INSTITUT D’ÉLECTRONIQUE ET DE … D’ÉLECTRONIQUE ET DE TÉLÉCOMMUNICATIONS DE RENNES Carrier...
Transcript of INSTITUT D’ÉLECTRONIQUE ET DE … D’ÉLECTRONIQUE ET DE TÉLÉCOMMUNICATIONS DE RENNES Carrier...
INSTITUT D’ÉLECTRONIQUE ET DE TÉLÉCOMMUNICATIONS DE RENNES
Carrier per Carrier Analysis of SDR Signals Power Ratio
Christophe MOYAssociate Professor at Supélec
SDRF Technical Conference,
Orlando, USA, November 2006
1UMR6164
SUPELEC - Campus de Rennes - FRANCE
SCEE team – Signal, Communication and Embedded Systems
IETR – UMR 6164
Institute of Electronics and Telecommunications of Rennes
Authors: Sidkiéta ZABREJacques PALICOTYves LOUETChristophe MOYChristian LEREAU
INSTITUT D’ÉLECTRONIQUE ET DE TÉLÉCOMMUNICATIONS DE RENNES
Ideal Software Radio
• Ideal software radio (SWR)– digitize the globality of the RF spectrum at Rx– generate all the channels of a multi-standard
BTS with a single DAC and RF amplifier at Tx
GSM DCS UMTSPCS cellular
Bluetooth
Home RF
SDRF Technical Conference’06 - November 2006 - Orlando 2
- -
82
48
49
86
98
94
88
0-9
15
92
5- 9
60
-
17
01-
17
85
18
05-
18
80
- -
18
50
19
10
19
30
19
90
-
19
20
19
80
211
02
17
0
-2
40
02
50
0
-5
15
05
72
0 frequences [MHz]
GSM
900DCS
1800
UMTS
IEEE 802.11a
Example ofa futuremini-BTSon-board atransatlantic
plane
PCS
1900
cellular
US
Home RF
WIFI : IEEE 802.11b
Hyperlan
INSTITUT D’ÉLECTRONIQUE ET DE TÉLÉCOMMUNICATIONS DE RENNES
Software Defined Radio
• Technlology constraints– ADC, DAC speed and quality – processing power of processors, FPGA… – power consumption – analog components bandwidth and linearity…
SDRF Technical Conference’06 - November 2006 - Orlando 3
• Restricted to sub-parts of the spectrum– channels of one standard for instance
Spectrum of a multiplex of GSM CarriersSpectrum of a multiplex of GSM Carriers
925.000 925.200 925.400 925.600 925.800
… …
INSTITUT D’ÉLECTRONIQUE ET DE TÉLÉCOMMUNICATIONS DE RENNES
SDR RF signal and PR issue
• A SDR RF signal at least, and moreover a Software Radio RF signal– multi-band signal– multiplex of carriers
• Major drawback of a multiplex of carriers
SDRF Technical Conference’06 - November 2006 - Orlando 4
• Major drawback of a multiplex of carriers– suffer from high Power Ratio (PR)– known also as the PAPR (Peak-to-Average
Power Ratio)
• Cause– non constant envelop of the signal
INSTITUT D’ÉLECTRONIQUE ET DE TÉLÉCOMMUNICATIONS DE RENNES
PR issue
• The difference between the average power and the max power is high – either under use the amplifier (input back-off)– or generate non linearities
• Non linearities1.8
2output power
SDRF Technical Conference’06 - November 2006 - Orlando 5
• Non linearities– area 1: linear– area 2: non linear– area 3: saturation
• Signal corruption• BER degradation 0 0.5 1 1.5 2 2 .5 3 3 .5 4
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
1 2 3
input power
INSTITUT D’ÉLECTRONIQUE ET DE TÉLÉCOMMUNICATIONS DE RENNES
Facing PR issue
• PR is an old issue– 70’s on satellite communications– 80’s till now: OFDM (DVB-T, DAB, 802.11g)
• OFDM signal– a multiplex of carriers
SDRF Technical Conference’06 - November 2006 - Orlando 6
– a multiplex of carriers– very close to SWR and SDR signal– particularly suffers from PR issue
• Idea: take advantage of solutions investigated for OFDM and apply to SDR signals
INSTITUT D’ÉLECTRONIQUE ET DE TÉLÉCOMMUNICATIONS DE RENNES
OFDM and PR
• OFDM PR definition– based on a time PR calculation– on the samples after IFFT (all generated from
a single time signal)– definition
{ }2
2
,
)(
)(
tSE
tSMaxPR mc =
SDRF Technical Conference’06 - November 2006 - Orlando 7
{ }2,
)(tSEmc
• SDR approach– a multiplex of signals on several carriers– each generated independently from the others– a frequency approach would be more suitable– each standard one by one
INSTITUT D’ÉLECTRONIQUE ET DE TÉLÉCOMMUNICATIONS DE RENNES
Time PR vs frequency PR
• Computing the PR in the frequency domain would better fit to SDR signal – SDR signals are a set of signals at different
carriers– compute the global PR from each carrier PR
SDRF Technical Conference’06 - November 2006 - Orlando 8
• Three questions– Is it equivalent to compute the PR in the time
and the frequency domains?– Is it correct to apply it to SDR signals?(Q1)
(Q2)
(Q3)– How to compute or approximate the PR in the frequency domain?
