[ieee 2010 7th workshop on positioning, navigation and communication (wpnc) - dresden, germany...
TRANSCRIPT
Maximal Operating Distance Estimation for
Ranging in IEEE 802.15.4a Ultra Wideband
Thomas Gigl⋆•, Florian Troesch◦, Josef Preishuber-Pfluegl•, and Klaus Witrisal⋆
{thomas.gigl, witrisal}@tugraz.at, [email protected], [email protected]⋆Signal Processing and Speech Communication Laboratory, Graz University of Technology, Austria
•CISC Semiconductor, Design and Consulting GmbH, Austria◦Communication Technology Laboratory, Swiss Federal Institute of Technology (ETH) Zurich, Switzerland
Abstract—The IEEE 802.15.4a standard for wireless sensornetworks is designed for highly accurate ranging with impulseradio Ultra Wideband (UWB) signals. The standard is designedfor coherent and non-coherent (energy detector) receivers, thusthe choice of complexity is left to the implementer. In thispaper, the maximal operating distance is analyzed for ranging,using a coherent receiver and an energy detector under FCC/ECregulations. The analysis is based on the working points of thereceivers and a link budget calculation. The preamble of thestandard is defined by several parameters, which influence theallowed FCC transmit power, the free space losses and thus themaximal operating distance. It is shown which code and channelcombinations achieve the largest operating distance. The bestperformance is achieved by using long spreaded codes, wherecoherent receivers can achieve a maximal operating distance upto several thousand meters and energy detectors up to severalhundred meters.
I. INTRODUCTION
The IEEE 802.15.4a standard [1] is a standard for wireless
sensor networks with sub-meter ranging accuracy even in
indoor environments. The physical layer is defined for bi-
directional communications based on impulse radio ultra-
wideband signals. A two-way time-of-arrival ranging scheme
is proposed. The standard allows the usage of high complexity
coherent receivers and low-complexity energy detectors. The
used preamble sequences show perfect autocorrelation prop-
erties for both receiver types [2].
Energy detectors promise low-cost and low-power consump-
tion. On the other hand, a performance loss [3] and more
vulnerability to interfering signals occur [4]. Sub-Nyquist
sampling is applicable for energy detectors, but limits on the
other hand the ranging performance [5]. The IEEE 802.15.4a
standard has a large number of selectable system parameters,
which are important for the achievable ranging performance
[6]. Their impact is studied in a signal-to-noise (SNR) analysis
for a coherent receiver and an energy detector. The influence
of the preamble parameters is also studied with respect to
the maximal allowed transmit power according to the FCC
[7] and EC [8] regulations. A detailed analysis about the
FCC regulations for specific signaling schemes can be found
in [9]. In this work, these concepts have been adopted to
the preamble design of IEEE 802.15.4a. A detailed analysis
shows the best preamble parameter setting for maximal al-
lowed transmit energy. Furthermore, the different frequency
Hilbert∑
w(t)∣∣ . ∣∣ Ranging
e−ωct ci
1T
××
Fig. 1: Coherent receiver architecture
(.)2∫TI
1TI
×
ci ∑w(t) Ranging
Fig. 2: Energy Detector
channels are discussed to optimize the system performance.
The performance is analyzed based on the maximal operating
distance, which is based on a link budget calculation according
to [10].
The rest of the paper is organized as follows: Section II
gives an overview of the introduced receiver architectures,
while Section III presents the signal model of the IEEE
802.15.4a standard, the received signal and the two receiver
architectures. The SNR analysis and the performance metric
definition is given in Section IV. The FCC and EC regulations
are discussed in Section V and the link budget is introduced
in Section VI. This is followed by performance results and a
short conclusion in Sections VII and VIII.
II. RECEIVER SYSTEM MODELS
This section shows the system models of the coherent
receiver and the energy detector.
A. Coherent Receiver
Fig. 1 shows the coherent receiver. The signal is received
by a UWB antenna and is filtered by transmit pulse shape
w(t). Thus, a matched filtering to the pulse shape is applied.
The signal is converted to discrete complex baseband using
the Hilbert transform, the carrier frequency ωc and Nyquist
sampling with 1/T . Next, the channel response is obtained by
code despreading, and the ranging algorithm is applied on the
magnitude of the complex baseband signal. A similar receiver
structure can be found in [11] and [12].
