Overview of Ultra Wide Band (UWB)
Impulse Radio
UMIACS/LTS Seminar
March 17, 2004
Dr. Jay Padgett
Senior Research Scientist
Telcordia Technologies/ Applied Research
Wireless Systems and Networks
jpadgett@telcordia.com
What is Ultra Wide Band?
Ÿ FCC definition of UWB signal: bandwidth is at least 20% of
center frequency, or a bandwidth of at least 500 MHz
Ÿ Bandwidths can exceed 1 GHz
Ÿ Example: Pulsed UWB signal (pulse width: 0.5 ns).
Pulse energy spectrum
0
Signal is a sequence of short pulses
Φ(f) / Φmax (dB)
2
-10
-20
τ = 0.5 ns
-30
-40
-50
0
1
2
3
4
f (GHz)
Ÿ The large UWB passband spans the operating frequencies of
many existing licensed services
Ÿ Interference concerns: UWB to narrowband systems (main
concern of FCC, NTIA), narrowband-to-UWB interference (UWB
design issue)
5
3
Brief FCC / UWB History
Ÿ FCC motivation: encourage development of potentially useful
new technology without causing harmful interference to existing
services and devices
Ÿ Sequence of FCC Notices and Orders
– Notice of Inquiry in September 1998
– Notice of Proposed Rule Making, June 2000
– First Report and Order, April 2002: made legal unlicensed operation
of UWB devices subject to certain restrictions
– Reconsideration Order and Further NPRM, March 2003: responsive
to Petitions for Reconsideration filed in response to the first R&O
Ÿ FCC categorizes its approach to legalizing UWB “conservative”
with respect to preventing harmful interference to other devices
and services. FCC has promised further evolution to UWB
Rules
Ÿ Allowed UWB emission levels are less than or equal to the level
allowed for unintentional radiators such as computers and other
electronic devices (-41.3 dBm/MHz).
FCC General UWB Emission Limits
UWB emission limit in dBm/MHz
4
-40
−41.3 dBm
1.99 GHz
UWB Passband (3.1 to 10.6 GHz)
-50
−51.3 dBm
−53.3 dBm
-60
−61.3 dBm
−63.3 dBm
-70
Indoor
Handheld
−75.3 dBm
-80
960
MHz 1.61 GHz
1
2
3
4
5
6
7
8
Frequency in GHz
9
10
11
12
5
Applications: DARPA’s NETEX Program
Objective: Develop military UWB sensors and communications
systems for operation in extreme environments
Ÿ Phase I: Gain understanding of effects of UWB system
operation on existing narrowband spectrum users
– Testing (NAVAIR/Pax River): 16 legacy system, 39 modes, 1600
tests completed.
– Modeling/Simulation (Telcordia) details below
– UWB propagation and channel modeling (Virginia Tech)
Ÿ Phase II: (DARPA/ATO BAA 03-20, Industry Day April 7 ‘03)
– Push UWB physical layer design to the point where it is capable of
reliably supporting advanced LPD, ranging, location and networking
protocols
– Develop algorithms, protocols, and distributed control for robust,
scalable ad hoc UWB networking
– Demonstrate 20-node hand held radios, 50-node video network, and
50-node radar sensor integrated into high data rate short range
network.
http://www.darpa.mil/ato/solicit/NETEX/documents.htm
6
UWB: An Undesired Result
7
UWB Interference Modeling
victim receiver or
measurement instrument
(e.g., spectrum analyzer)
impulse response
receiver
(baseband equivalent)
response
UWB impulse train
x(t ) =
g (t )
∞
∑ g (t − T )
k = −∞
y (t )
hb (t )
k
?
t
What is the effective
interference power due
to the UWB signal, as
seen by the non-UWB
receiver?
t
H(f )
G( f )
f
f
8
Victim Receiver Model and Interference Filtering
f0
fIF1
RF
~
fIF2
IF1
~
LO1
Detector/
Demodulator
Circuitry
IF2
LO2
Receiver Passband
UWB Pulse
Energy Spectrum
−f0
f0
IF Impulse Response
(envelope shown)
UWB Pulse Sequence
HIF(f )
f
Time Hopping PSD Example: Partial-Frame 2-ns Dithering
-10
-20
filter output power, dBm
9
-30
-40
M = 10 time hopping bins
εc = 2 ns bin separation
Puwb = 10 dBm
R = 10 MHz pulse rate
BIF = 1 MHz IF bandwidth
-50
-60
0
1
2
3
f0 (filter center frequency), GHz
4
5
Closeup – with Simulation Results
10
-10
filter output power, dBm
-20
-30
-40
-50
-60
1.4
M = 10
εc = 2 ns
Puwb = 10 dBm
R = 10 MHz
BIF = 1 MHz
Model
Simulation
1.5
f0 (filter center frequency), GHz
