Vous êtes sur la page 1sur 7

Comparison among FB techniques

ICI and ISI


SMT and CMT are more prone to frequency offset and timing jitter than FMT. But with ideal channel conditions SMT and CMT show zeros ISI and ICI. CMT requires RRC filter at transmitter and receiver for ICI free transmission when used for multi-carrier communication. Furthermore, all these techniques use equalization even in ideal channel conditions. Algorithms for blind detection in CMT can be developed.

Spectral Efficiency
CMT is most efficient as compared to SMT and FMT. Since no guard interval is used in CMT and SMT. FMT uses virtual carriers for better DAC performance. This results in some loss in spectral efficiency.

Resolution
Considering -6db bandwidth, we have following resolutions for FBSE methods.
CM T 0
X: 0 .5 8 7 1 Y : -6 .0 3 1 X: 0 .5 9 9 3 Y : -6 .0 9 5

-20

-40

-60

-80

-100

-120

-140 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Fig 1: CMT resolution

smt 0 -10 -20 m agnitude, dB -30 -40 -50 -60 -70 -80 -90
X : 0 .5 8 9 8 Y : - 6 .0 0 9 X : 0 .5 9 7 3 Y : - 5 .9 9 7

0.1

0.2

0.3

0.4 0.5 0.6 frequenc y , H z

0.7

0.8

0.9

Fig 2: SMT resolution.


F M T fo r E = 1 0 -10 -20 in d B -30 -40


X : 0 .6 2 5 8 Y : -5 .9 9 5

X : 0 .6 3 6 3 Y : -5 .9 9 7

M -60 -70 -80 -90

Fig 3: FMT (1) resolution

   

-50

0.1

0.2

0.3

0.4 F

0.5

0.6 , Hz

0.7

0.8

0.9

F M T fo r E = 0 .5 0 -1 0 -2 0 M a g n itu d e in d B -3 0 -4 0 -5 0 -6 0 -7 0 -8 0
X : 0 .5 8 1 3 Y : - 5 .9 9 X : 0 .6 0 2 Y : -6

0 .1

0 .2

0 .3

0 .4 0 .5 0 .6 F re q u e n c y , H z

0 .7

0 .8

0 .9

Fig 4: FMT(0.5) resolution.

CMT = 0.0122 SMT = 0.0075 FMT (0.5)=0.0207 FMT (1) = 0.0105 High resolution for SMT is misleading. With increasing power of the signal, side-lobe power also rise and lead to bad detection performance.

Probability of detection and false alarm:


Plots for probability of detection and false alarm are given below.

Fig 5: Probability of detection and false alarm.

From here we can see that good resolution of SMT is of little use, since its probability of false alarm is very unreliable. The reason for this is the increased side lobe power with increasing SNR. This side-lobe power is always undesirable.

Spectral containment ( mean out of band power)


It means how much of signal power is contained within the main lobe. Larger main lobe width or smaller side lobe width often results in good spectral containment. Clearly CMT seems to have smaller main lobe width but it then decays faster. (from fig combined plot and mathematical figure should be inserted)

Hardware complexity
Table shows the hardware complexity among FBSE techniques. CMT shows good complexity results.
Table 1: Hardware complexity of differect Estimation scheme.

For N = 128 Periodogram Modified Periodogram Blackman-Tukey method SMT CMT FMT ( = 1) FMT ( = 0.5)

Multipliers 3840 4096 28928 8448 2566 4224 8449

Adders 3712 3712 28415 8193 2310 4096 8193

Buffers 64 -

Comparison among PSE, Modified PSE, BTSE


SDR (Spectral containment)
It represents the power of the first side-lobe. BTSE is best among these three followed by Modified PSE and PSE

0 P eriodogram M odified P eriodogram w it h k ais er w indow B lac k m an-Tuk ey -10

-20

-30 m agnit ude, dB

-40

-50

-60

-70

-80

10

15

20

25 frequenc y , H z

30

35

40

45

50

Fig 6: PSE, Modified PSE and BTSE combined plot.

Resolution
BTSE = 0.0397 MPSE = 0.0251
M odified Periodogram w K aiser w ith indow 0 -10 -20 M agnitude in dB -30 -40 -50 -60 -70 -80 0
X : 0.587 Y : -5.9 9 9 X : 0.6131 Y : -5.9 9 8

0.1

0.2

0.3

0.4 0.5 0.6 F requency in H z

0.7

0.8

0.9

Fig 7: Modified PSE resolution.

B la c k m a n -Tu k e y S p e c tra l E s tim a tio n w ith K a is e r w in d o w o f B e ta = 4 0 -5 -1 0 M a g n itu d e in d B -1 5 -2 0 -2 5 -3 0 -3 5 -4 0


X : 0 .5 8 0 3 Y : - 5 .9 9 1 X : 0 .6 2 Y : - 6 .0 0 2

0 .1

0 .2

0 .3

0 .4 0 .5 0 .6 F re q u e n c y in H z

0 .7

0 .8

0 .9

Fig 8: BTSE resolution.

Probability of detection and False alarm


Probability of detection is better in case of PSE followed by MPSE and BTSE in order. For probability of false alarm, order is reverse.
P r o ba bilit y 1 0 .9 0 .8 0 .7 0 .6 Pd 0 .5 0 .4 0 .3 0 .2 0 .1 0 -3 0 P e rio d o g ra m M o d i fi e d P e r i o d o g r a m B la c k m a n -T u k e y o f d e t e c t io n , P d , w it h c o n s t a n t n o is e p o w e r

-2 5

-2 0

-1 5 S NR in

-1 0 d B

-5

Fig 9: Probability of detection.


P r o ba bilit y 0 .3 5 P e rio d o g ra m M o d i fi e d P e r i o d o g r a m B la c k m a n -T u k e y o f fa l s e d e t e c t io n , P f, w i t h c o n s t a n t n o is e p o w e r

0 .3

0 .2 5

0 .2 Pf 0 .1 5

0 .1

0 .0 5

0 -3 0

-2 5

-2 0

-1 5 S NR in

-1 0 d B

-5

Fig 10: Probability of false alarm.

Hardware Complexity
From table we can see that BTSE has more complex implementation than other two.

Conclusion
FBSE techniques give good performance while keeping complexity low at the same time. But these can only be better used with equalization at receiver if we are using it for both detection and transmission. Furthermore, CMT can be used to develop blind detection algorithms. A hybrid technique needs to be developed for accumulating maximum good properties of the existing techniques.

Vous aimerez peut-être aussi