0% found this document useful (0 votes)
52 views55 pages

Lecture Notes 6 5510

The document discusses satellite modem design. It describes the key components and operations of a satellite modem, including modulation, upconversion, transmission over the satellite channel, downconversion, and demodulation. It discusses the challenges of timing recovery, carrier recovery, and operating under time varying mobile satellite channels. It provides examples of modern satellite modems used for applications like internet access, video conferencing, and mobile communications.

Uploaded by

Chentao Yue
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
52 views55 pages

Lecture Notes 6 5510

The document discusses satellite modem design. It describes the key components and operations of a satellite modem, including modulation, upconversion, transmission over the satellite channel, downconversion, and demodulation. It discusses the challenges of timing recovery, carrier recovery, and operating under time varying mobile satellite channels. It provides examples of modern satellite modems used for applications like internet access, video conferencing, and mobile communications.

Uploaded by

Chentao Yue
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 55

ELEC5510 Satellite

Communication Systems
Lecture 6
Satellite Modem Design

Dr. Wanchun Liu


Outline

 Why Modulate?
 Complex Signal Representation of MPSK Modulation
 Continuous Time Modem Operations
 Discrete Time Modem Operations
 DSP Modem Implementations
 Timing Recovery Methods
 Carrier Recovery Methods
 Examples of Satellite Modem Products
Satellite Communication Systems

Earth
station
antenna
Data Channel Modu- Power
source encoder lator amplifier Uplink
channel

Satellite
Error
Channel
control Modem Multiple
Antennas propagation
coding access

Earth
station Downlink
antenna channel

Data Channel Demodu Power


sink decoder -lator amplifier
Satellite Earth Station Transceivers

DVBS2 Modem
Data rate up to 380Mbps
Up to 64APSK and 9/10 FEC

Modem for Internet Services

Modem for Video Conferencing, Surveillance and Elections


4
Satellite Mobile Transceivers

BGAN Modems
Land mobile
Maritime
TV broadcasting
M2M
Aeronautical
5
Satellite Mobile Phones

Thuraya dual phone Inmarsat Iridium Satellite


IsatPhone Phone

6
Why We Need Modulation?

 Baseband signal does not propagate through the air for a


long distance.
 Satellite channels can only pass signals within fixed
frequency limits defined by frequency allocation for a given
service.
 In order to send base-band signals over band-pass
satellite channels, modulation is needed prior to
transmission.
 A most commonly used modulation technique for satellite
communication is MPSK.

7
M-PSK Modulation
 Varying phase according to the message
s (t )  A cos(c t    k )
A A
 cos k 2 cos(c t   )  sin k 2 sin(c t   )
2 2
 I k 2 cos(c t   )  Qk 2 sin(c t   )
2
k  k ; k  0,1, 2,...M  1;(l  1)Ts  t  lTs ;
M
l  0,1, 2,...; c  2 f c
A A
Ik  cos k , Qk  sin  k
2 2

s (t )  Re ( I k  jQk ) 2e j (ct  )  8
M-PSK Demodulation

The received signal is multiplied by the coherent carrier


s (t ) 2 cos(c t   )
 2 I k cos 2 (c t   )  2Qk sin(c t   ) cos(c t   )
 I k (1  cos 2(c t   ))  Qk sin 2(c t   )
 Ik (Low pass filtering)

 s (t ) 2 sin(c t   )
 2 I k sin(c t   ) cos(c t   )  2Qk sin 2 (c t   )
  I k sin 2(c t   )  Qk (1  cos 2(c t   ))
 Qk (Low pass filtering)
9
Complex Signal Representation

 Modulation

s (t )  ( I k  jQk )e j (ct  )
~ modulate

0 fc f
 Demodulation
~  j (c t  )
s (t )e
demodulate
j (c t  )  j (c t  )
 ( I k  jQk )e e
 I k  jQk
0 fc f

10
Modulator and Demodulator
Modulator

 The mapper converts the input binary symbols into complex


valued numbers.
 The real and imaginary parts are In-phase (I) and Quadrature
(Q) signals.
 The signal spectrum is shaped by base-band filters.
 The up-converter translates the filtered signals to the carrier
frequency. The I and Q signals modulate two orthogonal
carriers which are added for transmission.

