Conference PaperPDF Available

Polarity reception for IR-UWB in wireless fading channel

Authors:
  • Institute for High Perfomance microelectronics and TU-Cottbus

Abstract and Figures

Detection performance of Impulse Radio Ultra wideband IR-UWB wireless communication was evaluated in noisy and dispersive channel environment. A polarity reception method was introduced which sets reference phase and does phase detection in relation to it. Simulation results confirm that the adopted detection scheme can achieve a packet error rate of 0.01 for SNR as low as-5 dB in fading channel even at high data rate 27.24 Mb/s. To the best knowledge of the authors, it is the first ever research work introducing results of polarity detection in IR-UWB systems. The wireless IR-UWB being considered here was implemented and shown to be fully functional. Thus the results reflect real practical system.
Content may be subject to copyright.
Polarity reception for IR-UWB in wireless fading
channel
Sonom Olonbayar, Dan Kreiser, Rolf Kraemer
IHP
Frankfurt Oder, Germany
sonom@ihp-microelectronics.com
Gunter Fischer, Denys Martynenko, Oleksiy Klymenko
IHP
Frankfurt Oder, Germany
kreiser@ihp-microlectronics.com
Abstract Detection performance of Impulse Radio Ultra
wideband IR-UWB wireless communication was evaluated in
noisy and dispersive channel environment. A polarity reception
method was introduced which sets reference phase and does
phase detection in relation to it. Simulation results confirm that
the adopted detection scheme can achieve a packet error rate of
0.01 for SNR as low as -5 dB in fading channel even at high data
rate 27.24 Mb/s. To the best knowledge of the authors, it is the
first ever research work introducing results of polarity detection
in IR-UWB systems. The wireless IR-UWB being considered here
was implemented and shown to be fully functional. Thus the
results reflect real practical system.
Index TermsIEEE802.15.4a, Synchronization, detection,
SNR, distortion, carrier frequency deviation, I and Q signal,
pulse position modulation, comparator, NLOS channel
I. INTRODUCTION
IR-UWB was specified in the standard IEEE802.15.4a as a
physical layer technology for short range low to medium data
rate wireless communications [2]. IR-UWB operates in the
frequency band of (3.1 -10.6) GHz occupying a large
bandwidth being more than 500 MHz. The transmit power is
limited to -41.3 dBm/MHz restricting the range and reducing
the interference to others. Due to its very short pulses, it is well
suited in accurate indoor ranging and positioning. It can also be
applied as wireless communication for transferring data
reliably, in particular for automation purposes. For such
applications, reliability to be measured by a packer error rate
(PER) has to be as low as possible. This paper explores
detection techniques in wireless environment and their
capabilities to exploit the processing gain and other advantages
offered by IR-UWB in achieving the required performance.
Most detection techniques realized so far focus on energy
reception schemes which establishes presence of energy at the
expected positions. Such methods are simpler to implement
however, they suffer from limited performance [1,4]. With
such non-coherent reception, polarity detection is impossible
limiting the performance. The standard recommends a
combination of pulse position modulation (PPM) and binary
phase shift keying (BPSK). Data bits are carried with IR-UWB
by a number of pulses placed at certain position known as burst
and their polarity. The burst length differs for varying data
rates being 16 for 850 kb/s, 2 for 6.81 Mb/s and 1 for 27.24
Mb/s. This paper discusses a detection method based on I and
Q signal that can receive both pulses and their phase. If error
occurs, convolutional and Reed-Solomon decoding do
correction as suggested in the standard. IR-UWB is usually
researched as a ranging [6]. This work focuses on treating it as
communication technique.
The paper is structured into five sections. In the first section
a short review is given on the modulation schemes specified for
the standard. The next section discusses IR-UWB detection
methods along with the one applied to this work. Wireless
channel models common to NLOS environment are explored in
the third section. The fourth section considers imperfections to
be caused during implementation and operation that can
degrade the bit error rate performance. The last section gives
simulation results obtained.