This receiver architecture suffers from the high sampling
rates according to the Nyquist theorem. Another disadvantage
978-1-4244-7157-7/10/$26.00 ©2010 IEEE 10
TABLE I: Preamble characteristicNs L Npr Tchip [ns] Tpr [µs] PRF [MHz] MRF [MHz] ERF [MHz] N1ms
31 16 16 2.0032 15.897 31.2 16.1 0.256 100631 16 64 2.0032 63.590 31.2 16.1 1.024 100631 16 256 2.0032 254.358 31.2 16.1 4.096 100631 16 1024 2.0032 1017.436 31.2 16.1 16.1 100631 16 4096 2.0032 4069.744 31.2 16.1 16.1 1006
31 64 16 2.0032 63.590 7.8 4.03 0.256 25131 64 64 2.0032 254.359 7.8 4.03 1.024 25131 64 256 2.0032 1017.433 7.8 4.03 4.03 25131 64 1024 2.0032 4069.744 7.8 4.03 4.03 25131 64 4096 2.0032 16278.974 7.8 4.03 4.03 251
127 4 16 2.0032 16.282 124.8 62.89 1.024 982127 4 64 2.0032 65.128 124.8 62.89 4.096 982127 4 256 2.0032 260.512 124.8 62.89 16.384 982127 4 1024 2.0032 1042.051 124.8 62.89 62.89 982127 4 4096 2.0032 4168.205 124.8 62.89 62.89 982
of this concept is the required synchronization of the carrier
frequency and the phase, which can harm the performance.
The energy detector is able to cope with these two problems,
thus a low complexity solution can be obtained.
B. Energy Detector
The energy detector works as shown in Fig. 2. The signal is
again received by a UWB antenna and filtered by a bandpass
filter, which is implemented as a filter matched to the pulse
shape. Next, the signal is squared and integrated for short time
periods TI , which also defines the sampling period. The length
of TI causes a mean absolute error (MAE) of at least TI/4[5], which limits TI to a few ns. The signal is correlated and
despread with the corresponding despreading code ci. This
delivers the channel response, where the ranging algorithm is
applied.
III. SIGNAL MODEL
The following notations are used: Column vectors are
denoted by boldface symbols and estimated values are marked
by hats.
A. IEEE 802.15.4a Preamble and Received Signal
The most important signal part for ranging is the preamble,
thus this section gives a short overview on the design of the
preamble. The IEEE 802.15.4a preamble sequence cs has a
length Ns of 31 or 127 [1]. It consists of ternary elements
∈ {−1, 0, 1}. The spread preamble code vector csp is created
as
csp = 1Npr⊗ cs ⊗ δL = c ⊗ δL (1)
where ⊗ denotes the Kronecker product, δL is the spreading
sequence of the preamble, which is defined as a unit vector
with a one at the first position and length L. 1Nprdenotes a
vector of ones with length of the preamble sequence repetitions
Npr and c is the periodically extended preamble code. The
transmitted signal s(t) is defined as
s(t) =ℜ{√
Ep
M−1∑m=0
cmw(t − mLTchip)ejωct
}
=√
Ep
M−1∑m=0
cm w(t − mLTchip)
(2)
where Ep is the energy per pulse, cm is the m-th element of c,
w(t) is the energy normalized pulse shape, M is the number
of code elements in the preamble, ωc is the carrier frequency,
Tchip is the chip duration, and w(t) is the up-converted pulse
assuming the carrier and the pulse are phase synchronous.
Table I shows the timing characteristics of the preamble,
where Tpr is the total duration of the preamble, PRF is
the peak pulse repetition frequency, MRF is the mean pulse
repetition frequency and N1ms is the number of preamble
sequences within 1 ms. ERF is the effective pulse repetition
frequency according to the regulations (see Section V).
The transmitted signal (2) is sent over a multipath channel
with channel impulse response hc(t), where also the effects of
the antenna are contained for simplicity. Furthermore, hc(t) is
assumed to be constant during Tpr. Thus, the analog received
signal
ra(t) = s(t) ∗ hc(t) + ν(t) (3)
where ν(t) is modeled as additive white Gaussian noise and
∗ is the convolution. Next, the received signal enters the two
receiver architectures, as described in the next sections.
B. Coherent Receiver
First the signal is pulse matched filtered by w(−t), where
w(t) = w(−t) due to symmetry. It follows
rf (t) = s(t) ∗ hc(t) ∗ w(t) + ν(t) ∗ w(t) (4)
Next, the negative frequencies are canceled by a Hilbert
filter hhilb(t) and the complex baseband signal is obtained
by down-converting the signal with the estimated carrier
11
frequency ωc. The resulting baseband signal is given by
r[n] =(rf (t) ∗ hhilb(t)) e−j(ωct+ϕ)
=M−1∑m=0
cmh(nT − mLTchip)ej[ωc−ωc]nT + νl[n](5)
where h(t) is an equivalent channel response incorporating
the channel hc(t), the unknown carrier phase ϕ, the pulse√Epw(t) and the downconverted matched filter w(t). The
noise νl[n] is the band limited input noise νl[n] = ν[n] ∗w[n]in complex baseband. The carrier frequency is assumed to be
known in this analysis. Thus, (5) simplifies to
r[n] =M−1∑m=0
cmh[n − mLNchip] + νl[n] (6)
where h[n]=h(nT ). The number of samples within a chip
is defined by Nchip=Tchip/T . Assuming synchronization, the
estimation of the despread channel response h[n] is given by
h[n] =Npr−1∑
q=0
Ns−1∑m=0
cm r[n + (m + qNs)LNchip]
=M1h[n] +M−1∑m=0
cmνl[n + mLNchip]
(7)
where the despreading is first performed sequence-wise (∑
m)
and then over the sequence repetitions (∑
q). c2m = 1 for
the non-zero coded elements, it follows that∑
q and∑
m
is simple the number of non-zero code elements in the
preamble M1 = Ns+12 Npr. In other words, M1 is the number
of transmitted pulses.