12
Demodulator
 The received signal is down-converted into I and Q base-
band signals.
 They are filtered to remove noise outside the message
bandwidth. The overall transmit and receive filter response
satisfies the Nyquist ISI minimization criterion.
 The signal is then sampled every T seconds, where T is the
symbol period duration (Note T= Ts).
 The sampled signals are passed to a decoder.
 For coherent demodulation the receiver needs to know the
carrier phase and frequency.
 When Nyquist filters are used the optimum sampling instants
are at the symbol midpoints.

13
Challenges in Demodulator

 Assume a time offset of  seconds for the symbol


midpoint and phase offset of  radians in the modulator.
 Due to the channel effects, the main tasks of the
demodulator include
 Estimate the carrier frequency fc -- carrier frequency
sync
 Estimate the phase offset ˆ -- carrier phase sync
 Estimate the time offset ˆ -- timing sync

14
15
Mobile Satellite Modem Design

 Modems on satellite mobile channels need to operate


under the following conditions
- time varying channels
- high Doppler shifts
- low signal-to-noise ratios
- burst mode operation due to time division multiple
access (TDMA) used in satellite communications
 Current satellite mobile modems operate with data rates of
up to several hundreds of Kbps and fixed satellite modems
up to several hundreds of Mbps. This allows the use of
digital signal processors (DSP) in modem design.

16
A Continuous-Time Modem

 a  (t  T   )

17
Baseband Modulator Operations

 The complex baseband modulated symbol at time lT is


denoted by al = Il + jQl.
 The impulse symbol sequence can be expressed as
a(t )   a (t  T   )

where  is a time offset and T is the symbol period.
 a(t) is filtered by a square root Nyquist filter with impulse
response hT(t). c(t )  a(t ) * hT (t ) (1)
where  represents the convolution operation.
 By expanding the convolution we obtain for the filter output

c (t )    (t  T   )
a h T

  
(2)

18
Up-conversion Modulator Operations

 The up-conversion operation can be represented as

s (t )  Re{c(t ) 2e j (ct  ) } (3)

where ωc = 2πfc is the carrier angular frequency.

 By using (2), the transmitted signal is given by

 
s (t )  Re  a hT (t  T   ) 2e j (ct  ) 
  

19
Down-conversion Demodulator Operations

 The received signal in an AWGN channel is

r (t )  s (t )  n(t )
where n(t) is white Gaussian band-pass noise.
 The received signal is first down converted to baseband

 j (c t  )
d (t )  r (t ) 2e

where ˆ is the carrier phase estimate at the receiver.

20
Baseband Real Value Demodulator
Operations

 The down-converted signal is filtered by a square root


Nyquist filter
R
d (t ) * h (t )
where hR(t) is the receive filter impulse response.
 The filtered signal is then sampled
R
[d (t ) * h (t )] t  nT ˆ

where ˆ is the sampling time estimate at the receiver.

21
Baseband Complex Value Demodulator
Operations
 In the absence of noise, the received signal r(t) is equal to the
transmitted signal s(t), given by (4) and the down-converter
output d(t) can be represented as
j (  )
d (t )  c (t )e  double frequency terms (4)

 Noise can be also represented in complex-valued notation as


n(t )  Re{z (t )e jct }
where z(t)= wI(t) + jwQ(t) and wI(t) and wQ(t) are Gaussian
processes with the same power spectral densities.
 The frequency translation operation in the receiver results in a
phase shift of 
    
