Professional Documents
Culture Documents
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on May 28, 2009 at 21:22 from IEEE Xplore. Restrictions apply.
where SIR represents Signal-to-Interference Ratio, which is
equal to Eb / N J . The optimal jamming fraction U * can be
obtained by maximizing (14) and (15) with respect to U for a
given SIR
UBPSK
*
argmaxPˆBPSK(U) (16)
U
U DBPSK
*
arg max PˆDBPSK (U ) (17)
U
4) PBJ under Rayleigh fading channel with AWGN Here we have two random variables: one is I J , which is
For the time-correlated Rayleigh fading channel, following uniformly distributed over [0,2S ] ; the other is VI1 , which
the same steps as before, the average BER for BPSK and satisfies the Gaussian distribution with N (0, N 0 2) . Now we
DPBSK becomes
ff
2D2Eb 2D2Eb DE D2 E2
define Y A J cos( I J ) Eb , X V I 1 and W X Y .
(18)
P BPSK ³³
(U Q(
00
) (1 U) Q(
N0 E NJ / U
2
)) exp(
N0
)dDdE
V 4
2V 2
Then
ff
U D Eb2
1 U D Eb DE
2
D E
2 2
(19) PBPSK _ MTJ Pr ( SI1,r1 0) Pr (W 0) (25)
PDBPSK ³³( 2 exp(N E N / U) (
2
2
) exp( )) exp(
N0 V4 2V2
)dDdE
00 0 J The Probability Density Function (PDF) of cos() function is
Here we do not consider the optimization of jamming fraction, given in [7] as
since this requires a complicated algorithm in a time-varying 1 1
Rayleigh fading channel. It is not realistic to do so just for ° z (1,1) (26)
fZ (z) ®S 1 z2
small improvement in jamming effect. Instead jamming °0
¯
fraction optimization table from previous section is used to
obtain some improvement. The PDF of Y can be represented as
1 (27)
fY ( y)
5) MTJ under AWGN y Eb 2
Multitone jamming (MTJ) divides its total power into q SAJ 1 ( )
AJ
distinct, equal power, random phase tones. Every jamming where y should satisfy E b A J y E b AJ
tone can be modeled as
Knowing the PDF of X, the PDF of W is
J (t ) AJ e j ( 2Sf J t I J ) (20) f
vector added to them. Projecting the compound signal onto I1 easily. Thus the BER for BPSK is given as
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on May 28, 2009 at 21:22 from IEEE Xplore. Restrictions apply.
2 Eb which shows good results in the simulation.
PBPSK ( Eb , N 0 , U , SIR) U PBPSK _ MTJ (1 U )Q ( )
N0 (30) In MTJ, optimal jamming fraction should also be consid-
0 ered, so that we get the optimal-fraction MTJ. Again,
2 Eb
U ³ fW ( w)dw (1 U )Q( ) assuming the background AWGN is negligible, from (31)
f
N0
following the same process as PBJ, we can get
If the AWGN is negligible, we can get U*
1 S 0.5 SIR
U S ( arcsin( UBPSK
*
SIR)) BPSK 0 (38)
PBPSK (Eb , U, SIR) = Pes1 = ° S ( 2 arcsin( USIR )) USIR < 1 S 2 S UBPSK SIR(1 UBPSK
* *
SIR )
®
°̄0 Solving it with different SIR will generate MTJ jamming
(31) fraction optimization table of BPSK. For DBPSK, the optimal
For DPBSK, let z(k-1) and z(k) be the reference and re- jamming fraction can be obtained from simulation results. The
optimization tables of BPSK and DBPSK will be used by
ceived symbol vectors respectively. The variable D is given as
optimal-fraction multitone jammer to achieve optimal
D Re(z(k ) z * (k 1)) performance.