C. Energy Detector
The signal model of the energy detector after squaring,
integration and sampling is given by
x[n] =
(n+1)TI∫nTI
(M−1∑m=0
cmg(t − mLTchip) + νf (t)
)2
dt (8)
where νf (t) is the passband filtered noise and the channel
response g(t)=√
Epφw(t) ∗ hc(t + mLTchip) and φw(t) is the
autocorrelation function of w(t). Thus, the estimated channel
response y[n] is obtained by despreading x[n],
y[n] =Npr−1∑
q=0
Ns−1∑i=0
ci x[n + iLNchip + q Ns LNchip]
=Npr−1∑
q=0
Ns−1∑i=0
ci
(n+1)TI+iTchip+qTs∫nTI+iTchip+qTs
(M−1∑m=0
cmg(t−mLTchip)
)2
dt
+2Npr−1∑
q=0
Ns−1∑i=0
ci
(n+1)TI+iTchip+qTs∫nTI+iTchip+qTs
M−1∑m=0
cmg(t−mLTchip)νf (t)dt
+Npr−1∑
q=0
Ns−1∑i=0
ci
(n+1)TI+iTchip+qTs∫nTI+iTchip+qTs
ν2f (t) dt
= yss[n] + ysν [n] + yνν [n](9)
where Nchip = Tchip/TI , is the number of samples in a
code frame, and the sequence duration Ts = NsLTchip. The
code despreading is performed sequence wise with∑
q and∑i with the despreading code ci. In contrast to the coherent
receiver the non-coherent receiver uses a different despreading
code than the spreading code, which shows also perfect
circular correlation properties for the squared sequences. The
despreading code c is created by squaring c and setting all
zeros to −1 [2]. As observable, the output of the energy
detector is dependent on a signal-by-signal yss[n], a linear
signal-by-noise ysν [n] and a quadratic noise-by-noise yνν [n]term.
IV. SNR ANALYSIS AND MAXIMAL OPERATING RANGE
ESTIMATION
The goal of a ranging system is to find the LOS component
in the channel response, due to the fact that a detection
of a multipath component may lead to very large ranging
errors. The LOS-SNR (LSNR) correlates very well with the
ranging performance [11]. This section shows the relation
between receiver input SNR, the energy of the despread LOS
component over the noise spectral density ELOS/N0, and the
receiver output SNR, written as LSNR. ELOS = M1E(1)LOS,
where E(1)LOS is the received energy for the LOS component
for a single pulse.
A. Coherent Receiver
If only the LOS component is taken into account, h[n] sim-
plifies to hLOS[n]=√
E(1)LOSw[n]∗w[n]ejϕ. Thus, (7) simplifies
to
hLOS[n] =M1
√E
(1)LOSφw [n]ejϕ
+Npr−1∑
q=0
Ns−1∑m=0
cmνl[n + (m + qNs)LNchip]
=ys[n] + yν [n]
(10)
where φw[n] is the autocorrelation of w[n], ys[n] is the signal
part and yν [n] is the noise part of the receiver output signal.
12
1) Expected value of hLOS[n]: The expected value of
hLOS[n] is given by
E{hLOS[n]
}= M1
√E
(1)LOSφw[n]ejϕ (11)
2) Expected value of |ys[n]|2: The expected value of the
signal power is given by
E{|ys[n]|2
}= E
(1)LOSM
21 φ2
w[n] = ELOSM1φ2w[n] (12)
3) Variance of hLOS[n]: The variance of hLOS[n] is defined
by the variance of the noise component in (10) and is given
by E{|yν [n]|2
}due to the zero mean noise process. Thus, it
follows
E{|yν [n]|2
}=
Npr−1∑q=0
Npr−1∑q′=0
Ns−1∑m=0
Ns−1∑m′=0
cmcm′
×νl[n+(m + qNs)LNchip]ν∗
l [n+(m′+q′Ns)LNchip] .(13)
The noise νl is uncorrelated for a spacing of LNchip, thus
it follows m′ = m, q′ = q and (13) simplifies to
E{|yν [n]|2
}= N0
M−1∑m=0
c2mφw[0] = N0M1φw [0] (14)
where φw[0] is the equivalent bandwidth of the matched
filter.
B. Energy Detector
The statistical analysis of the energy detector for ranging
in IEEE 802.15.4a has been derived in [6]. Here only the
final equations for the LOS component without inter-pulse-
interference are shown.
