Sf P tdt (6)
From the transmitter model, when M orthogo-
³0 n 1 nal codes and orthogonal frequencies have been
detected correctly, i.e. one bit information has been
After substituting Eq.(5) into Eq.(6), ZP can be
transmitted, the BER in AWGN is
rewritten into
­° NTc 2 Pq  Q (O , r ), P O and f P f O 1
Pb Pc ˜ {1  [erf ( Eb / N 0 )]M 1} (14)
ZP ® (7) M 1
°̄ Q( P , r ), P z O or f P z f O
· 144 · Huang Wenzhun et al. / Chinese Journal of Aeronautics 21(2008) 141-148

Ws
4 Anti-jamming Performance Analysis Njm  N0  fO  ª2 º
³ W
2 Sinc2 « f  O  fO  fIF » dO (19)
Ws  fO  s ¬Ws ¼
The current main jamming methods aimed at 2

the UAV data link are of full-band jamming, par- With the definition of J= f – O– fO– fIF, Eq.(19)
tial-band jamming, multitone jamming, and pulse can be simplified into
jamming[8]. Later, the analysis and simulation of the Ws
N 0  N jm f  f IF  § 2J ·
first two jamming methods are presented. N f SR f | ³ W
2 Sinc 2 ¨ ¸ dJ
Ws f  f IF  s © Ws ¹
2
4.1 Performance in full-band jamming
(20)
Assume that the input signal of UAV is
and the double-sided spectral density of noise near
r (t ) S (t )  n(t ) (15) f = ± fIF is

where n(t) is wide-band noise. Eq.(15) can also be Nn N 0 N jm


N f IF SR f IF |  (21)
expressed as 2 2 2

r t S t  2nI t cos 2Sft  2nQ t sin 2Sft (16) It shows that when the system is subjected to
full-band jamming, the noise band is much wider
where nI(t) and nQ(t) are two mutual independent than that of the subchannel, and Nn/2 approaches
low-pass Gaussian noises with double-sided power N0/2 or Njm/2, meaning that broadband jamming
spectral density (N0+Njm)/2, where N0 denotes sin- noise possesses features similar to the receiver’s
gle-sided power spectral density of thermal noise, inner noise.
and Njm single-sided power spectral density of jam-
4.2 Performance in partial-band jamming
ming signal. Therefore, the component near the
intermediate frequency after correlation is Compared with full-band jamming, it is easier
M to generate partial-band jamming having a higher
H (t ) ¦ {UO (t  OTs / M ) cos[ZIFt  Td (t )]  power spectral density. Assume that W is the band-
O 1
width of radio frequency, Wjm the jamming band-
[ 2nI (t ) cos ZO t  2nQ (t ) sin ZO t ] ˜
width, Ws the subchannel bandwidth, Pjm the jam-
2 cos[(ZO  ZIF )t  Ts (t )]} (17)
ming power, K the rate of W and Wjm, Np the spectral
where UO(·) is the self-correlated function for or- density of partial-band jamming, Njm the spectral
thogonal code in Oth subchannel, and Ts(t) the ran- density of full-band jamming, and N0 single-sided
dom modulation phase. In the receiver, the power spectral density of thermal noise. Thus the final
spectrum of each sebchannel output is spectral density of noise is N0+Np, and only N0 is the
effects in the remaining partial-band. The receiver
2 ª 2 º
SR f Sinc2 « f  f O  f IF »  signal is
Ws ¬ Ws ¼
2 ª 2 º r (t ) S (t )  n(t )  n jm (t ) (22)
Sinc 2 « f  f O  f IF » (18)
Ws ¬Ws ¼ where S(t) is the useful transmitting signal, n(t) the
where Ws is the bandwidth of subchannel. When the thermal noise, and njm(t) the partial-band jamming.
frequency components excluding fIF have been fil- After correlative calculation, the output of the des-
tered, the following equation holds. pectral mixer is
M
N f SR f |
Ws
H (t ) ¦ {UO (t  OTs / M ) ˜ cos[2ZIFt  Td (t )]} 
N jm  N 0 fO  ª 2 º O 1
³ W
2 Sinc 2 « f  O  f O  f IF » dO 
Ws fO  s ¬Ws ¼
2
Huang Wenzhun et al. / Chinese Journal of Aeronautics 21(2008) 141-148 · 145 ·