1.6
Simulation – Zero Span – 1.46 GHz
-10
-20
-30
-10
-20
filter output power, dBm
filter output power in dBm
11
-40
-50
-30
-40
-50
-60
1.4
M = 10
εc = 2 ns
Puwb = 10 dBm
R = 10 MHz
BIF = 1 MHz
Model
Simulation
M = 10, εc = 2 ns, Puwb = 10 dBm
R = 10 MHz, BIF = 1 MHz
1.5
1.6
f0 = f1.46
GHz
(filter center frequency), GHz
0
-60
0
20
40
60
time in microseconds
80
100
Simulation – Zero Span – 1.47 GHz
12
-10
-30
-10
-20
filter output power, dBm
filter output power in dBm
-20
-40
-30
-40
-50
-50
-60
1.4
M = 10
εc = 2 ns
Puwb = 10 dBm
R = 10 MHz
BIF = 1 MHz
Model
Simulation
M = 10, εc = 2 ns, Puwb = 10 dBm
R = 10 MHz,
BIF = 1 MHz
1.5
1.6
(filter center
frequency), GHz
f0 = f1.47
GHz
0
-60
0
20
40
60
time in microseconds
80
100
Simulation – Zero Span – 1.48 GHz
-10
-20
-30
-10
-20
filter output power, dBm
filter output power in dBm
13
-40
-30
-40
-50
-50
-60
1.4
M = 10
εc = 2 ns
Puwb = 10 dBm
R = 10 MHz
BIF = 1 MHz
Model
Simulation
M = 10, εc = 2 ns, Puwb = 10 dBm
R = 10 MHz, BIF = 1 MHz
1.5
1.6
f0 =f 1.48
GHz
(filter center frequency), GHz
0
-60
0
20
40
60
time in microseconds
80
100
Simulation – Zero Span – 1.5 GHz
14
-10
-30
-10
-20
filter output power, dBm
filter output power in dBm
-20
-40
-30
-40
-50
-50
-60
1.4
M = 10
εc = 2 ns
Puwb = 10 dBm
R = 10 MHz
BIF = 1 MHz
Model
Simulation
M = 10, εc = 2 ns, Puwb = 10 dBm
R = 10 MHz, BIF = 1 MHz
1.5
1.6
f0 =f (filter
1.5center
GHz
frequency), GHz
0
-60
0
20
40
60
time in microseconds
80
100
Example: PSD of a Swept-PRF Signal
15
r(t)
r (t ) = r + at
T
T
− <t≤
2
2
a = 2rd T
v (t ) =
t
r(t)
pulse rate
M
∑ δ (t − T )
n
n = − M +1
V(f )=
M
∑e
− j 2πfTn
n = − M +1
rd
r
T
−
2
T
2
t
Tn =
(
− r + r 2 + 2a n + T 2 4
a
)
Swept PRF PSD and Approximation
16
4
Swept PRF
100 MHz average PRF
100 kHz sweep rate
4 MHz PRF span (98 to 102 MHz)
1 watt total average power
1.05 GHz pulse bandwidth (rectangular spectrum)
milliwatts
3
exact
first-order analytic approximation
2
1
0
0
200
400
600
frequency (MHz)
800
1000
Closeup of Swept PRF Spectrum
17
1.0
Swept PRF
100 MHz average PRF
100 kHz sweep rate
4 MHz PRF span (98 to 102 MHz)
1 watt total average power
1.05 GHz pulse bandwidth (rectangular spectrum)
milliwatts
0.8
0.6
0.4
0.2
exact
first-order analytic approximation
0.0
580
590
600
frequency (MHz)
610
620
Individual Tones of the Swept PRF PSD
18
1.0
Swept PRF
100 MHz average PRF
100 kHz sweep rate
4 MHz PRF span (98 to 102 MHz)
1 watt total average power
1.05 GHz pulse bandwidth (rectangular spectrum)
milliwatts
0.8
0.6
exact
first-order analytic
approximation
0.4
0.2
0.0
585
586
587
588
589
590
591
frequency (MHz)
592
593
594
595
19
UWB Effects on Receiver Performance
Ÿ N u is the ratio of the pulse rate to the IF bandwidth
Ÿ There are 3 primary interference cases:
– N u ≤ 1, which gives pulsed interference to the detector
– N u > 1 with constant pulse rate, which can give a tone
within the passband
– N u > 1 with dithering or modulation, which can give a
noise-like signal
Ÿ The first two cases were explored for fixed-frequency
digital communications receivers and for an analog FM
receiver. The third case, in the limit, gives results similar to
those of Gaussian noise, which are well-known.
Ÿ With Nu > 1, there can also be a combination of a tone and
noise (as seen previously).
Ÿ Results are calculated based on the average UWB
interference power within the receiver passband.
Modeling the Test Procedure for an FM Receiver
40
Nu = PRF ÷ IF bandwidth
Nu = 0.1
30
Baseband SINAD (dB)
20
Nu = 1.0
Nu = 0.01
Gaussian noise
20
∆f = 2.4 kHz
fm = 1 kHz
BIF = 30 kHz
Bbb = 3 kHz
10
0
-10
0
10
20
Carrier-to-interference ratio (CIR), dB
30
Analysis vs. dB-Average Test Results (FM)
CIR for 10 dB SINAD receiver upset, dB
21
20
Gaussian
noise
15
10
5
0
dB average from measurements
analysis
-5
1
2
3
4
5
Test Waveform (TW) Number
6
UWB into a Noncoherent FSK Receiver
0
10
Nu=0.01
Nu=0.02
Nu=0.05
-1
Nu=0.10
10
BER
22
Nu=1.00
-2
10
-3
10
Equivalent AWGN BER
1
 SIR 
exp  −