INSTITUT D’ÉLECTRONIQUE ET DE TÉLÉCOMMUNICATIONS DE RENNES
Carrier per carrier analysis
• Multi-standard SDR signal
( ) ( ) tfiP
p
pii
pi
i
etrtS ,2
1
,
π
∑=
=
• SDR single-standard signal made of a plurality of equidistant carriers
SDRF Technical Conference’06 - November 2006 - Orlando 9
analogy( ) ∑−
=
=1
0
2N
k
tfi
kkeCtZ
π
( ) ( ) ( ) tppfiP
p
piiii
i
etrtS∆−+
=
∑=)1(2
1
,0,π
plurality of equidistant carriers
• OFDM signal
(Q1)
INSTITUT D’ÉLECTRONIQUE ET DE TÉLÉCOMMUNICATIONS DE RENNES
OFDM signal
ft1
OFDM symbol 1
CNsN-1CNs1CNs0
C2N-1C21C20
C1N-1C11C10
PRfr(0)
fN-1f1f0
t
f
PR
PRfr(1) PRfr(N-1)
CNsN-1CNs1CNs0
C2N-1C21C20
C1N-1C11C10
CNsN-1CNs1CNs0
C2N-1C21C20
C1N-1C11C10
PRfr(0)
fN-1f1f0
t
f
t
f
PR
PRfr(1) PRfr(N-1)
SDRF Technical Conference’06 - November 2006 - Orlando 10
t
OFDM symbol 2
OFDM symbol NS
t2
tNs
PRNs,freqPRNs,freq
f0 fp fN-1
INSTITUT D’ÉLECTRONIQUE ET DE TÉLÉCOMMUNICATIONS DE RENNES
Demo for an OFDM signal
• PR upper-bound usuallycomputed in the timedomain: PRtemp
– on each OFDMsymbol (line) CNsN-1CNs1CNs0
C2N-1C21C20
C1N-1C11C10
PRfr(0)
fN-1f1f0
t
f
PR
PRfr(1) PRfr(N-1)
CNsN-1CNs1CNs0
C2N-1C21C20
C1N-1C11C10
CNsN-1CNs1CNs0
C2N-1C21C20
C1N-1C11C10
PRfr(0)
fN-1f1f0
t
f
t
f
PR
PRfr(1) PRfr(N-1)
PRtemp(2)
PRtemp(Ns)
SDRF Technical Conference’06 - November 2006 - Orlando 11
PRNs,freqPRNs,freq
PRfr(0)
PRfr(p)
PRfr(N)
• New calculation of PR– frequency PR on each column: PRfr(p) then max
– then the max PRNs,freqis computed from the set of time PR
• We find the same PRNs,freq(Q2)
INSTITUT D’ÉLECTRONIQUE ET DE TÉLÉCOMMUNICATIONS DE RENNES
•10•-1
PR upper-bounding
• PR approximation in the frequency domain – probabilistic
∑
∑=
+
≤N
N
p
frP
temp
KpPRA
PR1
)(.
•10•0
•Upper bound
•Temporal PR
OFDM sig
nal
SDRF Technical Conference’06 - November 2006 - Orlando 12
•0 •2 •4 •6 •8 •10 •12 •14•10
•-3
•10•-2
•10
– P[PRtemp>threshold](complementary cumulativedistribution fonction)
∑=
N
p
pA1
S(dB)
•Pro
b(P
ME
PR
>S
)
obtained by the new PRfreq approach
classical approach
OFDM sig
nal –
64 ca
rriers –
1000 sy
mbols
(Q3)
INSTITUT D’ÉLECTRONIQUE ET DE TÉLÉCOMMUNICATIONS DE RENNES
Example on a GSM signal
• Time PR and upper-bound– 10 GSM carriers– 1000 symbols– 20 experiments
•
7
8
9
10
Temporal PR
Upper bound
• Results
SDRF Technical Conference’06 - November 2006 - Orlando 13
0 2 4 6 8 10 12 14 16 18 200
1
2
3
4
5
6• Results
– only 2 dB upper-bound
– perfectly followstime PR variations
INSTITUT D’ÉLECTRONIQUE ET DE TÉLÉCOMMUNICATIONS DE RENNES
Conclusion
• This new PR approach (in the frequency domain) may help in the SDR domain for – applying OFDM reduction methods to SDR
signals – for the PR evolution analysis through a
SDRF Technical Conference’06 - November 2006 - Orlando 14
– for the PR evolution analysis through a transmission chain
• Equivalence for SDR single-standard• Perspective
– to be proven for multi-standard SDR signals