(32)
o o
jIJ ( k ) jIJ ( k 1)
Re(( AJ e s(k ) V (k ))( AJ e s (k 1) V (k 1)))
* *
6) MTJ under Rayleigh fading channel with AWGN
where s(k ) and s (k 1) are transmitted constellations at time Similar as before, the average BER for BPSK and DPBSK
k and k-1. D is used by differential detector to decide which for MTJ under Rayleigh fading channel with AWGN is
ff
DE D 2 E 2
symbol was transmitted. Due to symmetry, we can assume a PBPSK ³³ (P (D 2 Eb , N0 , U,D 2 SIR E 2 )) exp( )dDdE (39)
V4 2V 2
BPSK
given phase difference zero to compute the error probability 00
o
(33) where PBPSK (D 2 Eb , N 0 , U ,D 2 SIR E 2 ) is derived from (30), and
o
Re(( AJ e jIJ (k ) Eb V (k ))( AJ e jIJ (k 1) Eb V * (k 1))) when you calculate it, you should change all Eb and SIR to
If D is less than 0, then a decision error is made. That is, the
D 2 Eb , D 2 SIR E 2 correspondingly. Eq. (37) gives the value of
BER equals to Pr(D 0) when MTJ exists. However, the
PDBPSK (D 2 Eb , N 0 , U ,D 2 SIR E 2 ) . Here Eb and SIR should be
probability of D is difficult to calculate from (33), subse-
quently some approximations are necessary. For SIR >> SNR changed as well.
(Eb/N0), we can neglect some small items in (33), and get 4. Simulation results and analysis
D | Eb AJ Eb (cos(I J (k )) cos(I J (k 1))) Eb (VI1 (k ) VI1 (k 1))
In this section, the BER performance of different jamming
AJ (cos(I J (k ) I J (k 1)))
2
techniques for OFDM system is evaluated by the means of
(34) software simulation. Based on the 802.11a standard [10], the
Dividing (34) by Eb yields main parameters used in the simulation are summarized as
Table 1. In this table, to simplify the problem, we use 64 as
D | Eb AJ (cos(I J ( k )) cos(I J (k 1))) VI1 (k ) VI1 (k 1)
2
(35) the number of sub-carriers instead of 52 in the 802.11a
AJ standard. Hence the occupied bandwidth is changed from 16.6
(cos(I J (k ) I J (k 1)))
Eb MHz to 20 MHz correspondingly.
Compared with the case of coherent BPSK (22), there are five Fig. 3 shows the comparison between simulation results and
noise terms instead of two. Approximately, the MTJ is theoretical results of all non-optimal jamming types (BNJ, the
2 1 / USIR times larger than that of BPSK and AWGN is 2 fixed-fraction PBJ and the fixed-fraction MTJ) in the paper.
times larger than that of BPSK, which gives a simple way of In the simulation, every test is repeated 200-1000 times to
getting differential modulation BER from coherent modula- eliminate the fluctuations caused by intrinsic random nature of
tion BER Eq. (30). the OFDM communication system. It is shown that the
SIR (36) simulation results of all non-optimal jamming types are in
PDBPSK ( Eb , N 0 , U , SIR) PBPSK ( Eb ,2 N 0 , U , )
1
(2 )
USIR Table 1: Main parameters used in simulation
This equation is valid only for SIR >> SNR. From simulation, Modulation
Signal bit rate 20 MHz BPSK/DBPSK
scheme
we found that when SIR is close to SNR, the simulation Number of
values of BER will deviate from theoretical values to some 64 Cyclic prefix 0.8 us
sub-carriers
smaller values. In order to compensate this deviation, Eq. FFT length 64
Channel Rayleigh fading
(36) was modified empirically to model channel with AWGN
Doppler OFDM
SIR (37) 40 Hz 3.2 us
PDBPSK ( Eb , N 0 , U , SIR) PBPSK ( Eb , N 0 , U , ) frequency symbol period
1 Signal Jamming Depend on different
(2 0.6) 20 MHz
USIR bandwidth bandwidth jamming techniques
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on May 28, 2009 at 21:22 from IEEE Xplore. Restrictions apply.
agreement with the analytical prediction perfectly under BNJ are compared in Fig. 5. Under AWGN, the optimal MTJ
AWGN. On the contrary, in Rayleigh fading channel, there clearly outperforms the other two for BPSK and DBPSK.
are some small deviations between simulation results and When the channel condition degenerates to the worst case
theoretical values. They are caused by precision errors of channel model — Rayleigh fading channel, BNJ performs best
numerical integration (for all non-optimal jamming types) and since BNJ is the special case of PBJ, whose jamming fraction
equation approximation (only for DBPSK of MTJ). All these equals to 1. From observation it is noticed that even BNJ is
deviations are less than 10%, therefore they are acceptable. best in Rayleigh fading channel, its advantage over optimal-
To verify the optimization process about PBJ and MTJ, fraction MTJ is not so obvious. So it is reasonable to believe
optimal jamming fraction data is listed as Table 2, in which that the optimal-fraction MTJ can be used to obtain improved
SNR is fixed to 20db and SIR is varied from -2db to 10db jamming effect under different channel conditions with low
under AWGN. We show both analytical predictions and complexity.
