Multi-band OFDM Physical Layer Proposal Update

1
Project IEEE P802.15 Working Group for Wireless
Personal Area Networks (WPANs) Submission Title
Multi-band OFDM Physical Layer Proposal
Update Date Submitted 10 November,
2003 Source Presenter 1 Roberto Aiello
Company Staccato Communications
Presenter 2 Anand Dabak Company Texas
Instruments
see page 2,3 for the complete list of
company names, authors, and supporters Address
12500 TI Blvd, MS 8649, Dallas, TX
75243 Voice214-480-4389, FAX [phone],
E-Maildabak_at_ti.com Re This submission is
in response to the IEEE P802.15 Alternate PHY
Call for Proposal (doc. 02/372r8) that was issued
on January 17, 2003. Abstract This document
describes the Multi-band OFDM proposal for IEEE
802.15 TG3a. Purpose For discussion by IEEE
802.15 TG3a. Notice This document has been
prepared to assist the IEEE P802.15. It is
offered as a basis for discussion and is not
binding on the contributing individual(s) or
organization(s). The material in this document is
subject to change in form and content after
further study. The contributor(s) reserve(s) the
right to add, amend or withdraw material
contained herein. Release The contributor
acknowledges and accepts that this contribution
becomes the property of IEEE and may be made
publicly available by P802.15.
2
This contribution is a technical update authored
by

• Texas Instrument 03/141 Batra
• Femto Devices 03/101 Cheah
• FOCUS Enhancements 03/103 Boehlke
• General Atomics 03/105 Askar
• Institute for Infocomm Research 03/107 Chin
• Intel 03/109 Brabenac
• Mitsubishi Electric 03/111 Molisch
• Panasonic 03/121 Mo
• Philips 03/125 Kerry
• Samsung Advanced Institute of Technology
  03/135 Kwon
• Samsung Electronics 03/133 Park
• SONY 03/137 Fujita
• Staccato Communications 03/099 Aiello
• ST Microelectronics 03/139 Roberts
• Time Domain / Alereon 03/143 Kelly
• University of Minnesota 03/147 Tewfik
• Wisair 03/151 Shor

For a complete list of authors, please see page
3.
3
Authors from 17 affiliated companies/organization
s

• Femto Devices J. Cheah
• FOCUS Enhancements K. Boehlke
• General Atomics N. Askar, S. Lin, D. Furuno, D.
  Peters, G. Rogerson, M. Walker
• Institute for Infocomm Research F. Chin,
  Madhukumar, X. Peng, Sivanand
• Intel J. Foerster, V. Somayazulu, S. Roy, E.
  Green, K. Tinsley, C. Brabenac, D. Leeper, M. Ho
• Mitsubishi Electric A. F. Molisch, Y.-P.
  Nakache, P. Orlik, J. Zhang
• Panasonic S. Mo
• Philips C. Razzell, D. Birru, B. Redman-White,
  S. Kerry
• Samsung Advanced Institute of Technology D. H.
  Kwon, Y. S. Kim
• Samsung Electronics M. Park
• SONY E. Fujita, K. Watanabe, K. Tanaka, M.
  Suzuki, S. Saito, J. Iwasaki, B. Huang
• Staccato Communications R. Aiello, T. Larsson,
  D. Meacham, L. Mucke, N. Kumar, J. Ellis
• ST Microelectronics D. Hélal, P. Rouzet, R.
  Cattenoz, C. Cattaneo, L. Rouault, N. Rinaldi,,
  L. Blazevic, C. Devaucelle, L. Smaïni, S.
  Chaillou
• Texas Instruments A. Batra, J. Balakrishnan, A.
  Dabak, R. Gharpurey, J. Lin, P. Fontaine, J.-M.
  Ho, S. Lee, M. Frechette, S. March, H. Yamaguchi
• Time Domain / Alereon J. Kelly, M. Pendergrass
• University of Minnesota A. H. Tewfik, E.
  Saberinia
• Wisair G. Shor, Y. Knobel, D. Yaish, S.
  Goldenberg, A. Krause, E. Wineberger, R. Zack, B.
  Blumer, Z. Rubin, D. Meshulam, A. Freund

4
In addition, the following 19 affiliated
companies support this proposal

• Adamya Computing Technologies S.Shetty
• Broadcom J. Karaoguz
• Fujitsu Microelectronics America, Inc A.
  Agrawal
• Furaxa E. Goldberg
• Hewlett Packard M. Fidler
• Infineon Y. Rashi
• JAALAA A. Anandakumar
• Microsoft A. Hassan
• NEC Electronics T. Saito
• Nokia P. A.
  Ranta
• Realtek Semiconductor Corp T. Chou
• RFDomus A. Mantovani
• SiWorks R. Bertschmann
• SVC Wireless A. Yang
• TDK P. Carson
• TRDA M. Tanahashi
• tZero O. Unsal
• UWB Wireless R. Caiming Qui

5
Why did 10 Companies Propose Multi-Band
Solutions in March 2003 ?

• Some of the reasons include
• Spectrum Flexibility / Agility
• Regulatory regimes may lack large contiguous
  spectrum allocations
• Spectrum agility may ease coexistence with
  existing services
• Energy collected per RAKE finger scales with
  longer pulse widths used
• Fewer RAKE fingers
• Reduced bandwidth after down-conversion mixer
  reduces power consumption and linearity
  requirements of receiver
• Fully digital solution for the signal processing
  is more feasible than a single band solution for
  the same occupied bandwidth
• Transmitter pulse shaping made easier
• Longer pulses easier to synthesize less
  distorted by IC package antenna properties
• Have the ability to utilize an FDMA mode for
  severe near-far scenarios

6
Most of the Multi-Band Proposals in March 03
used Pulses, What Happened ?

• Energy collection under severe multipath (CM3,
  CM4) required improvement
• We needed a computationally efficient method of
  multipath combining
• Parallel receivers? Infinite RAKE? OFDM?
• OFDM in each sub-band was selected as a successor
  to the pulsed multi-band approaches

7
Why are 37 Companies Now Supporting the
Multi-band OFDM Approach ?

• Multi-band OFDM kept the unique Multi-Band
  benefits and solved the energy collection problem
  very elegantly
• Feasibility studies of FFT and Viterbi cores
  showed encouraging numbers for gate-count and
  power consumption
• Multi-band OFDM suitable for CMOS implementation
  (all components)
• Antenna and pre-select filter are easier to
  design (can possibly use off-the-shelf
  components)
• Low cost low power CMOS integrated solution
  early market adoption
• Scalability
• Digital section complexity/power scales with
  improvements in technology nodes (Moores Law).
• Analog section complexity/power scales slowly
  with technology node
• Much more can be said in detail about the
  Multi-band OFDM PHY performance, but first we
  should review our proposal

8
Overview of OFDM

• OFDM was invented more than 40 years ago
• Adopted by numerous standards effort
• Asymmetric Digital Subscriber Line (ADSL)
  services.
• IEEE 802.11a/g IEEE 802.16a
• Digital Audio Broadcast (DAB) Home Plug
• Digital Terrestrial Television Broadcast DVD in
  Europe, ISDB in Japan
• OFDM is also being considered for 4G, IEEE
  802.11n and 802.20
• OFDM is spectrally efficient.
• IFFT/FFT operation ensures that sub-carriers do
  not interfere with each other
• OFDM has an inherent robustness against
  narrowband interference.
• Narrowband interference will affect at most a
  couple of tones.
• Information from the affected tones can be erased
  and recovered via the forward error correction
  (FEC) codes
• OFDM has excellent robustness in multi-path
  environments.
• Cyclic prefix preserves orthogonality between
  sub-carriers.
• Cyclic prefix allows the receiver to capture
  multi-path energy more efficiently

9
Overview of Multi-Band OFDM

• Basic idea divide spectrum into several 528 MHz
  bands
• Information is transmitted using OFDM modulation
  on each band
• OFDM carriers are efficiently generated using an
  128-point IFFT/FFT
• Internal precision is reduced by limiting the
  constellation size to QPSK
• Information bits are interleaved across all bands
  to exploit frequency diversity and provide
  robustness against multi-path and interference
• 60.6 ns prefix provides robustness against
  multi-path even in the worst channel environments
• 9.5 ns guard interval provides sufficient time
  for switching between bands
• Solution is very scalable and flexible
• Data rates, power scaling, frequency scaling,
  complexity scaling

See latest version of 03/268 for more details
about the Multi-Band OFDM system
10
Band Plan

• Group the 528 MHz bands into 4 distinct groups.
• Group A Intended for 1st generation devices (3.1
  4.9 GHz).
• Group B Reserved for future use (4.9 6.0 GHz).
• Group C Intended for devices with improved SOP
  performance (6.0 8.1 GHz).
• Group D Reserved for future use (8.1 10.6
  GHz).
• Use of Group A is mandatory, while use of Group
  AC is optional.

11
FCC Compliance of Multi-band OFDM

• Presenter Anand Dabak

12
Interference and the FCC (1)

• In 03/153r9 (July 2003), XSI stated
• The issue today is NOT whether or not there is
  more or less interference, the issue is, what
  are the rules
• XSI/Motorola filed a petition with the FCC for
  declaratory ruling immediately after the San
  Francisco meeting
• Q Should a multi-band OFDM waveform be
  transmitted at a lower power than other UWB
  systems?
• The FCC response (full response in back-up
  slides)
• FCCs concern is not with interpretations of the
  rules, but rather with interference.
• We urge that IEEE perform technical analyses to
  ensure that any UWB standard it develops will not
  cause levels of interference beyond that already
  anticipated by the rules.

13
Interference and the FCC (2)

• Interference
• We have identified several systems that need to
  be considered by any UWB transmitter.
• Most of these systems are out of band (OOB) and
  require adoption of an appropriate spectral mask
  to ensure appropriate level of protection.
• Several simulation studies completed looking at
  impact of MB-OFDM waveform on various FEC schemes
  often employed in wideband FSS systems.
• Several (measurement based) experiments are being
  conducted to determine impact to real systems.
• MB-OFDM does not cause any more interference than
  already anticipated by current FCC rules.
• FCC compliance Contrary to XSIs claims, the
  multi-band OFDM system is FCC compliant and
  should not have to reduce its transmit power.

14
Study of Potential Victim Receivers

• Most US government and commercial systems are
  out-of-band (OOB)
• For OOB systems, victim receiver performance is
  expected to be the same for MBOK MB-OFDM type
  UWB interference.
• MB-OFDM has a slight advantage due to better OOB
  rejection capabilities.
• No impact expected on CW Pulsed Altimeter
  systems due to high tolerable UWB TX power limits
  (NTIA Report 14 dBm/MHz).
• Analog FSS systems are quickly being replaced by
  digital FSS systems.
• In 1995, there were 2 million analog FSS systems.
  In 2002, only 500K.

15
FSS Simulation results

• 35 MSPS, rate 7/8 coding, no interleaving, Iuwb/N
  -6 dB XSI filing to FCC for typical operating
  scenarios, Sept. 2003

Very little difference between UWB radios under
realistic scenarios Note SINRC/(NIsIuwb),
Issatellite intra-system interference
16
Interference into FSS band

• Interference and coexistence studies depend on a
  number of factors
• Application dependent variations expected
  minimum separation distance between UWB emitter
  and victim receiver under realistic usage models,
  probability of this minimum separation distance
  seen in reality, pervasiveness of victim receiver
• Implementation variations antenna gain response,
  available link margin, FEC and other signal
  processing techniques adopted to mitigate noise
  and interference
• Other interference sources intra-system
  interference sources, noise floor of device
• Allowed interference margins minimum criteria
  for interference level and impact on probability
  of outage, built-in margin for external
  interference sources (all systems must expect
  some level of interference)
• Potential interference caused by multi-band OFDM
  is lower than that generated by impulse radios,
  which are allowed under FCC rules.

17
Interference to Out-of-band Systems

• Many of the government systems (FAA, DOD) and
  commercial systems (GPS, PCS) are out of band
  (OOB) for the proposed UWB systems.
• The OOB rejection capability for UWB systems is
  important when analyzing interference to these
  systems.
• Since the multi-band OFDM system employs a
  narrower bandwidth than MBOK, it can achieve
  better OOB rejection
• OFDM has an inherent steep roll-off at the band
  edges due to modulating narrow tones (4 MHz)
  relative to the occupied bandwidth (528 MHz).
• To achieve similar roll-off, an MBOK system would
  require sharp (higher-order) filters, which can
  be expensive in terms of die area and insertion
  loss.
• Hence, interference into out of band FAA, GPS,
  and PCS systems can be much less from MB-OFDM
  systems than from MBOK systems.

18
FCC Chairman Michael PowellKey Steps toward
Spectrum Reform
Future In-band Interference Mitigation Techniques

• International regulatory agencies are supportive
  of frequency agile solutions to help protect
  different services in different locations
• Recent ITU meeting shows uncertainties still
  exist around international regulations
• Multi-band OFDM is an efficient method for
  enabling frequency agility

• There is a substantial amount of white space
  out there that is not being used by anybody.
• A software-defined radio may allow licensees to
  dynamically rent certain spectrum bands when
  they are not in use by other licensees.

Broadband Migration III New Directions in
Wireless Policy University of Colorado at
Boulder, Oct 30, 2002
19
HomePlug Power Line Spectral Mask -- A Precedent
for Low-Cost Sculpting via OFDM Technology
Source HomePlug Alliance, HomePlug ARRL Joint
Test Report, January 24, 2001
30dB Notches Protect Amateur Radio
Frequency, MHz
OFDM enables simple interference reduction
techniques
20
Conclusions and Summary

• Conclusions about FSS
• Both multi-band OFDM and WGN waveforms are less
  harmful than impulse radios, which are allowed
  under the FCC.
• Contrary to XSIs claims, multi-band OFDM is FCC
  compliant and should not have to reduce its
  transmit power.
• Summary
• The multi-band OFDM proponents are committed to
  ensuring that no harmful interference is caused
  to potential victim receivers.
• Both simulations and real experimental testing
  will continue in order to determine if anything
  in the current proposal needs to be changed to
  help mitigate potential interference
• This should be the case for ANY draft proposal
  adopted by the IEEE.
• The combination of multi-banding and OFDM
  provides a unique capability to tightly control
  OOB emissions as well as enables spectrum
  flexibility to protect future systems and
  differences in international allocations.

21
Comparison Between theMulti-band OFDM and MBOK
Proposals

• Presenter Anand Dabak

22
Apples to Apples Comparison

• Similar frequency bands MB-OFDM 3.1-4.9 GHz,
  MBOK 3.15-5.15 GHz
• Compared multi-band OFDM versus MBOK with respect
  to
• Performance and range in multi-path channel
  environments.
• Robustness to interference from a single tone
  jammer.
• Analog and RF implementation considerations.
• ADC precision requirements.
• Digital complexity.
• Comparison based upon widely available
  information for MBOK system.
• Digital architectures for MBOK/DS-CDMA have been
  selected for the comparison.
• