Sei sulla pagina 1di 3

Doppler Shift Compensation Techniques for LEO Satellite On-board Receivers

Umair Naeem*, Zeeshan Jawaid* and Salman Sadruddin* Space and Upper Atmosphere Research Commission (SUPARCO), Karachi, Pakistan

Abstract-In
compensating

this the

paper

two

techniques

employed in

for

Doppler

shift

encountered

LEO

To Bascb'nd Demodulator

satellite receivers are presented. These include the analog compensation technique involving the carrier tracking loop and the compressive receiver technique. The compressive receiver technique which involves precise detection of the rapidly time varying carrier frequency received by the satellite receiver is discussed in further detail. A compressive receiver with a center frequency of 100 kHz MATLAB to test the technique.

and a Doppler shift up to 100

kHz is simulated on

I.

INTRODUCTION

In LEO satellite channels, the relative motion of the satellite with respect to ground station results in significant Doppler shifts. The drift in the received frequency causes many problems including band overlap, loss of available spectrum, higher bit error rate (BER), etc. Therefore, for reliable communications, this Doppler shift should be accurately estimated and compensated. Traditionally, work has been done on Doppler correction techniques which involve reducing the Doppler shift by applying a linear decreasing frequency sweep to the received signal [1]. However, this requires prior information of the famous S-shaped Doppler-time curves [2], [3]. Therefore, recent work has been directed towards the real time estimation and compensation of Doppler shift from the received frequency without the need of Doppler characterization [4]. This paper first introduces an analog compensation technique. Although this technique does not require any prior information of the Doppler-time curves, but there are still some difficulties and drawbacks in the implementation of this technique. Next section describes the technique for Doppler correction which makes use of the principle of compressive receivers to estimate the Doppler shift introduced in the received signal. The solution presented in this technique assumes that the RF front-end of the receiver has already mixed the received signal down to the baseband frequency of 100 kHz so that it is easily handled by the DSP. The simulation results are then presented and finally, last section highlights the important conclusions drawn from this paper.
II. ANALOG COMPENSATION TECHNIQUE

Fig. I Block diagram of the double-conversion heterodyne receiver employing the analog compensation technique.

The analog compensation technique is explained with reference to the double-conversion heterodyne receiver [5] shown in Fig. 1.

This technique is based on the principle that the 2nd intermediate frequency (IF) stage is always maintained at a fixed frequency f2 Hz, irrespective of any Doppler shift in the received RF signal. This is done by tuning the two local oscillators, LOI and L02 in a phase locked carrier tracking loop. A Doppler shift fD causes the input RF frequency to vary in the range fRF fD' neglecting the frequency shift due to frequency stability of uplink LO. The frequency variation is translated down to the 2nd IF stage and thus, the 2nd IF varies in the range f2 fj, 2f where fj,f2 is a function of fD' The phase of this dynamically varying IF signal is then compared with a fixed frequency crystal oscillator (XO). The phase comparator outputs an R.M.S. error voltage, Ve, which is proportional to the error fj,f2 ,between its two input frequencies. This voltage signal is low pass filtered to produce a D.C error control voltage, Vc which, in turn, controls a voltage controlled temperature compensated crystal oscillator (VCTCXO). The VCTCXO is used to generate the reference frequency, fref' for the two local oscillators. By applying the control voltage to the VCTCXO, the two LOs are shifted in such a way that the final IF returns back to f2 Hz, thus cancelling the effect of Doppler shift. The expression for the maximum frequency pulling, fj,fref' of the VCTCXO required to cancel a peak Doppler shift of fD can be found by equating the two frequencies at the input of the phase comparator. For this, we need to find the frequencies at various points inside the control loop. The frequency at the output of VCTCXO is:

f VCTCXO

fref fj,fref

(1)

The frequencies generated by the two local oscillators, LO1 and L02, are:

fLOl
978-1-4577-1929-5/12/$26.00 2011 IEEE

IcTCXO

Nl

Rl

(2)

Proceedings of 2012 9th International Bhurban Conference on Applied Sciences & Technology (IBCAST) Islamabad, Pakistan, 9th - 12th January, 2012