1) Signal-by-Signal Term yss[n]: The signal-by-signal term
is given by
yss[n] = E(1)LOSM1
(n+1)TI∫nTI
φ2w(t)dt (15)
2) Signal-by-Noise Term ysν [n]: The variance of the signal-
by-noise term is given by
var{ysν [n]} = E(1)LOS(Ns + 1)NprN0
(n+1)TI∫nTI
φ2w(t)dt (16)
where it is shown that this term depends on the pulse shape
and the received signal energy ELOS. Note, this term does not
occur in the coherent receiver and degrades the performance
of the energy detector.
0 10 20 30 40 50 60 70−10
0
10
20
30
40
50
60
ELOS
/N0 [dB]
LS
NR
[dB
]
ED: Npr
=16, ND=992
ED: Npr
=64, ND=3968
ED: Npr
=256, ND=15872
ED: Npr
=1024, ND=63488
ED: Npr
=4096, ND=253951
CR: all Npr
Working Point ED
Working Point CR
Fig. 3: Relation between input SNR ELOS/N0 and out-
put SNR LSNR with respect to the noise dimensionality
ND = Ns Npr TI WRRC for the energy detector ED and the
coherent receiver CR. The static parameters are Ns = 31,
TI = 2.0032 ns, and WRRC = 1 GHz.
3) Noise-by-Noise Term yνν [n]: The variance of the noise-
by-noise term is given by
var{yνν [n]} =N2
0
2MTIWRRC (17)
where WRRC is an equivalent bandwidth defined as
WRRC =∫ Tw
−Twφ2
w(µ)dµ. A comparison to the coherent re-
ceiver (see (14)) shows that this variance scales with M instead
of M1, which means that approx. two times more noise occurs
due to the despreading method and is further proportional to
N20 .
C. Input-to-Output SNR Relation
The Input-to-Output SNR relation is given for the coherent
receiver, using (14) and (12), by
LSNRCR =|ys[nLOS]|2E{|yν [n]|2} =
ELOSφw [0]N0
=ELOS
N0(18)
as φw[0] is energy normalized, where nLOS is the sample
containing the LOS component at φw[0]. The energy detector
relation is obtained from (15)-(17) and is given by
LSNRED = 10 log2(
ELOS
N0
)2
4ELOS
N0+ NsNprTIWRRC
(19)
The first term of the denominator corresponds to the linear
and the second to the quadratic noise term. The quadratic
noise term depends on the receiver parameters, which can be
combined to the noise dimensionality ND=Ns Npr TI WRRC
[4].
Fig. 3 shows the relation of the detector input SNR
ELOS/N0 and the output LSNR based on (18) and (19). The
specific curves for the energy detector ED are obtained by
13
increasing ND by factors of 4, where the depicted curves corre-
sponds to Npr=[16, 64, 256, 1024, 4096], Ns = 31, TI = 2 ns,
and WRRC = 1 GHz, . Note, Npr = 256 is not intended in
the standard, but is shown for visibility in this analysis. The
curves are separated if the quadratic noise term dominates and
they merge if the linear noise term is dominant. Increasing
ELOS/N0 by 6 dB leads to LSNR +12 dB in the quadratic part
and to +6 dB in the linear one. The horizontal line illustrates
the LSNR at the working point LSNRWP = 12 dB for the
energy detector (cf. [6]), where 80% of the range estimations
are within 1 m. The coherent receiver shows a working point
LSNRWP = 9 dB (cf. [11]). Both working points are evaluated
by extensive simulations. It can be seen for the energy detector,
that 3 dB more ELOS/N0 is required when the number of
sequences is increased by a factor of 4. Note, the linear noise
term is negligible at this working point. A detailed analysis of
the energy detector for IEEE 802.15.4a can be found in [6].
As observeable from (18), LSNR for the coherent receiver is
independent of the number of the pulses. It depends only on
the transmitted energy. In other words, it does not matter if
this energy is transmitted in one pulse or in a sequence of
pulses. The coherent receiver shows a gain of ≥ 11 dB in the
working point in comparison to the non-coherent receiver.
D. Maximal Operating Distance Estimation
As N0 is constant in the scenario, ELOS/N0 is modeled
by the well-known pathloss model for the maximal operating
distance dmax. Thus
ELOS
N0(dmax)dB =
ELOS
N0(d0)dB − 10n log
(dmax
d0
)(20)
where n is the pathloss exponent and d0 is a reference distance.
The maximal operating distance for the coherent receiver is
obtained from (18) and (20):
dmax =
(ELOS
N0(1)
ELOS
N0(dmax)
)1/n
=
(ELOS
N0(1)
LSNRWP
)1/n
(21)
where the reference distance is assumed to be 1 m. It follows
for the energy detector
dmax =(
ELOSN0
(1)
LSNRWP+√
LSNRWP (LSNRWP+ND/2)
)1/n
(22)
using (19) in (20).