2
2


-4
10
-20
-10
0
10
SIR (dB)
20
30
Coherent PSK – Comparison of Rates
10
10
10
BER
23
10
10
10
10
0
Blue : N u = 0.05
Green : N u = 0.1
-1
Red : N u = 0.25
-2
-3
-4
-5
-6
-5
I/N = -30 dB
I/N = 0 dB
I/N = 30 dB
0
5
SINR (dB)
10
At low SINR, BER ≅ N u 2
At low BER, SINR ≅ −10 log N u
15
UWB/Legacy NB Coexistence Example
24
duwb
Maximum UWB path distance (meters)
1000
f0 = 300 MHz
Buwb = 200 MHz
dnb = 20 meters
Fnb = 10 dB
Fuwb = 1 dB
(Eb/N0)uwb = 10 dB
A(f0) = 0 dB
GTX,uwb = 2 dB
GRX,uwb = 2 dB
GTX,uwb->nb = 2 dB
GRX,nb = 2 dB
LTX = LRX = 1 dB
Dithered pulses
800
600
400
I/N = 0 dB
I/N = 5 dB
I/N = 10 dB
I/N = 15 dB
I/N = 20 dB
200
0
1
10
100
Rb
UWB data rate (kb/s)
1000
Aggregate Interference – Closed form CDF
99
P { Z < z } [percent]
Distribution of normalized
aggregate interference power
25
90
Nearest single interferer
70
50
30
γ = 4.0
γ = 3.5
γ = 3.0
10
1
Multiple interferers
0.1
-10
0
10
20
10 log z
• Uniformly random distribution of interfering transmitters
• Equal power
• No exclusion zone
1 ∞ Γ(kν )  Γ(1 − ν ) 
FZ ( z ) = Pr (Z < z ) = 1 − ∑
sin kπ (1 − ν ) , z > 0
ν