391

(3) lL02 = IVCTCXO X Nl N2 are the division ratios of LOI and L02 where l and R R2
-

N2 R2

respectively. The frequency at the output of first mixer, i.e. 151 IF is:

where and At the output of second mixer the 2n IF is:

11 = IRF - lLOl

Nl /j.ll = d - /j. fe, x - . Iv Rl N2 . R2

(4)

1lF2 = where 12 = 11 - lL02 1lF2 12'

E. /j. {

(5)

and /j./- /j. fe, x The two frequencies at the input of the phase comparator and By equating these frequencies, the are maximum required frequency pulling of the VCTCXO is found to be: (6)

/j.12 =

these compressed pulses. First, the input signal is applied to a mixer whose LO is a frequency modulated (FM) signal. The mixer's output is then passed through the DDL to compress it into narrow RF pulses. Let us assume that the frequency versus time curve of the FM signal is linear, i.e. it changes frequency from to in the time interval to with a frequency versus time slope of as shown in Fig. 3(a). If the frequency versus time slope of the DDL isas shown in Fig. 3(b), the output from the DDL is a burst of energy at the time instant This phenomenon occurs because the leading edge of the signal which enters the DDL at with a frequency of is delayed by while the trailing edge of the signal entering the DDL at with a frequency of is delayed by Thus, both the leading and trailing edges of the FM signal come out from the DDL at t2resulting in the entire FM pulse from to being compressed at time The bandwidth of the DDL equals the frequency difference and the dispersive delay time of the DDL equals the time difference

12

tl t2

11

m,

tl + t2. 12

11

t2,

tl

t1

t2

t2

tl +

tl + t2.

tl

12 - 11'

t2 - t1.

f
f, ..

Equation (6) suggests that a high pulling range of the VCTCXO is required for a large Doppler shift. High pulling range VCTCXO has its own design constraints and is quite difficult to design. Moreover, the division ratios of the two LOs also have a direct effect on the pulling range. Besides this, the design of PLLs also depends on the division ratios. Thus, careful selection of these ratios should be done in order to achieve an optimized design. Furthermore, the phase detector threshold parameter has a great effect on the dynamic performance of this kind of multi loop. Phase detector threshold is defined as the signal to noise ratio (SNR) at the phase comparator input below which it ceases to function effectively. This means that the RF power budgeting of the receiver should be done very carefully in order to maintain an SNR which is well above the phase detector threshold. Lastly, this analog technique adds a lot of circuit complexity to the overall design of the receiver. However, still it may not be as accurate as desired to cancel a large Doppler shift. Due to the above stated reasons, a simpler and comparatively more accurate technique, which makes use of the principle of compressive receivers, is presented in the next section. III. COMPRESSIVE RECEIVER TECHNIQUE

:71
I,

(a)

(b)

Fig.3. Input FM signal and matched DDL. Frequency versus time for (a) the input FM signal and (b) the matched DDL.

Before we start the mathematical analysis, let us assume that the DDL has a bandwidth B and a dispersive delay time /j.T. At the mixer's output, a pulsed FM signal with constant amplitude can be expressed as [6]:

Set)

2 = cjot+ mt )1
o

or

1 /j.T < t < /j.T}


elsewhere

(7)

where Wo is the center angular frequency of the DDL and m is the frequency versus time slope of the FM signal and is given by:

2rrB m = I1T

(8)

The block diagram of a compressive receiver [6] is shown in Fig. 2.


Compressed output pulse

The exponential form of Equation (7) can be written as :

Set)

/j.T t { e x[(wot+mt2)] for-elsewhere /j.T}


< < o

(9)

The transfer function of the DDL can be expressed as :

H(w)
Fig.2. Block diagram of compressive receiver.

=e

xp

(r:o)2]

(10)

The signal at the output of the DDL is given by :

In a compressive receiver, the input radio-frequency (RF) signal is compressed to a narrow pulse by using a dispersive delay line (DDL). The frequency of the input signals can be determined by measuring the position of

G(w)

/qw)S(w)

(11)

where Sew) is the Fourier transform of Se ) . In order to find the time domain output of the DDL, i.e.g ( ) , the

Proceedings of 2012 9th International Bhurban Conference on Applied Sciences & Technology (IBCAST) Islamabad, Pakistan, 9th - 12th January, 2012

392

inverse Fourier transfonn of G(w) is needed. By doing some complex integral manipulations which are given in detail in [6], the final time domain expression for g(t)comes out to be:

get) =!J. rffi

exp

(wot - mt2 )l sinc(m!J.Tt) (12)


+

A quick glance on (12) leads to the conclusion that the compressive receiver generates a pulse with many side lobes in the time domain for one input signal. The position of the main lobe is then measured to determine the frequency of the input signal. Once the frequency of the input carrier is measured, it is then compensated rendering a constant frequency signal which can then be processed for demodulation and bit synchronization. The simulation results of this technique are presented in the next section. IV. SIMULATION

index of the main lobe represents the frequency point of the input signal. The index of the reference of 100 KHz has already been calculated for comparison. The input signal's frequency index is then subtracted from the reference index to calculate the Doppler shift. Once the Doppler shift is calculated, it can be removed from the input signal by multiplying it with the input signal. The output is passed through a band pass filter to remove any unwanted sidebands. This model was run on Simulink and it was found that, for ideal conditions a Doppler shift of up to 100KHz for a carrier frequency of 100 kHz can be compensated. The Simulink model of the Doppler shift compensation technique is shown in Fig. 6.

To validate the proposed technique, a compressive receiver was simulated on MATLAB Simulink. The block diagram is shown in Fig 4:
InJiut timc varying frequency __

---, _ ___+{

Fig.6. MATLAB Simulink model of the implemented compressive receiver for Doppler shift compensation.

V.

CONCLUSION

(
I I

Fig.4. Block diagram of the implementation of compressive receiver for Doppler shift compensation.

The centre frequency of the input test signal is considered to be 100 KHz with a variable Doppler shift. Frequency range of the sweeping LO is taken as 200 KHz. Therefore, ideally a Doppler shift of 100KHz can be compensated. The test signal and the sweeping LO are mixed together and the output is passed through the DDL. The output of the DDL is a compressed pulse as shown in Fig. 5.

Doppler shift compensation techniques in the analog and digital domains are discussed and compared. It has been demonstrated that Doppler compensation can be done in the digital domain with reasonable accuracy and low complexity, thus eliminating the need for complex analog compensation techniques. The digital technique that is implemented makes use of a compressive receiver to estimate the incoming carrier frequency. The proposed technique has been simulated and verified on MATLAB. Further work includes the implementation of the technique on a digital signal processor with large Doppler shifts in noisy environments.
REFERENCES

[I]

[2]

G. P. Ah-Thew, Doppler Compensation for LEO Satellite Communication Systems, Open Access Dissertations and Theses, McMaster University, Canada, 1998. Irfan Ali,Pierino G. Bonanni, Naofal AI-Dhahir, John E. Hershey,
Doppler Applications in LEO Satellite Communication Systems,

09 08 0.7 06 05

[3]

[4]

[5] [6]

o
03 02 01

Newyork: Kluwer Academic Publishers, 2002. Irfan Ali, Naofal AI-Dhahir, John E. Hershey, "Doppler Characterization for LEO Satellites," IEEE Transactions on Communications, vo1.46, No.3, pp. 309-313, March 1998. Ewald van der Westhuizen and Gert-jan van Rooyen, "Baseband Carrier Recovery and Phase Tracking as a Doppler Compensation Technique for a zero-IF SDR," SATNAC Conference on Access Networks, 2009. David M. Pozar, Microwave and RF Design of Wireless Systems, New York: John Wiley & Sons, 2001. James Bao-Yen Tsui, Microwave Receivers with Electronic Warfare Applications, New York: John Wiley & Sons, 1986.

tN12

67loC 67]6 6.133 6.7 6 .1

,J DJI.

.7

6746 6.1018 6.75'

Iff

Fig. 5. Compressed output pulse of DDL.

Since the output is in frequency domain, every frequency point is represented by a unique index. The

Proceedings of 2012 9th International Bhurban Conference on Applied Sciences & Technology (IBCAST) Islamabad, Pakistan, 9th - 12th January, 2012

393

Potrebbero piacerti anche