E. Maximal Allowed Pathloss PLmax
A more general look at achieveable range is given by the
maximal allowed pathloss, which is independent of the channel
model, fading margins or implementation losses. It follows
from the pathloss model
ELOS
N0(dmax)dB =
ELOS
N0(d0)dB − PLmax,dB (23)
where ELOS is the energy of the received LOS component
at 1 m, which does not take fading margins or implementation
losses into account (see Table II). Thus, PLmax for the coherent
receiver at 1 m is given by
PLmax,dB =ELOS
N0(1)dB − LSNRWP,dB (24)
using (18) and (23). For the energy detector, using (19) and
and (23), it follows
PLmax,dB =ELOS
N0(1)dB
−10 log(
LSNRWP +√
LSNRWP (LSNRWP + ND/2))
.
(25)
ELOS/N0(1) and ELOS/N0(1) are defined by the transmitted
preamble energy Epr (see Section V) and the link budget (see
Section VI).
V. FCC REGULATIONS
In this section the maximal allowed transmit power is
calculated w.r.t. the FCC regulations [7]. In principle, the same
regulations have been adopted by the EC in Europe for the
band between 6 − 8.5 GHz [8]. In the band between 3.1 and
4.8 GHz the EC requires detect and avoid (DAA) or low duty
cycle (LDC) mitigation additionally, which does not influence
this analysis. This analysis is done in accordance to [9].
The FCC constraints essentially consist of an average and
a peak power limit. In any band of bandwidth Bav = 1 MHz,
the average transmit power is limited to P FCCav = −41.3 dBm
by averaging over an integration window Tav = 1 ms. The
peak power within the bandwidth Bpk = 50 MHz is restricted
to P FCCpk = 0 dBm. Both peak and average transmit power are
defined by the equivalent isotropically radiated power (EIRP).
The 802.15.4a preamble is a sequence of non-uniformly
spaced pulses whose polarities are chosen pseudo-randomly
by the codes. According to [9], its average and peak power is
determined by ERF and PRF respectively and the pulse energy
spectral density Ep,av|V (f0)|2 for the average power limit is
given by
Ep,av|V (f0)|2 =
TavPFCCav
2BavERF ≤ 1/Tav
P FCCav
2BavERFERF ≥ 1/Tav
(26)
where V (f0) is the spectrum of the UWB pulse at the center
frequency f0. ERF is defined as
ERF =
{MRF
Tpr
Tav= M
TavTpr < Tav
MRF Tpr ≥ Tav
(27)
where (Ns +1)/2 is the number of code elements 6=0. ERF
is the compressed MRF due to stretching the preamble over
the averaging time Tav = 1 ms (FCC), or for greater Tav it
is MRF. The mean power limit is limited by the number of
pulses within one 1 ms (N1ms) (see Table I). The peak power
14
102
103
104
105
106
107
108
109
10−23
10−22
10−21
10−20
10−19
10−18
10−17
10−16
Effective Repetion Frequency (ERF) [Hz]
Ep
|V2(f
0)|
[J/H
z]
FCC: Avg
FCC: Peak
PL−ERF: Ns=31,L=16
PL−ERF: Ns=31,L=64
PL−ERF: Ns=127,L=4
Active Power Contstraint
Fig. 4: Pulse Power Spectral Density
limit is defined by the PRF, due to sequenced pulses within
the observation window 1/Bpk = 20 ns at the long preamble
codes. The peak power limit is obtained by
Ep,p|V (f0)|2 =
P FCCpk
9B2pk
PRF ≤ (3/2)Bpk
P FCCpk
4PRF2 PRF ≥ (3/2)Bpk
and ∃f ′ ∈ [−Bpk/2, Bpk/2]
(28)
The maximal FCC compliant pulse energy spectral density
(ESD) with respect to peak and average power are shown
in Fig. 4. To find the active ESD for a specific preamble,
the smaller value between ESDpk at PRF and ESDAv at ERF
is considered. It is observable that only the short preamble
sequences with Ns = 31 with a spreading of L = 16 and
L = 64 are peak power limited, where ERF = 0.256 MHz
(see also Table I). As PRF is constant for these values the
markers are set at ERF of the peak power limit.
Assuming a pulse with rectangular spectrum, the energy per
pulse Ep=2BEp|V (f0)|2, where B is the pulse bandwidth.
Thus, the achieveable preamble energy can be calculated as
shown in Fig. 5. The long preamble codes with Ns = 127are always restricted by the average power limit, thus in-
creasing Npr up to 1024 does not lead to a preamble energy
improvement due to Tpr ≤ 1 ms. Npr = 4096 leads to an
improvement, due to Tpr > 4 ms, which means the preamble
is more than 4 times longer than Tav. At Npr = 16, the long
preamble sequences lead to an advantage of approximately
5 dB in comparison to the short sequences. The short preamble
codes with spreading 64 imply a 4 times longer preamble in
contrast to the others, thus a gain of 6 dB can be achieved.