22
Baseband Complex Value Demodulator
Operations
By substituting c(t) from (2) into (4) and adding the noise,
the filter output at the n-th sampling instant is
 z( t )  R
  a (t  T   ) * h (t )e
 j 
T
  * h (t ) t  nT ˆ
  2
 By replacing hT(t)hR(t) by the combined filter response he(t) and
expanding the convolution we get for the filter output

y n (  ,  )   
a h
 
e
[( n   )T   ]e  j 
 wn

 where   ˆ   and wn is a complex valued noise sample of


z (t ) * h R (t ) / 2 23
Baseband Complex Value Demodulator
Operations

 If there are no time and phase offsets


y n (  ,  )    [(n  )T ]  w n  an  w n
a h
 
e

since for a Nyquist filter


an

 1, i  0
h (iT )  
e

0, i  0

Ts 2Ts 3Ts nTs 24


Filtering

 Filters limit the signal bandwidth and remove noise


outside the signal bandwidth.
 Most commonly used filters are raised cosine Nyquist
filters.
 The transfer function of a Nyquist raised cosine filter is
T , | f | (1   ) / 2T

H (f )  T 
e
T (f  1/ 2T )  (1   ) 1 
 2 1  sin   ,
2T
| f |
2T
  
where T is the symbol period and α is the roll-off factor.
 The Nyquist transfer function is usually divided into two
square root Nyquist filters in the transmitter and
receiver.
25
Raised Cosine Nyquist Filters
rs
fc  (1   )
2
Transfer Function He(f)

rs is the symbol rate


rs=1/T fc  (1)
rs

Bandwidth B=(1+α) rs 2

he  t 
Impulse Response

T 2T 3T
26
Nyquist Filter Implementation
 Discrete time implementations are based on DSP hardware.
 DSP filters are mostly implemented as Finite Impulse Response
(FIR) filters.
 The coefficients of an FIR filter are the samples of the filter
impulse response. That is, the filter output yn is given by
L
yn   hi xn i
i 0
where L is the filter order, xn is the input (a sample) and hi are the
filter impulse response samples given by
hi  h RN (t ); t  iTsm , i  0, L
where h RN (t ) is a truncated impulse response of a root Nyquist
filter and Tsm is the sampling period Tsm =1/ fs and fs is the
sampling rate.

27
FIR Nyquist Filter

xn xn-L+2 xn-L+1 xn-L


Tsm Tsm Tsm

x h0
x x hL-1 x hL
hL-2

yn
+

28
Selecting Sampling Rate

 According to the Nyquist-Shannon theorem the sampling rate fs


should be

1
f   2W
s Tsm

where W is the maximum


frequency in the
baseband spectrum.
Nyquist Filter Implementation

 The sampling rate of the FIR Nyquist filter output:


From the Nyquist-Shannon sampling theorem
f  2W  2r (1   ) / 2
s s Nyquist
f  rs filtering
s r
(1   )
s
2
r is the symbol rate
s
Oversampling -2rs 0 2rs
f  M r  M / T , where f s  2rs
s ss s
M is the number of samples per symbol.
s
 Typically Ms is 4 and the impulse response of an FIR
root Nyquist filter extends over 7T. 30
Low IF Signal Interfaces

 For high speed modems quadrature mixers are


implemented as analog devices.
 In low speed modems the quadrature mixer can be
implemented with DSP devices at low intermediate
frequencies (IF) and later shifted to higher carrier
frequencies by an up-converter. The advantage of using a
low IF interface is in elimination of imperfections of analog
quadrature mixers, like nonlinearities and DC offsets.

31
DSP Modulators

32
DSP Modulator Operations

 The input to the filter has to be an impulse in order to get a


sinc-like Nyquist impulse response. The interpolator inserts
Ms – 1 zeros after each input symbol sample, so that the
total number of samples per symbol is Ms, where Ms is
typically 4.
 Frequency translation is performed by multiplying the
complex filter output signal by the exponential
2e j 2 f I t |t  nTsm  2e j 2 nf I / f s  2e jn
where fI is the IF frequency and fs is the sampling frequency
equal to rsMs.
 The result is a complex valued number. However, only the
real part is applied to the D/A converter and filtered to
remove high frequency components.
33
DSP Demodulators

34
DSP Demodulator Operations

 The received signal is down-converted to the low IF and


filtered. After Analog-to-Digital (A/D) conversion the
discrete signal is translated to the base-band in a DSP
down converter by multiplying it with the exponential
2e  j 2 f I t |t  nTsm  2e  j 2 nf I / f s  2e  jn
 Coherent demodulation requires that the receiver carrier in
the down converter is synchronous relative to the carrier in
the transmitter. The estimation of phase offset is called
carrier recovery.
 Due to Doppler shifts and local oscillator frequency drift,
the frequency of the receiver carrier must also be adjusted.
 The demodulator must determine correct symbol timing.