M N Eb Eb
n(t ) ˜ ¦¦ {C (O , i ) ˜ p (t  O Ts M ) ˜ cos[2(ZO  ZIF )t  Pa 1  K Pb ( )  K Pb ( ) (26)
O 1i 1 N0 N 0  N c jm
M N
T d (t )]}  n jm (t ) ˜ ¦¦ {C (O , i ) ˜ where Pb(·) is determined by Eq.(14). In such cases,
O 1i 1
p (t  O Ts M ) ˜ cos[2(ZO  ZIF )t  T d (t )]} receiver can still obtain a majority of self-correlated
(23) peaks and recover the data as long as the number of
interferential subchannels is less than half of the
where the second term is the noise caused by co-
number of subchannels.
actions of the thermal noise and the despectral re-
ference waveform, and the bandwidth is much wider 5 Simulations and Discussion
than that of the spectral waveform, which has been
In the simulation, the OC-TDMSCSSM system
discussed in Section 4.1. Hereafter, two cases are
parameters are assumed as follows: the source rate
considered as follows:
is 8 Kbps; the former spread spectrum segment
Case one when Wjm<Ws, i.e. one of the
adopts BPSK modulation; the length of former
sub-channels is jammed, for the single subchannel,
spread spectrum code is 64 with a chip rate of 512
the output of despectral mixer is
kc/s; the number of sub-channels 16 with a chip rate
H1 t U cos[ZIFt  T d (t )] 
of 2 048 kc/s; the system bandwidth is 32.768 MHz,
n t {2c ˜ cos[(ZO  ZIF )t  Ts (t )]}  and the intermediate frequency 85 MHz. The DS/
n jm (t ) ^2c ˜ cos[(ZO  ZIF )t  Ts (t )]` (24) FH SS system parameters are chosen as follows: the
source rate is 8 Kbps; the length of DS-SS code 64
where U is the self-correlated function for the or-
with a chip rate of 1 024 kc/s; the number of hop-
thogonal code of the single subchannel, and c the
ping frequency 16; the FH rate 16 000 hop/s, 1bit
orthogonal code of the single subchannel. The
per two hoppings; the system bandwidth is 32.768
power spectral density of the third term, generated
MHz, and the intermediate frequency 22.36 MHz.
by jamming, is the convolution of jamming spectral
The jamming signal parameters are taken as follows:
density and spread spectrum waveform spectral
the full-band jamming signal is applied by Gaussian
density. Along the similar lines cited in Section 4.1,
white noise signal with an alterable gain amplifier;
it can be deducted that the output spectral density in
the partial-band jamming signal is adopted by DS-
partial-band jamming is determined by
SS signal with a bandwidth of 16.384 MHz. In the
Np f jm Wjm / 2 ª 2 º BER performance analysis, the simulation is based
S r f IF ˜³ Sinc 2 « f O  x » dx (25)
Ws f jm Wjm / 2 W
¬ s ¼ on the MATLAB platform and the Monte Carlo
where O 1, 2," , M . In such cases, receiver can method with a sampling rate of one per chip. In the
still obtain M – 1 self-correlated peaks and receive anti-jamming ability analysis, the simulation is
data normally while only one subchannel is sub- based on the systemview dynamic platform with a
jected to interference. sampling rate of 500 MHz and sampling number of
Case two when Wjm>Ws, for simplifying 80 000. The Rician channel is chosen with a Rician
analysis, if subchannel central frequency falls in the factor of 11 dB, and the maximum Doppler fre-
jamming frequency band, the subchannel is sub- quency shift is 500 Hz[9].
jected to complete jamming. Herein, when sub- The Monte Carlo simulation results of BER in
channel is not in the jamming frequency band, the AWGN channel are shown in Fig.3, from which it is
noise double-sided spectral density of the despectral clear that the simulation curve accords well with
correlator output is N0/2; otherwise, it is (N0+Np)/2, that of the theoretical results, and when Eb/N0 is
and the average BER is equal to or larger than 9 dB, the two curves are su-
perimposed thus leading to the theoretical credibi-
· 146 · Huang Wenzhun et al. / Chinese Journal of Aeronautics 21(2008) 141-148

lity of the system BER.