simulated values for PBJ and MTJ except MTJ for DBPSK, to
which only simulated values are shown since precise theoreti- 5. Conclusion
cal equation is hard to obtain. Since 64-FFT is used in the The BER performance of different jamming strategies for
proposed OFDM system, every U * has been rounded to the OFDM system is investigated. Both analytical form and
simulation values are given. In addition, two new jamming
closest integer multiple of 1/64. In Table 2, the bold part is
methods — optimal-fraction PBJ and optimal-fraction MTJ
the analytical prediction and its right side is the corresponding
are proposed in this paper. Through analysis and simulation, it
simulated results. The biggest error between them is less than
is shown that under the best channel condition (AWGN only),
5%, which validates the correctness of the analytical model.
the optimal-fraction MTJ clearly outperforms other jamming
Fig. 4 shows the comparison between the optimal-fraction
types in the paper, and as the channel condition gets worse
jamming and the fixed-fraction jamming of PBJ and MTJ. It is
into deep fading, the optimal-fraction MTJ still shows
revealed that under AWGN only, the optimal-fraction
competitive performance. The results of the experiment and
jamming always gives the best jamming effect. On the
the analysis of those results show that the optimal-fraction
contrary, this can not be promised in Rayleigh fading channel.
MTJ is a very effective jamming technique for OFDM system
In fact, in Fig. 4 (c), (d), (g) and (h), it is found that the best
in various channel conditions.
jamming effect can be achieved by just setting jamming
fraction to 0.9 simply. Thus in deep fading channel, the
jamming power should be distributed to the whole bandwidth REFERENCES
to gain the best jamming effect. On the other hand, from Fig. [1] A. Goldsmith, Wireless Communications, Cambridge University
4, it is found that the optimal-fraction jamming also performs Press, 2005.
[2] R. F. Ormondroyd and E. Al-Susa, “Impact of multipath fading
quite well even in Rayleigh fading channel. Therefore the
and partial-band interference on the performance of a
optimal-fraction jamming can obtain optimal performance COFDM/CDMA modulation scheme for robust wireless com-
under AWGN (best case channel model) and relatively good munications,” IEEE MILCOM, vol. 2, pp. 673-678, 1998.
performance under Rayleigh fading channel (worst case [3] H. Zhang and Y. Li, “Anti-jamming property of clustered
channel model). In general, the optimal-fraction jamming OFDM for dispersive channels,” IEEE MILCOM, vol. 1, pp.
gives us a simple way to obtain good jamming effect under 336-340, Oct. 2003.
[4] J. Park, D. Kim, C. Kang and D. Hong, “Effect of partial band
various channel conditions.
jamming on OFDM-based WLAN in 802.11g,” ICASSP 2003,
Finally, optimal-fraction MTJ, optimal-fraction PBJ and vol. 4, pp. 560-563, Hongkong, China, 6-10 April 2003.
Table 2: Optimal jamming fraction (SNR = 20dB) [5] S. Lijun, T. Youxi and L. Shaoqian, “BER Performance of
Frequency Domain Differential Demodulation OFDM in Flat
SIR U * U * U * U * U * U * U *
(dB) Fading Channel,” GLOBECOM, vol. 1, pp. 1-5, 2003.
PBJ PBJ PBJ PBJ MTJ MTJ MTJ
BPSK BPSK DBPSK DBPSK BPSK BPSK DBPSK
[6] Y. R. Zheng and C. Xiao, “Improved models for the generation
(A1) (S2) (A) (S) (A) (S) (S) of multiple uncorrelated Rayleigh fading waveforms,” IEEE
-2 1 62/64 1 62/64 1 1 1 Communications Letters, vol. 6, no. 6, pp. 256-258, 2002.