II. MODULATION IN IR-UWB
As IR-UWB is based on transmitting very short pulses to
be measured in few nanoseconds even shorter creating large
bandwidth, the standard recommends to use PPM. Here a
number of pulses generated according to certain sequence, is
placed at either first or second half of a symbol which has
differing lengths depending on the data rate. For example the
symbol duration is 1024 ns at 850 kb/s which accommodates
16 pulses each occupying 2 ns. For higher data rates such as
6.81 Mb/s and 27.24 Mb/s, the symbol durations are shortened
to 128 ns and 64 ns respectively. The number of pulses is
reduced to 2 and 1 accordingly. The packet structure for the
above mentioned three data rates is shown in Figure 1. Beside
the PPM that sends one bit every symbol period, a second bit is
transmitted with BPSK by reversing the polarity of sequence.
Hence, the receiver has to detect the presence of sequence (bit
1-if at the first half, 0 if at the second half of symbol) and
establish its reversal of polarity (bit 1 if reversed, 0 if not
reversed).
Figure 1. IR-UWB packet (frame) with a packet length of
17 bytes for various data rates, fc=7.9872 GHz
0 0.5 1 1.5 2 2.5 3 3.5 4
x 106
-10
-8
-6
-4
-2
0
2
850 Kb/s
6.81 Mb/s
27.24 Mb/s
It can be seen that for lower data rates, there is a large number
of pulses allowing PPM bit to be detected with high reliability.
One burst extracted from all considered data rates are
illustrated in Figure 2.
850 kb/s 6.81 Mb/s 27.24 Mb/s
Figure 2. Data PPM burst corresponding to various data rates
A preamble sequence of the packets is shown in Figure 3 and it
contains negative and positive pulses arranged according to
certain sequence. Thus synchronization scheme has to be
capable to recover polarity of every single pulse for detecting
the preamble (see Figure 3). Thus polarity recovery plays a role
in synchronization performance.
1024ns t
Figure 3. Preamble sequence used for the preamble
III. WIRELESS CHANNEL OF IR-UWB
The most influencing factor to the performance of wireless
communication is wireless channel through which the signal
propagates. It can distort signal in numerous ways. Wireless
channel is perceived to manifest itself producing multiple
copies of the signal through reflection, diffraction and etc.
These copies arrive at the receiver with various delays and it
can reach as long as 40 ns for IR-UWB and can be large in
numbers. The main task of receiver in this case is to sort out all
these paths and to get focused on the strongest. IR-UWB
channel is rather unique according to [5] which models late
arriving paths to possess even higher level than the first.
Moreover, the channel impulse response occurs in clusters
extending largely in time. Once channel impulse response is
known, its affect to the signal is expressed by convolutional
operation. It is clear that longer the response, higher the delay
of the signal to be received. If this delay exceeds 32 ns, it can
cause inter-symbol interference to IEEE802.15.4a. For clarity,
one 850 kb/s burst (16 pulses with 32 ns duration) is illustrated
in Figure 4 which is subjected to both noise and multipath
fading.
a) 850 kb/s noisy burst b) channel response c) faded burst
Figure 4. IR-UWB burst influenced by wireless fading channel
NLOS office environment and SNR=0 dB
It can be observed that bursts can get disturbed drastically in
particular the polarity may get altered randomly making
detection hard.
IV. DETECTION OF IR-UWB
Detection of bits contained in position of bursts is made by
checking expected positions for energy as visualized in Figure
5. After the preamble and start of frame delimiter (SFD) phase,
data detection commences by observing if the burst is
positioned in the first or in the second half of symbols. The
position is calculated at the receiver for every symbol and
energy detection is followed at these positions only. If burst is
received in the first half, the decision is made for bit ‘1’
otherwise for ‘0’. In case there is a shift for bursts from their
original location due to clock offset, channel and any other
reasons, the receiver can miss them making wrong decision.
Figure 5. PPM bit detection scheme for 850 kb/s
To receive polarity bits conveyed by reversing the phase of
burst sequence, a polarity of the individual pulses needs to be
detected. For instance, each polarity of 16 pulses ought to be
recovered. One way of detecting polarity appears in Figure 6
where detection is made for 1 and -1 when the signal exceeds
threshold Vthr and Vthr respectively. This method was
employed by the receiver being presented which integrates the
signal over 2 ns and compares the result with the threshold set.
The comparator here serves as an analogue to digital converter
having a sampling rate of 499.2 MHz and a resolution of 1.5
bit. It is obvious even in ideal case it requires the threshold to
be set accurately.
1.915 1.9155 1.916 1.9165 1.917 1.9175 1.918
x 105
-1
-0.5
0
0.5
1
7.9165 7.917 7.9175 7.918
x 105
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1.3493 1.3494 1.3495 1.3496 1.3497 1.3498
x 105
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
2.7 2.72 2.74 2.76 2.78 2.8 2.82 2.84
x 105
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
3.8305 3.831 3.8315 3.832 3.8325 3.833 3.8335 3.834 3.8345 3.835 3.8355 x 105
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
050 100 150 200 250 300
0
0.05
0.1
0.15
0.2
0.25
3.831 3.8315 3.832 3.8325 3.833 3.8335 3.834 3.8345 3.835 3.8355 3.836
x 105
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0
t (ns)
1
-1
992
2016
0
1
-1
32ns
0ns
0ns
32ns
0ns
20ns
0
32 ns
4 ns
2 ns
Figure 6. Polarity recovery for the 850 kb/s burst at the
baseband in ideal situation
It is not straightforward to recover phase of every single pulse
as the polarity can easily get altered due to reasons like fading,
clock offset or implementation inaccuracies. For coping with
this, I and Q signals were generated at the receiver when
down-converting the signal from 7.9872 GHz to 499.2 MHz
baseband as shown in Figure 7. The receiver contains low
noise amplifier (LNA) and I and Q path generation stages that
involve down-conversion from 7.9872 GHz to 499.2 MHz,
variable gain amplifier chain and comparators. After it was
made digital through the comparator synchronization, bit
detection and decoding operations take place. Finally the
received packet is stored in data memory.
Figure 7. I and Q signal generation at the receiver
Since these I and Q signals are 90° apart in phase to each other
(see Figure 7), they experience different signal level.
Furthermore, initial phase of the carrier can affect the
performance too as there is no phase aligning. By combining I
and Q, the robustness of detection can be improved
dramatically even in highly fading channel. As shown in the
Figure 8, I and Q paths undergo varying envelope distortion
where phase of individual pulses gets rotated by 180° due to
carrier frequency deviation causing phase ambiguity. This
phase rotation occurs periodically being more often for higher
deviation. Therefore, the receiver should be able to
differentiate whether the phase rotation is because of data
itself (-1 in data bursts) or other reasons mentioned before.
The polarity detection adopted works as follows. With SFD
overhead of the frame (packet) ends indicating the beginning
of the data field. As the SFD has the structure 0 1 0 -1 1 0 0-1
whereby 1 refers to preamble sequence and -1 preamble with
phase reversal, the last reversed preamble sequence is set as a
phase reference. Here, it was assumed that the polarity
remains unchanged following SFD till data reception
terminates. This assumption does not hold if carrier frequency
deviation becomes large so that the phase gets rotated earlier
the end of packets. However, one of these I and Q signals
gives an indication on correct phase if analyzed properly.
Following the SFD, polarity detection proceeds by comparing
recently received phase with the reference set. If for example
reversed preamble sequence is detected at I path, it is assumed
that the phase on that path is reversed across the whole frame.
The same principle is applied to Q path considering
effectively both paths to the detection process. Figure 8 shows
the frame corresponding to maximum allowable carrier
frequency (fc) deviation which is 160 kHz.
Figure 8. I and Q signal generated for 850 kb/s with carrier
frequency deviation of 160 KHz at 7.9872 GHz.
This implies that fc=7.9872 GHz ± 160 KHz either at the
transmitter or receiver. Since the two signals are 90° apart, at a
point of time if one sees weak a signal level the other shows
detectable amplitude. Thus both are taken into decision for
position as well as polarity bit.
