Information Technology Reference
In-Depth Information
E
N
added to each generated template pulse. The
target TOA is determined from the receiver clock
phase when the output energy of the multiplied
incoming and template signals is maximal. This
value is set to correspond to the TOA of the in-
coming signal. In more detail, the analog sliding-
correlator is composed of a multiplier and buffer
stages. The TOA of the incoming signal is deter-
mined by sweeping the phase of reference clock
with a predetermined one-degree-per-step value,
and the TOA is set to correspond to the maximum
detected cross-correlation energy. Time-step is
p
P z
min
( ) =
Q
(
ρ
(0)
−
ρ
( ))
z
(17)
pv
pv
0
Since the ZZLB transforms the estimation
problem into a binary detection problem, this
simplifies the bound estimation in multi-path en-
vironments. The derivation of
P
min
(z)
requires the
a-priori
knowledge of the multi-path phenomena
(Dardari, Chong et al. 2006).
As the human locomotion tracking system
under investigation should work in indoor and
outdoor environments, we chose the seventh de-
rivative Gaussian pulse to satisfy the FCC masks
for both environments. In order to obtain results
for the TOA bounds using realistic BAN channels,
the ZZLB was determined using a semi-analytic
simulation approach in the IEEE 802.15.6a UWB
CM3 channel model (Yazdandoost 2008). The
resulting ZZLB is depicted in Figures 2 for the
distance estimate of the TOA bound assuming
optimal, sinusoidal (suboptimal), and complex
suboptimal templates (also known as quadrature
analog correlator (QAC)). As seen from the results,
an accuracy of 0.11 cm is achievable at an
E
p
/
N
0
= 18 dB, which defines the target
E
p
/
N
0
. The
achievement of this target SNR was justified in
our link budget design. Also, different link budget
calculations for low-power and low-cost detectors
for BAN applications showed that the achievable
E
p
/
N
0
is greater than 18 dB with appropriate link
margins (Simic, Jordan et al. 2007; Nict 2009).
1
calculated as ∆
t
=
o
in s/1
o
, where
CLK
RX
.360
CLK
RX
is the reference clock frequency (Dederer,
Schleicher et al. 2007).
Even though the multiplier output pulses are
spaced by the pulse-repetition frequency (PRF),
this spacing is transformed to clock sweeping step
when estimating the TOA (Dederer, Schleicher et
al. 2007; Dederer, Schleicher et al. 2008). Obvi-
ously, in order to be able to determine the maximum
energy, which corresponds to the TOA, an ADC
placed after the buffer stage is required. It needs
to have a sampling frequency equivalent to PRF.
For our proposed gait analysis system, initially we
assume 400 possible positions of the correlation
pulse peaks, which correspond to the maximum
expected 4 ns
5
delay divided by the 10 ps step.
Thus, for this stage, an ADC with 9 bit resolution
and a sampling frequency equivalent to PRF =
50 MHz is sufficient. According to (Das, Bhasin
et al. 2009), an ADC with 9.7 effective number
of bits (ENOF) and 80 Msample/s sampling fre-
quency requires a power consumption = 10 mW.
Thus, for a 50 Msample/s PRF, the ADC power
consumption is approximately 6.25 mW. In our
calculations, we further perform TOA estimation
ten-times, and take the average of these estimates.
Essentially, averaging provides values that are not
limited to the 10 ps resolution assumption. For
instance, when we average 10 ps and 20 ps it gives
15 ps, which is a value that would not have been
attained by solely using the 10 ps limit.
Employed Correlator Receiver
Architecture for Ranging
In our design, we assume the analog sliding-
correlator proposed in (Dederer, Schleicher et al.
2007), which capable of achieving our target 1
mm ranging accuracy with the appropriate choice
of the design parameters. In this receiver, incom-
ing pulses are multiplied by a series of template
pulses generated at the pulse repetition rate (PRR)
with a predetermined phase offset increment
Search WWH ::
Custom Search