Fig.6 Self-correlated peaks of OC-TDMSCSSM under


full-band jamming (1, 4, 7, 11 subchannels).

times. The signal-to-noise ratio (SNR) is 0 dB; the


jamming to signal ratio (JSR) is 22 dB and 24 dB.
Fig.3 BER performance with theoretic and Monte Carlo
simulation in AWGN channel. In the case of JSR 22 dB, the self-correlated peaks
can be extracted completely and the system can re-
The output spectrum of OC-TDMSCSSM sys-
cover the data rightly, but this is completely differ-
tem and DS/FH SS system are described in Fig.4
ent with 24 dB.
and Fig.5, respectively, which show that the trans-
Fig.7 shows the self-correlated peaks of DS/FH
mitted signals of OC-TDMSCSSM system look like
SS system under the condition of full-band jamming.
the Gaussian white noise signals in line with the
The program runs two times. The SNR is 0 dB; the
theoretical results, while in the DS/FH SS system,
JSR is 17 dB and 19 dB. In the case of JSR 17 dB,
there are 16 distinct envelops without the character-
the self-correlated peaks can be extracted com-
istics of Gaussian white noise signals. Fig.6 shows
pletely and the data can be recovered rightly, but,
the self-correlated peaks of OC-TDMSCSSM sys-
for 19 dB, the self-correlated peaks of the 3rd, 4th,
tem of the subchannels 1, 4, 7, 11 under the condi-
6th, 8th, 12th, 13th hoppings cannot be detected
tion of full-band jamming. The program runs two
normally leading to the losing of synchronization.
Therefore, the jamming margin of the OC-
TDMSCSSM system is 5 dB more than that of the
DS/FH SS system under the condition of Rician
channel and full-band jamming.

Fig.4 Output signal spectrum of OC-TDMSCSSM system.

Fig.7 Self-correlated peaks of DS/FH under full-band


jamming.

Fig.8 describes the output spectrum of the


OC-TDMSCSSM system under the condition of
partial-band jamming, which shows that the eight
anterior subchannels are covered by the major lobe
Fig.5 Output signal spectrum of DS/FH system.
of jamming signal spectrum, and the eight posterior
Huang Wenzhun et al. / Chinese Journal of Aeronautics 21(2008) 141-148 · 147 ·

subchannels by the side lobe of jamming signal tracted except those of the 4th, 12th subchannels
spectrum. Here, SNR is 0 dB and JSR is 25 dB. and the right data can be recovered, but, for 21dB,
Fig.9 shows the self-correlated peaks of OC-TDMS- the self-correlated peaks of the 2nd, 4th, 6th, 11th,
CSSM system under the condition of partial-band 12th, 14th hoppings cannot be detected normally
jamming. The program runs two times. The JSR is leading to the losing of synchronization. Therefore,
25 dB and 27 dB. For JSR 25 dB, the self-correlated the jamming margin of the OC-TDMSCSSM system
peaks can be extracted except those of the 3rd, 4th, is 6 dB more than that of the DS/FH SS system un-
5th subchannels and the right data can be recovered, der the condition of Rician channel and partial-band
but, for 27 dB, the self-correlated peaks of the 1st, jamming.
2nd, 3rd, 4th, 5th, 6th, 7th, 10th subchannels cannot
be recovered normally leading to the losing of syn-
chronization.

Fig.10 Output signal spectrum of DS/FH system under


partial-band jamming.

Fig.8 Output signal spectrum of OC-TDMSCSSM system


under partial-band jamming.

Fig.11 Self-correlated peaks of DS/FH under partial-band


jamming.

6 Conclusions
The paper proposes the OC-TDMSCSSM
Fig.9 Self-correlated peaks of OC-TDMSCSSM under
partial-band jamming (1, 4, 7, 11 subchannels). scheme as a novel anti-jamming technique for UAV
data link, for which the system and mathematical
Fig.10 describes the output signal spectrum of
model are established; the signal format derived;
the DS/FH system under the condition of partial-
and the BER and anti-jamming performance in
band jamming. It is noted that the eight anterior sub-
AWGN channel analyzed. Theoretical analysis and
channels are covered by the major lobe of jamming
simulation results demonstrate that the derived sys-
signal spectrum while the eight posterior subchan-
tem BER formula is correct and the proposed sys-
nels by the side lobe of jamming signal spectrum.
tem is technically feasible. The proposed scheme
Here, the SNR is 0 dB and the JSR is 19 dB. Fig.11
possesses much better anti-jamming capability than
shows the self-correlated peaks of DS/FH system
the DS/FH system, i.e. the jamming margin of the
under the condition of partial-band jamming. The
OC-TDMSCSSM system is 5 dB more than that of
program runs two times. JSR is 19 dB and 21 dB.
the DS/FH SS system under the condition of Rician
For JSR 19 dB, the self-correlated peaks can be ex-
channel environment and full-band jamming, and
· 148 · Huang Wenzhun et al. / Chinese Journal of Aeronautics 21(2008) 141-148

the jamming margin of the OC-TDMSCSSM system tem capacity. Proceedings of IEEE International Conference on

is 6 dB more than that of the DS/FH SS system un- Emerging Technology. 2006; 32-37.

der the condition of Rician channel environment [7] Oz E. A comparison of timing methods in orthogonal frequency

and partial-band jamming. division multiplexing (OFDM) systems. PhD thesis, Naval Post-

graduate School, 2004.


References
[8] Poisel R A. Modern communications jamming principles and
[1] Ye W. Some key techniques about modulation and coding in UAV
techniques. Norwood, MA: Artech House, Inc, 2004.
communication. PhD thesis, Northwestern Polytechnical Univer-
[9] Jin S, Zhang X L, Zhou Q. A statistical model for the UAV com-
sity, 2002. [in Chinese]
munication channel. Acta Aeronautica et Astronautica Sinica 2004;
[2] Scott W T, King D, Maschino M. Information assurance in the
25(1): 62-65. [in Chinese]
UAV/UCAV environment. Proceedings of AUVSI’s Unmanned

Systems North America 2005. 2005; 1307-1320.


Biographies:
Huang Wenzhun Born in 1968, a Ph.D. candidate in
[3] Murphy M J, Donaldson M. Digital microwave link for unmanned
Northwestern Polytechnical University. He received the B.S.
vehicles. Proceedings of AUVSI’s Unmanned Systems North
in 1994 and M.S. in 1997 from Air Force Telecommunica-
America 2005. 2005; 1029-1040.
tion Engineering Institute. His main academic interest is
[4] Cho J Y, Kim Y H, Cheun K W. A novel frequency-hopping
radio communication.
spread-spectrum multiple-access network using M-ary orthogonal
E-mail: wenzhunw@sohu.com
Walsh sequence keying. IEEE Trans on Communications 2003;

51(11):1885-1896.
Wang Yongsheng Born in 1941, he is a professor and
[5] Wen L, Zhu J K, Costa E, et al. M-ary MC CDMA in up-link for
doctor supervisor in Northwestern Polytechnical University.
next generation mobile communication. Proceedings of IEEE 58th
His main research interests are radio communication and
Vehicular Technology Conference. 2003; 4: 2526-2530.
navigation.
[6] Mohammad A R K, Shao D R, Mohammad B S. Analytical ex-
E-mail: wangys@nwpu.edu.cn
pression for hybrid DS/FH spread spectrum communication sys-

You might also like