Such phase detection method based on the reference set around
SFD can offer a good performance under frequent phase
reversal as will be shown later. If still errors are to be made, it
will be corrected by the Viterbi and Reed Solomon decoders
that follow. The ADC, implemented as comparators, creates
two bit digital output every 2 ns for both I and Q path being 01
for positive, 10 for negative pulses and 00 for null. The digital
baseband thus receives 4 bits (see Figure 9). There are four
comparators: two for each I and Q path one for detecting
positive and the other for negative pulses. Since this reception
method is implemented both in FPGA and ASIC as a part of
wireless physical layer, it is worth explaining it in connection
with its realization. Each digital output (Ip In Qp Qn) is
parallelized (1:16) into 16 bits producing four streams each 16
bits long within one cycle of 31.2 MHz (499.2 MHz/16). The
reason of doing so is to make the implementation simpler and
make the baseband to operate at low clock rate. All operations
taking place afterwards run at 31.2 MHz. Synchronization
algorithm then runs on those four streams searching the
preamble individually on every path. Synchronization
performance plays an important role to the robustness of
detection. If synchronization is not achieved, there is no
reception of data regardless of any sophisticated algorithms to
1.915 1.9155 1.916 1.9165 1.917 1.9175 1.918 1.9185
x 105
-1
-0.5
0
0.5
1Vthr
-Vthr
1 1 -1 1 1 -1 -1 -1 1 -1 -1 1 -1 -1 -1 1
Polarity detection
0 1 2 3 4 5 6 7 8 9
x 105
-2
-1
0
1
2
3
4
I
Q
16µs
32µs
t
0µs
32 ns
0 ns
t
Amplitude
be applied afterwards in detecting data. Thus, this work
analyzes data detection in conjunction with the
synchronization.
Figure 9. Comparator , Ser/Par and correlators at the receiver
Synchronization or preamble and SFD detection is carried out
in the following way. The received data is correlated with the
expected 31 length preamble sequence performing a cross
correlation. The correlation continues with a shift of one step
(2 ns) establishing every time if there is match thus offering a
synchronization accuracy of 2 ns. This accuracy may seem
unsatisfactory for detecting pulses which have the duration of 2
ns. However, with a suitable threshold supported by an
appropriate pulse shaping, it is possible to receive every single
pulse with their polarity. The standard specifies to use 31
length ternary sequence for the preamble. The correlator
defines how much the receive signal matches with the
preamble sequence. In case of perfect match it produces a peak
which is 31 indicating a complete agreement.
As far as the hardware doing the correlation is concerned, there
are 16 correlators working parallel for each signal being in total
32. In other words, 16 bits applied to the correlators from the
Ser/Par every 32 ns, enters 16 correlators at the same time.
Ser/Par outputs corresponding to comparators which detect I
signal with positive and negative threshold (Vthr, -Vthr) are
combined through a logical OR (see Figure 9). If the threshold
voltage is set in a way to capture negative and positive pulses,
after combining, preamble as well data burst should be
recognizable with the correlators that follow. The same holds
for the Q path. Correlation results on I and Q path are then
further used for the synchronization and data bit detection.
Although all 16 correlators for each I and Q operate
simultaneously, with the index of individual correlators a
synchronization accuracy of 2 ns is possible. The one among
these 16 correlators, which spots the preamble as well as SFD,
is taken as a timing base for further detection. From the
moment where last SFD preamble is recognized, the receiver
lands at the position of the first burst as it has the knowledge
where to look. Since the burst could be at one of the two
positions depending on the data bit, the receiver determines
where it detects more pulses: if at the first deciding for bit ‘0’
and if at the second for ‘1’. It is clear that 850 kb/s can achieve
better detection for position bits as there are 16 pulses forming
one burst. The detection is robust even the position gets shifted
slightly. Since only 2 or 1 pulse is transmitted for data rates
6.81 Mb/s and 27.24 Mb/s, the detection performance can get
degraded if their position happens to be shifted.
Detection process continues this way processing every symbol
and observing presence of energy and what the polarity of burst
is. As regard to position bits, the detection is usually reliable as
long as synchronization is achieved especially if the burst is
long. Half rate (R=1/2) convolutional encoding with k=3 and
generator polynomial g0[010]2, g1[101]2 and Reed Solomon
coding RS6(63,55) need to correct bit errors made in detecting
position as well as polarity of burst. One goal of this work has
been to establish which correction suits more in which
circumstance.
As regard to detecting polarity, following the decision made for
positive and negative pulses, the resulting burst is correlated
with the expected one producing peak when the detected
phases match. Once the polarity bits are decided this way
showing which path delivers -1 and 1 in reference to the phase
in the SFD, it is polished finally by examining if any
systematic error is made.
V. SIMULATION OF IR-UWB
Extensive and thorough simulation was carried out to
evaluate the performance of IR-UWB wireless
communications at low to medium data rates. The simulation
made here deviates dramatically from papers based on
theoretical considerations where practical aspects influencing
the performance are quite often neglected. The detection
algorithm discussed was implemented both in FPGA and as
ASIC. Therefore, much effort was made to incorporate
imperfections that can arise during various stages of
implementation. Furthermore, signal was generated in frames
according to the IEEE802.15.4a in the baseband and then up-
converted to 7.9872 GHz in Matlab. A frame signal produced
by the simulation had been proved to be identical with the
practical signal generated by the ASIC implemented. At the
receiver firstly down conversion was made bringing the signal
down to 499.2 MHz followed by a low pass filter (5th order
Butterworth). The down-converted signal is then digitized as
stated earlier through a comparator with a certain threshold.
The digital baseband in this case receives 4 bits (Ip comparator
output working with positive threshold for I signal, In
comparator output working with negative threshold for I signal
and the same holds for Q signal) every 2 ns. The receiver
performs detection according to the method explained in the
previous section.
White noise was added to the up-converted signal
according to the signal to noise ratio needed. The noisy signal
was then introduced fading channel convolving the channel
impulse response with the noisy signal generated.
VI. SIMULATION RESULTS OF IR-UWB
The performance of the IR-UWB wireless communications
compatible to IEEE802.15.4a was evaluated in terms of packet
error rate. The detection scheme was implemented too. Thus
the simulation results reflect the performance of the real
system. The simulation was conducted in the following
manner. IR-UWB frame generated at the carrier frequency
7.9872 GHz was introduced noise and multipath effect and the
resultant signal was then applied to the receiver. First the
synchronization performance was tested seeing how often the
receiver can synchronize. The result is presented in Figure 10.
Here for each SNR, one thousand frames (packets) each with
the length of 17 bytes were received. Each frame undergoes
varying channel condition according to the known Saleh-
Valenzuela model for office NLOS environment. The adopted
synchronization method was shown to be very robust being
nearly 100% correct even under very low SNR which is -5 dB.
This implies that the receiver was synchronized to all one
thousand packets. Preamble symbol recognition threshold was
set to 20 not 31 allowing up to 10 errors. In order to receive as
many detectable paths as possible, threshold of the comparator
was made to vary in the range 0.02 to 0.1 with the step 0.02.
The threshold was varied till synchronization is achieved and
packets at the output of Reed Solomon decoder becomes all
correct. With a single threshold, the performance would be not
satisfactory because multipath signals demonstrate varying
amplitude. Thus the reception continues with a different
threshold for the same packet till the received data gets correct.
If no synchronization happens across the whole range of the
threshold, it is considered not synchronized for the SNR.
Figure 10. IR-UWB Synchronization performance
Beside the synchronization, data detection performance was
evaluated for the data rates mentioned. Upon detection,
position and polarity bits pass through Viterbi and Reed-
Solomon decoders where errors that could have been made, are
corrected. It is of interest to analyze how much error is
introduced and how effective the decoders are in correcting it.
The obtained packet reception performance is visualized in
Figure 11 for various data rates. Two types of receiver were
considered: the first is the one without any correction
mechanisms and the second is with Viterbi and Reed Solomon
decoding involved. The simulation was made in NLOS
channels in office environment and the carrier offset was set to
160 KHz. As shown with dashed lines, lower data rates 850
kb/s and 6.8 Mb/s achieve 100% correct reception for SNR of -
5 dB although no any correction is applied indicating
robustness of the detection scheme adopted. For SNR as low as
-10 dB the performance starts to get degraded falling down to -
78 % and 47% for data rates 850 kb/s and 6.81 Mb/s
respectively. Further lowering of SNR to -15 dB leads to 10%
reception. As for 27.24 Mb/s, the performance without
correction is under 40% even for higher SNR 10 dB suggesting
its sensitivity to fading and carrier offset.
The performance of receivers employing correction methods
are illustrated in solid lines in Figure 11. With the SNR of -5
dB and -10 dB, the decoders start to correct notably. Since the
850 kb/s sends 16 pulses for 1 bit data, it is received mostly
correctly offering less than 1 dB coding gain in SNR compared
to the dashed line. This gain increases substantially for data
rates 6.81 Mb/s and 27.24 Mb/s reaching approximately 5 dB
and 12 dB respectively. Since for the higher bit rates only 2
and 1 pulse is sent, these can get shifted from their position due
to channel influence and other imperfections producing errors.
It is particularly obvious for the 27.24 Mb/s where one pulse is
sent.
Figure 11. Performance at various data rates over NLOS
channel for 100 packets each with the length 17 bytes
In evaluating packet detection, packets were considered not
received even there is only one erroneous bit out of whole
packet with 204 bits.
Assessing robustness of polarity bit reception is particularly
important as it presents main challenge in implementing IR-
UWB systems. Therefore, the standard is made in a way so that
polarity bits can be ignored at the receiver ending up only with
RS decoder without the Viterbi. 850 kb/s and 6.81 Mb/s in this
case display a slight improvement compared to the receiver
performing both RS and Viterbi decoding as appear in Figure
12. However, the combination becomes effective for the
highest rate 27.24 Mb/s despite large number of polarity bit
errors made. This result suggests that only at 850 kb/s Viterbi
decoder can be excluded when SNR is large. Otherwise use of
both decoders performs always better.
10 5 0 -5 -10 -15 -20
0
20
40
60
80
100
SNR (dB)
Synchronisation percent (%)
-15 -10 -5 0 5 10
0
20
40
60
80
100
SNR (dB)
Packet detection (%)
850 Kb/s no correction
6.81 Mb/s no correction
27.24 Mb/s no correction
850 Kb/s corrected
6.81 Mb/s corrected
27.24 Mb/s corrected
1 dB 5 dB
12 dB
Figure 12. Performance comparison of fully decoding only RS
decoded
If only polarity bit detection is to be shown separately, then
half of the packets would be corrupted even for SNR 10 dB at
850 kb/s and this figure would go down to 90 % and 100 % for
the remaining data rates. This gives a hint to the complexity of
detecting polarity bits. The reception algorithm presented in
this work concentrates on to keep varying threshold till
receiver gets synchronized and a packet is received correctly.
The packet is counted as corrupted even there single bit error is
present. Varying threshold helps to get focused on the path that
can be resolved and possess detectable SNR. The lower the
step to change the threshold better is the performance. It is
extremely hard to implement though and most practical
systems implement one threshold limiting the achievable
performance.
VII. CONCLUSION
Performance of IR-UWB communication system which
operates at data rates 850 kb/s, 6.81 Mb/s and 27.84 Mb/s and
is compatible to the IEEE802.15.4a was evaluated. A position
bit detection scheme based on energy reception was explored
in great detail. Beside it, polarity bit detection scheme was
incorporated into reception which decides for polarity bit
referring to a reference phase set in the SFD region of the
preamble. The receiver was shown to achieve very good
synchronization performance being nearly 100% correct for
SNR of -5 dB in highly faded channel. This performance was
obtained by varying threshold in the receiver within certain
interval till the reception becomes correct. The PER was found
to be 0.01 at all data rates when both Viterbi and Reed
Solomon decoders are applied. The detection algorithm was
implemented and tested to be fully functional.
REFERENCES
[1] S. Olonbayar, D. Kreiser, R. Kraemer.; „Performance and
implementation of multi-rate IR-UWB baseband transceiver for
IEEE802.15.4a“, ICUWB2013, Sydney, Sep. 2013
[2] Standard specification IEEE802.15.4a, 2007
[3] S. Olonbayar, G. Fischer, R. Kraemer “Synchronisation
performance of wireless sensor networks“, ICUWB2008, pp.
59-62, no.2 Hannover, Germany, Sep 2008
[4] S. Olonbayar, D. Kreiser, D. Martynenko, G. Fischer, O.
Klymenko, R. Kraemer.; „Board implementation and its
performance for IR-UWB IEEE.802.15.4a from multiple ASIC
chips“, Invited paper, EW2012, Poznan, Poland, April 2012
[5] Andreas F. Molisch, Kannan Balakrishnan and etc.:
IEEE802.15.4a channel model final report, 2003
[6] Tingcong Ye, Michael Walsh etc.: Experimental Impulse Radio
IEEE802.15.4a UWB based wireless sensor localization
technology: characterization, reliability and ranging, ISSC 2011
-15 -10 -5 0 5 10
0
20
40
60
80
100
SNR (dB)
Packet reception (%)
850 Kb/s only RS decoded
27.24 Mb/s only RS decoded
850 Kb/s fully decoded
6.81 Mb/s fully decoded
27.24 Mb/s fully decoded
6.81 Mb/s only RS decoded
Amplitude
... At the receiver, the data frame is first down-converted to the baseband with the Matlab function " amdemod " and a Butterworth low-pass filter is applied to remove highfrequency components. If the carrier frequency at the receiver (nominally 7.9872 GHz) deviates by 160 kHz due to unsynchronized oscillators, the waveform will be distorted as shown in Figure 8(d) [6]. Not only the envelope is distorted but also the polarity of pulses gets reversed periodically. ...
... By observing both paths in the preamble phase, the polarity reversal can be recognized and taken into account during payload detection phase. Such detection is possible if the preamble sequence is known at the receiver [6]. Hence polarity reception is possible by combining the two paths. ...
... Otherwise despite robust detection of position bits, errors that are made in the polarity bit reception, can degrade the performance significantly and introduce errors to actually correctly detected position bits. A discussion on when the decoding schemes should be turned on and how much performance they offer especially in detecting the polarity bits is given in [6]. ...
Article
An in-depth treatment of impulse an radio ultra-wideband (IR-UWB) wireless system is provided reviewing theoretical background, proceeding with detailed implementation procedure, and finally giving simulation and test results. This is the first research and prototyping work to be published in the field of IR-UWB that operates in the 6–8 GHz band. The aim of this work is to implement an IR-UWB wireless system for industrial automation that is robust and reliable. To achieve this, an analogue bandwidth of 250 MHz and digital baseband processing at the clock frequency 499.2 MHz were realized in a 250 nm
Conference Paper
Full-text available
Design, simulation, implementation and performance of IR-UWB baseband conforming to IEEE802.15.4a are discussed. The baseband can support various data rates such as 850 Kb/s, 6.81 Mb/s and 27.24 Mb/s. The design and parameter selection were considered carefully taking into account all possible imperfections that IR-UWB high frequency signal can experience. Energy detection receiver employing a comparator clocked at 499.2 MHz was adopted for the digitisation. Using I and Q path both positive and negative pulses were detected with a high reliability leading to a very good synchronisation performance. Simulation results confirm that the synchronisation is very robust being always correct for office NLOS environment and a large clock deviation between transmitter and receiver. The algorithm presented in this paper was implemented with discrete components, FPGA and signal generators. Experimental results show a good agreement with the simulation for all the data rates and the implemented baseband offers around six meter communication range tested along with a high frequency frontend from discrete components.
Conference Paper
Full-text available
One energy efficient way of ASIC implementation for the standard IEEE802.15.4a was considered in greater detail. The main focus was on developing a solution for short range communication based on impulse radio ultra wideband. The ASIC implementation is performed with the 250 nm SiGe technology from IHP, Germany. The current system consists of three chips: radio frequency frontend, digital baseband and an integrating analogue digital converter. All the chips were fabricated and designed at IHP. A non-coherent energy detection receiver was realised which collects energy over 16 ns in analogue fashion. The performance of the complete system was evaluated bringing the chips on a board. Our multiple ASIC approach was a necessary step towards a complete integration of the system into a single chip. This paper discusses the function of individual chips and their interaction on a board. The performance was assessed by frame error rates, approximate coverage and waveforms at certain points. Initial laboratory tests and measurements suggest that the implemented system can support wireless communications in a distance of up to five meters and consume reasonable power.
Conference Paper
Ultra Wide Band (UWB) transmission has recently been the object of considerable attention in the field of next generation location aware wireless sensor networks. This is due to its fine time resolution, energy efficient and robustness to interference in harsh environments. This paper presents a thorough applied examination of prototype IEEE 802.15.4a impulse UWB transceiver technology to quantify the effect of line of sight (LOS) and non line of sight (NLOS) ranging in real indoor and outdoor environments. Results included draw on an extensive array of experiments that fully characterize the 802.15.4a UWB transceiver technology, its reliability and ranging capabilities for the first time. A new two way (TW) ranging protocol is proposed. The goal of this work is to validate the technology as a dependable wireless communications mechanism for the subset of sensor network localization applications where reliability and precision positions are key concerns.
Conference Paper
Ultra Wideband (UWB) has been gaining growing research and industry interests in many different areas of wireless communications. Among them using UWB as a radio interface for wireless sensor network is of high interest due to its precise localisation and low power consuming capabilities. In this paper we carry out quantitative analysis into the synchronisation performance of UWB when it operates under usual wireless channel environments. Simulation results confirm that coherent receiver needs around 10 dB less signal to noise ratio to achieve synchronisation compared to the noncoherent detection. Higher spreading factors lead to noticeable improvement for both coherent and noncoherent detection.