-1 57/64 57/64 1 1 51/64 48/64 63/64 [7] A. Papoulis, Probability, Random Variables and Stochastic
0 45/64 46/64 1 61/64 40/64 41/64 55/64 Processes, 3rd Edition, McGraw-Hill, New York, Feb. 1991
1 36/64 35/64 51/64 52/64 32/64 32/64 43/64 [8] N. Kingsbury, “Approximation Formulae for the Gaussian Error
2 29/64 31/64 40/64 41/64 25/64 24/64 32/64 Integral, Q(x),” Connexions, June 7, 2005.
3 23/64 26/64 32/64 32/64 20/64 20/64 27/64 [9] L. Hanzo, M. Münster, B. J. Choi, T. Keller, OFDM and MC-
4 18/64 21/64 25/64 25/64 16/64 15/64 21/64 CDMA for Broadband Multi-User Communications, WLANs
5 14/64 15/64 20/64 21/64 13/64 12/64 16/64 and Broadcasting, Wiley-IEEE Press, September 2003.
6 11/64 11/64 16/64 17/64 10/64 11/64 12/64 [10] IEEE 802.11a, “Part 11: Wireless LAN Medium Access Control
7 9/64 9/64 13/64 14/64 9/64 8/64 10/64 (MAC) and Physical Layer (PHY) Specifications: High-speed
8 7/64 7/64 10/64 10/64 6/64 7/64 8/64 Physical Layer in the 5GHz Band,” supplement to IEEE 802.11
9 6/64 6/64 8/64 8/64 5/64 5/64 6/64 Standard, Sept. 1999.
10 5/64 4/64 6/64 7/64 4/64 4/64 5/64
* (A1): Analytical * (S2): Simulated
* This table is generated under AWGN channel without Rayleigh fading
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on May 28, 2009 at 21:22 from IEEE Xplore. Restrictions apply.
0 0
10 10
-1
10
-2 -1
BER
BER
10 10
-3
10
-4 -2
10 10
-2 0 2 4 6 8 10 -2 0 2 4 6 8 10
SIR SIR
(a) BNJ under AWGN (b) BNJ under Rayleigh fading channel
0 0
10 10
-1
10
-1
BER
BER
10
-2
10
-3 -2
10 10
\ -2 0 2 4 6 8 10 -2 0 2 4 6 8 10
SIR SIR
(c) PBJ under AWGN ( is fixed to 0.5) (d) PBJ under Rayleigh fading channel ( is fixed to 0.5)
0 0
10 10
-1
10
-2 -1
BER
BER
10 10
-3
10
-4 -2
10 10
-2 0 2 4 6 8 10 -2 0 2 4 6 8 10
SIR SIR
(e) MTJ under AWGN ( is fixed to 0.5) (f) MTJ under Rayleigh fading channel ( is fixed to 0.5)
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on May 28, 2009 at 21:22 from IEEE Xplore. Restrictions apply.
0 0
10 10
-1
10
-1
10
-2
BER
BER
10
-2
10
-3
10
-4 -3
10 10
-2 0 2 4 6 8 10 -2 0 2 4 6 8 10
SIR SIR
(a) PBJ under AWGN (BPSK) (b) PBJ under AWGN (DBPSK)
0 0
10 10
-1 -1
BER
BER
10 10
-2 -2
10 10
-2 0 2 4 6 8 10 -2 0 2 4 6 8 10
SIR SIR
(c) PBJ under Rayleigh fading channel (BPSK) (d) PBJ under Rayleigh fading channel (DBPSK)
0 0
10 10
-1
10
-1
10
-2
BER
BER
10
-2
10
-3
10
-4 -3
10 10
-2 0 2 4 6 8 10 -2 0 2 4 6 8 10
SIR SIR
(e) MTJ under AWGN (BPSK) (f) MTJ under AWGN (DBPSK)
0 0
10 10
-1 -1
BER
BER
10 10
-2 -2
10 10
-2 0 2 4 6 8 10 -2 0 2 4 6 8 10
SIR SIR
(g) MTJ under Rayleigh fading channel (BPSK) (h) MTJ under Rayleigh fading channel (DBPSK)
: Optimal : =0.1
: =0.9 : =0.5
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on May 28, 2009 at 21:22 from IEEE Xplore. Restrictions apply.
0 0
10 10
-1
10
-1
10
-2
BER
BER
10
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on May 28, 2009 at 21:22 from IEEE Xplore. Restrictions apply.