VI. LINK BUDGET
As mentioned before, ELOS/N0(1) is the input SNR of the
receiver at 1 m, which depends on the link budget. Table II
101
102
103
104
−50
−48
−46
−44
−42
−40
−38
−36
−34
−32
Npr
Pre
am
ble
Energ
y [dB
mW
s]
N
s=31,L=16
Ns=31,L=64
Ns=127,L=4
Tpr
≤ 1ms
Tpr
>1ms
Fig. 5: Achievable preamble energies. The dotted line shows
the boundary for the mean power limit, which limits preambles
with a length ≤ 1 ms.
TABLE II: Link BudgetParameter Values
Pulse Ep (incl. GTX ) −116.38 dBWsPreamble Epr −74.31 dBWs
Free space loss @ 1 meter −45.5 dBReceiver antenna gain GRX 0 dBi
Received LOS component energyELOS −119.81 dBWs
Noise Spectral Density N0 −198, 93 dBW/HzImplementation Loss 4 dB
Fading Margin 3 dB
ELOS/N0(1) 72.12 dB
shows an example link budget calculation for 802.15.4a chan-
nel 3 using Npr = 1024, Ns = 31, L = 16, fc = 4.4928 GHz,
and a bandwidth of 499.2 MHz. In that case the average power
limit of the FCC regulations applies and Ep is calculated from
(26), where also the antenna gain is included. Ep is limited for
1006 sequences due to averaging over 1 ms (see Table I). The
preamble energy Epr = EpM1, the free space loss L at 1 m is
given by 45.5 dB using Friis’ equation [10] and the receiver
antenna gain GRX is defined by 0 dBi. These values yield
the received preamble energy without multipath components,
meaning the energy of the line of sight component ELOS.
Assuming the input structure of the receiver is linear, the
noise spectral density is given by N0 = kT0F [10], where the
Boltzmann constant k = 1.38 10−23 Joule/Kelvin [J/K], the
temperature of the environment T0 = 293 K, and the noise
figure of the receiver input structure F = 5 dB (cf. [13]).
Implementation losses of 4 dB1 and a LOS fading margin of
3 dB are assumed. Thus, ELOS/N0(1) is achieved and can be
used to calculate the maximal operating range according to
(20) and (22).
1This value was provided by DecaWave (www.decawave.com)
15
65 70 75 80 85 90 95 100
102
103
104
ELOS
/N0(d
0) [dB]
Max. O
pera
ting D
ista
nce [m
]
ED: N
s=31, L=16
ED: Ns=31, L=64
ED: Ns=127, L=4
CR: Ns=31, L=16
CR: Ns=31, L=64
CR: Ns=127 L=4
65 70 75 80 85 90 95 100
45
50
55
60
65
70
75
80
85
90
Max. A
llow
ed P
ath
loss [dB
]
Fig. 6: Code sequence analysis on the maximal operating
distance and the maximal allowed pathloss. Fig. 5illustrates the mapping of Npr to the different energy levels.
VII. RESULTS
The maximal operating distance and the maximal acceptable
pathloss are analyzed in this section. The maximal operating
distance is based on the free space link budget, because
only the LOS component is analyzed in this paper. The
maximal acceptable pathloss is shown as a more general
value, which allows the implementer to analyze the effect
of specific channel models, e.g. NLOS scenarios, or specific
system parameters, e.g. lower noise figures.
A. Effect on Codes
Fig. 6 shows the maximal operating distance dmax and
the maximal allowed pathloss PLmax with respect to the
preamble sequences. As expected from Section V the best
performance is achieved by the short preamble with long
spreading (L = 64) due to the highest transmitted energy,
where a maximal operating distance of approximately 430 m
(PLmax≈60 dB) is achieved for the energy detector. The
performance of the short preamble sequences with L = 16 is
up to 220 m (PLmax≈53 dB). As observable, the increasing
of Npr does not necessarily lead to a better performance
because of the non-coherent combining loss. Thus, a per-
formance degradation is seen at around Elos/N0(d0)≈79 dB
for increasing numbers of transmitted pulses due to constant
transmitted energy. This effect also harms the performance
of the long preamble codes significantly and leads to the
lowest achieveable performance. The performance with short
preamble sequences and spreading L = 16 is best at Npr = 64and with the long preamble sequences at Npr = 16. In
contrast to the energy detector, the coherent receiver is only
dependent on ELOS/N0, thus only the allowed transmit energy
per preamble limits the performance. It achieves for the
short preamble code with L = 64 the maximal distance of
≈6000 m (PLmax≈82 dB), where the other two codes achieve
approx. the half maximal operating distance with ≈3000 m
(PLmax≈76 dB).
0 5 10 15 20 25 30 35 40−10
−5
0
5
10
15
20
25
30
ELOS
/N0 [dB]
LS
NR
[dB
]
Ch 3: fc ≈ 4.5, B ≈ 0.5
Ch 4: fc ≈ 4, B ≈ 1.33
Ch 7: fc ≈ 6.5, B ≈ 1.1
Ch 9: fc ≈ 8, B ≈ 0.5
Ch11: fc ≈ 8, B ≈ 1.33
Ch15: fc ≈ 9.5, B ≈ 1.35
CR all channels
Working Point ED
Working Point CR
Fig. 7: Relation between input SNR ELOS/N0 and output SNR
LSNR for the short preamble codes with a spreading L = 64and Npr = 16 with respect to the IEEE 802.15.4a channels.
Note, the channel parameters in the legend are in GHz
The IEEE 802.15.4a standard defines also different channels
with specific bandwidths and carrier frequencies, which also
influences the achieveable operating range. Thus, an analysis
for the specific channels is given in the next section.
B. Effect on Channels
The IEEE 802.15.4a standard defines 16 channels in three
frequency bands, namely the Sub-Giga-Hertz (<1 GHz),
the Low-Band (3.2 − 4.8 GHz) and the High-Band (5.9 −10.3 GHz). The channel bandwidths B range from 499.2 to
1354.97 MHz. As mentioned in Section V, keep in mind that
the EC allows only the usage of the frequency bands 3.1−4.8and 6 − 8.5 GHz for UWB, where for LDC the signals have
to be shorter than 5 ms. Thus, the short preamble symbol with
spreading L = 64 and Npr = 4096 is not allowed for LDC
transmission. As it is well-known, that the carrier frequency fc
shows an influence on the received signal strength according
to Friis’ equation and a larger bandwidth leads to a higher
allowed transmit power (see Section V). Thus, 6 channels
are analyzed in this paper with their effect on the maximal
received signal strength. The short preamble codes with a
spreading of L = 64 are used for this analysis.
Fig. 7 shows the relation between input and output SNR for
the specific channels, where the variations occur due to the
different pulse and low-pass filter bandwidths. It is observable
that the channels with the large bandwidths need up to 1.5 dB
more Ep/N0 in the working point to achieve the same LSNR.
As seen from (18), the coherent receiver is again independent
of the channel bandwidths.
Fig. 8 shows the allowed preamble energies for the specific
channels. The large bandwidths of the preambles delivers a
gain of up to 4 dB, which is sufficient to compensate the
SNR loss of the energy detector (see Fig. 7). This is also seen
from (19), where the equivalent bandwidth WRRC influences
LSNR linearly and the additional energy improves the SNR
16
101
102
103
104
−50
−48
−46
−44
−42
−40
−38
−36
−34
−32
−30
Npr
Pre
am
ble
En
erg
y [
dB
mW
s]
Ch 3: f
c ≈ 4.5, B ≈ 0.5
Ch 4: fc ≈ 4, B ≈ 1.33
Ch 7: fc ≈ 6.5, B ≈ 1.1
Ch 9: fc ≈ 8, B ≈ 0.5
Ch11: fc ≈ 8, B ≈ 1.33
Ch15: fc ≈ 9.5, B ≈ 1.35
Fig. 8: Allowed preamble energy for the specific
IEEE802.15.4a channels.
quadratically in the working point. Thus, a gain of up to 2.5 dB
can be achieved. For the coherent receiver the additional
energy will directly improve the performance.
Fig. 9 shows the maximal operating distance and the
maximal allowed pathloss for the specific channels. The best
performance is achieved by Channel 4, which shows a low
carrier frequency fc ≈ 4 GHz and a large bandwidth of
B ≈ 1.33 GHz. It reaches a dmax ≈ 10620 m (PLmax ≈ 88 dB)
for the coherent receiver and dmax ≈ 620 m for the energy
detector (PLmax ≈ 63 dB). The mandatory Ch 3 of the low
band shows a significantly better performance in comparison
to the mandatory channel in the high band Ch 9 due to much
lower fc. It can be seen that a shift in the carrier frequency
leads to a change of ELOS/N0 but it does not change the
relation of input and output SNR. A shift of the bandwidth
changes the relation between them for the energy detector
(change of ND), which is observable for Ch 9 and Ch 11.
Even the higher bandwidth of the high frequency channels are
not able to compensate the higher free space losses. Again
the characteristic of the coherent receiver only depends on
ELOS/N0.
VIII. CONCLUSIONS
A coherent matched filter and an energy detector are studied
in detail for ranging in IEEE 802.15.4a, in the sense of
allowed transmission energy and maximal operating distance.
The FCC/EC regulations are studied with the 802.15.4a-
defined preamble sequences and frequency channels. The
effects of the pulse repetition frequency and the bandwidth
are analyzed. The maximal operating distance of the energy
detector strongly depends on the system parameters, where the
coherent receiver is independent. The energy detector suffers
from non-coherent combining losses and achieves maximal
operating distances of several hundred meters, while the co-
herent receiver can achieve maximal distances up to several
thousand meters in free space. Furthermore, the maximal
allowed pathloss is shown as performance metric in a more
general sense. It is shown that the preamble sequences with
a length of 127 do not lead to a better performance. The
preamble sequences with a spreading factor of 64 lead to the
best performance due to additional allowed transmit energy.
This leads to a trade off between complexity and performance
in the sense of the used preamble length and further receiver
complexity. The performance of the channels from the low
frequency band show better performance due to the lower
pathloss, which is also not compensateable by the expanded
channel bandwidths in the high band.
ACKNOWLEDGMENT
The authors would like to warmly thank G. Kubin, TU Graz,
and M. Pistauer, CISC Semiconductor GmbH, for their support
in the project. The project was funded by the Federal Ministry
for Transport, Innovation and Technology (BMVIT) and the
Austrian Research Promotion Agency (FFG).
REFERENCES
[1] IEEE Working Group 802.15.4a, Wireless Medium Access Control(MAC) and Physical Layer (PHY) Specifications for Low-Rate Wireless
Personal Area Networks (WPANs), tech. rep. IEEE 2007[2] Y.S. Kwok, F. Chin, and X. Peng; Ranging Mechanism, Preamble Gener-ation, and Performance with IEEE802.15.4a Low-Rate Low-Power UWB
Systems, Ultra-Wideband, ICU2006, IEEE Int. Conf. on, Pages:525-530[3] M. Weisenhorn and W. Hirt; Robust Noncoherent Receiver ExploitingUWB Channel Properties, Ultra Wideband Systems, 2004. Joint withConference on Ultrawideband Systems and Technologies. Joint UWBST& IWUWBS. 2004 International Workshop on Page(s): 156-160
[4] K. Witrisal, G. Leus, G.J.M. Janssen, M. Pausini, F. Troesch, T.Zasowski, and J. Romme Noncoherent ultra-wideband systems, IEEESignal Processing Mag., July 2009
[5] I. Guvenc and Z. Sahinoglu, Threshold-based TOA estimation forimpulse radio UWB systems, Ultra-Wideband, ICU2005, IEEE Inter-national Conference on, Pages: 420-425
[6] T. Gigl, J. Preishuber-Pfluegl, and K. Witrisal, Statistical Analysis of aUWB Energy Detector for Ranging in IEEE 802.15.4a, Ultra-Wideband,ICU2009, IEEE International Conference on, In press
[7] FCC, Revision of part 15 of the commission’s rules regarding ultra-wideband transmission systems, First Report and Order, ET-Docket 98-153, FCC 02-28, adoped/released Feb.23, 2004/Apr. 22, 2002
[8] European Commission; Commission Decision of 21 April 2009: Onallowing the use of the radio spectrum for equipment using UWB
technology, Official Journal of the European Union L 105/9, 2009[9] F. Troesch, Novel Low Duty Cycle Schemes From Ultra Wide Band toUltra Low Power, PhD-Thesis, ETH Zurich
[10] L.W. Couch, Digital and Analog Communication Systems, SeventhEdition, Pearson Prentice Hall, 2007
[11] T. Gigl, J. Preishuber-Pfluegl, D. Arntiz, K. Witrisal ExperimentalCharacterization of Ranging in IEEE 802.15.4a using a Coherent
Reference Receiver, in IEEE Intern. Symp. on Personal, Indoor, andMobile Radio Communications, PIMRC, Tokyo, Japan, September 2009.
[12] I. Guvenc, Z. Sahinoglu, P.V.Orlik, TOA Estimation for IR-UWB SystemsWith Different Transceiver Types, IEEE Transactions on MircrowaveTheory and Techniques, Vol.54, No.4, 2006
[13] G. Fischer, O. Klymenko, D. Martynenko, Time-of-Arrival Measure-ment Extension to a Non-Coherent Impulse Radio UWB Transceiver,Workshop on Positioning, Navigation and Communication 2008, WPNC2008.
17
65 70 75 80 85 90 95 100
102
103
104
ELOS
/N0(d
0) [dB]
Ma
x.
Ope
rating
Dis
tan
ce
[m
]
ED: Ch 3: f
c ≈ 4.5, B ≈ 0.5
CR: Ch 3: fc ≈ 4.5, B ≈ 0.5
ED: Ch 4: fc ≈ 4, B ≈ 1.33
CR: Ch 4: fc ≈ 4, B ≈ 1.33
ED: Ch 7: fc ≈ 6.5, B ≈ 1.1
CR: Ch 7: fc ≈ 6.5, B ≈ 1.1
ED: Ch 9: fc ≈ 8, B ≈ 0.5
CR: Ch 9: fc ≈ 8, B ≈ 0.5
ED: Ch11: fc ≈ 8, B ≈ 1.33
CR: Ch11: fc ≈ 8, B ≈ 1.33
ED: Ch15: fc ≈ 9.5, B ≈ 1.35
CR: Ch15: fc ≈ 9.5, B ≈ 1.35
65 70 75 80 85 90 95 100
45
50
55
60
65
70
75
80
85
90
Ma
x.
Allo
we
d P
ath
loss [d
B]
Fig. 9: Maximal operating distance and maximal allowed pathloss for the specific IEEE 802.15.4a channels. The dashed lines
show the characteristic for a channel with B = 499.2 MHz with the ND of Fig. 6.
18