π k =1 k!  z

k
(
FZ ( z ) = exp − z −ν
)
z > 0 (nearest interferer)
30
These curves show the average
percentage of handsets blocked
on the downlink as a function of
UWB deployment density and
transmit power spectral density
Probability density functions for minimum
UWB to PCS handset distance
Pb (percent)
blocking probabilityaveraged over cell area
and UWB-to-PCS distance
IS-95 PCS Blocking vs. UWB Density
100
80
Non-uniform distance distribution
γ = 3.5
ΦTXuwb = −50 dBm/MHz
60
−55 dBm/MHz
40
−60 dBm/MHz
20
−65 dBm/MHz
0
0.00
0.7
0.05
0.10
0.15
0.20
Truncated uniform (d0 = 1 meter)
Uniform
Non-uniform
0.5
0.4
ρ = 0.1 active UWB
transmitters/m2
0.3
0.2
0.1
0.0
0
1
2
3
r (meters)
4
5
6
Pb (percent)
blocking probabilityaveraged over cell area
and UWB-to-PCS distance
ρ (active UWB transmitters/m2)
0.6
fd(r)
probability density function for
distance to nearest UWB transmitter
26
100
80
Uniform distance distribution
γ = 3.5
ΦTXuwb = −50 dBm/MHz
60
−55 dBm/MHz
40
−60 dBm/MHz
−65 dBm/MHz
20
0
0.00
0.05
0.10
0.15
ρ (active UWB transmitters/m )
2
0.20
27
C/I with Aggregate Interference
Ÿ Desired signal to NB receiver Rayleigh-faded (terrestrial
multipath).
Ÿ UWB transmitters uniformly randomly distributed spatially, no
fading of interfering signals (short distances).
Ÿ May be an “exclusion zone” surrounding victim receiver within
which there will be no UWB transmitters.
Rayleigh-faded signal path
Desired NB transmitter
narrowband receiver
dmin
exclusion zone
(no UWB transmitters
within this area)
C/I CDF with Aggregate UWB Interference
and no Exclusion Zone
28
99.99
γ = 2.5
γ=3
γ = 3.5
γ=4
99.9
Pr(Cnb/IUWB < Λ), percent
99
Monte Carlo, γ = 3
Nearest interferer, γ = 3
90
70
50
30
10
2


γ
 Cnb



 2 
Λ

 Γ1 − 
P r
< Λ  = 1 − exp − 

I
X
 γ 
 UWB

  nb 

1
0.1
-40
-30
-20
-10
0
γ >2
10
20
Λ
, dB
X nb
normalization : X nb =
Cnb
γ
αPtx (πρ a )
2
29
The Larger Picture: UWB – Challenges,
Questions, and Opportunities Going Forward
Ÿ Design of UWB radios that can operate adaptively in the
presence of strong in-band narrowband signals.
Ÿ UWB networking: MAC layer design, ad hoc routing protocols,
QoS management, especially with narrowband interference.
Ÿ Applications – what is UWB best suited for (military and
commercial)?
– Radar/localization
– Data networking?
– Sensors?
– Integrated applications (e.g., localization + communications).
Ÿ Can UWB be used to provide high data rate hotspots with
connectivity to CMRS networks?
Ÿ Can UWB be used to increase the accuracy of E-911 indoors?