35
DSP Demodulator Operations

 It calculates the sampling time error and digitally shifts the


signal by this amount, to be able to sample the signal at its
peak.
 The process of selecting one sample and discarding the
other Ms – 1 samples is called decimation.
 The selected signal sample is then rotated by the negative
phase offset to correct phase error.
 The resulting sample is then passed to the decoder which
makes estimates of the transmitted binary symbols.

36
Timing Recovery

 The tasks of timing recovery is to sample Nyquist filter output


at its peaks to get the maximum SNR.
 We consider all digital timing recovery, where the signal is
sampled at the analog-to-digital (A/D) converter at a constant
sampling rate and a timing error is passed to the baseband.
 At the baseband a timing error, within one symbol period, is
estimated and corrected by an interpolator.
 The timing errors larger than one symbol are corrected by
frame synchronisation, based on transmitting and correlating
short known sequences.

37
Timing Recovery

Correct timing

T 2T

Timing error

 Timing error

T 2T

38
Timing Error Detection Methods

 The timing error signal n can be obtained by a variety of


timing error detection (TED) algorithms.
 Data-aided (DA) methods
- Maximum likelihood (ML) algorithm
 n  R e  x n a n*  1  x n  1a n* 
- Modified Mueller and Muller (mM&M) algorithm

 n  R e ( x n  x n  2 ) a n*  1  ( a n  a n  2 ) x n*  1 
where xn is the complex signal at the output of the
decimation unit and an is the complex decoder output
*
estimate and an is the signal complex conjugate.
 These methods could be used only for low error rates. 39
Timing Recovery by Interpolation

 If the symbol timing error is n, it can be represented as a


fraction μn of a sampling period Tsm
 μn =  n / T sm
 The signal can be shifted digitally to its correct position by
n = μn Tsm by an interpolation FIR filter.
 The interpolation filter computes the values of time
adjusted signal vn , from the input signal vn.
 A simple linear interpolation filter calculates
vn (nTsm )  vn (nTsm  nTsm )

when vn (nTsm ) and vn 1[(n  1)Tsm ] are known.


40
Timing Recovery by Interpolation

 A linear interpolator approximates the sinc curve


between vn (nTsm ) and vn 1[( n  1)Tsm ] by a straight line and
interpolates them to find vn (nTsm  nTsm )
The linear FIR interpolator has coefficients
cn ( n )  n , cn 1 ( n )  1   n

vn (nTsm ) vn 1[(n  1)Tsm ]


TTsm

X cn 1
cn vn (nTsm  nTsm )
X +
T
Timing Recovery by Interpolation

 A parabolic interpolator approximates the sinc curve


between vn-2(n Tsm), vn-1(n Tsm ), vn(n Tsm ) and vn+1[(n+1) Tsm ]
by a parabola and calculates by interpolation vn (nT  nT ) .
 The parabolic FIR interpolator coefficients are
cn  2 (  n )   n 2   n , cn 1 (  n )   n 2  (1   )  n
cn (  n )   n 2  (1   )  n  1, cn 1 (  n )   n 2   n
  0.5
 The cubic FIR interpolator coefficients are
c2 (  n )   n 3 / 6   n / 6, c 1 (  n )    n 3 / 2   n 2 / 2   n
c0 (  n )   n 3 / 2   n 2   n / 2  1, c1 (  n )    n 3 / 6   n 2 / 2   n / 3
42
Carrier Recovery

 Carrier recovery for coherent demodulation includes


estimation of phase and frequency of the receiver carrier
and adjusting it to the transmit carrier.
 The modulator can generate and transmit a discrete
component at the carrier frequency. The receiver then can
use a phase-locked loop (PLL) to track this component.
This method is power inefficient and PLLs are not suited
for DSP implementation.
 For M-PSK modulation schemes M-th power devices can
be used to produce the carrier frequency in the signal
spectrum and to track the carrier component by PLLs. M-th
power loops have phase ambiguity problems.
43
Carrier and Timing Recovery in Digital
Demodulator Implementations
 In digital demodulators the carrier in the down converter is
free running. The down-converted signal contains phase
and timing errors. These errors are estimated in the base-
band and subsequently corrected.
 There are two most important carrier recovery techniques:
data aided and nonlinear feed forward phase estimation.
 If there is a phase offset  in the receiver carrier relative
to the transmitted carrier, the bit error probability for
QPSK modulation becomes

 2 Eb 2 

Pb  Q  cos  
 No 
44
Data Aided Carrier Recovery

 The data aided method is based on maximum likelihood


estimation.
 The likelihood function is defined as the conditional
probability of receiving a signal r(t) over a given interval To
and for a set of parameters, such as carrier phase ,
symbol timing  and transmitted symbols al. The likelihood
function can be written as

Prob {r(t), t ≤To | , , for lT ≤ To}

 Signal phase can be estimated by maximising the


likelihood function with respect to the set of transmit
parameters, providing that the transmitted symbols are
known.
45
Data Aided Carrier Recovery

 In the data aided recovery method we assume that the


incoming symbols are known to the receiver and equal to
the decoded signals.
 By maximising the likelihood function we get for the phase
estimate at the sampling instant (n+1)

ˆn 1  ˆn   I m { yn an* }


where ˆn is the phase estimate at instant n,  is a small
positive constant, yn is the soft decision output from the
demodulator and an* is the complex conjugate of the hard
decision decoder output associated with yn .

46
Data Aided Carrier Recovery

 In block structured data transmission the performance of


the phase estimation can be improved by sending a
known preamble sequence at the beginning of each
block.
 When the phase offset ˆ is determined, the phase of the
n

signal xn is corrected by rotating xn by  ˆ . This is


n

equivalent to multiplying xn by e  j .
n
ˆ

47
Feed-forward Nonlinear Phase Recovery

 If the received M-PSK signal at sampling instant k is



j [( 2 n  1 )  ˆk ]
x k | x k | e M

where n=0,1,2,… M-1. Raising it to the M-th power we get


j [( 2 n 1)  Mˆk ]
| xk | e
M M
xk
 All received symbols after raising to the M-th power are
shifted to the same point in the signal set.
j (  M ˆk )
z  xk M
| xk | e
M


ˆk  angle( z )  , where angle is the phase of z
M 48
Feed-forward Nonlinear Phase Recovery


The average phase offset av can be estimated over a
number of received symbols in order to remove the
effect of noise. The phase estimate is given by

1  N
M  
av  angle    xk   

M  k  N  M
where 2N + 1 is the number of symbols over which the
phase is estimated and an g le is the average angle
over 2N+1 symbols.

49
Feed-forward Nonlinear Phase Recovery

 xk M

50
Feed-forward Nonlinear Phase Recovery

 The above operation causes an M-fold ambiguity in the


phase estimate. That is, the soft output demodulator
output could be rotated by any multiple of 2/M radians.
 Unique words known to the receiver can be used to
remove phase ambiguity by searching over all permuted
versions of the unique word.

51
Australian Mobilesat Modem Characteristics

 Used for voice, fax, data GPS and messaging


 Modulation type: /4 QPSK
 Bandwidth: 5 KHz
 Modem mounted in a vehicle or temporarily fixed
 Raised cosine Nyquist filters with 40% roll-off
 Carrier frequency: 1646.5 to 1600.5 MHz uplink and 1545.0 to 1559.0
MHz downlink
 IF frequency: 71.15MHz
 Carrier recovery: Nonlinear feedforward
 Timing recovery: Decision directed
 Phase ambiguity resolved by a unique word transmitted in each frame
of 120ms
 DSP implementation of baseband and IF functions 52
Optus Mobilesat Modem

53
Inmarsat Satellite Mobile Transceivers

54
Iridium Satellite Phones

55

You might also like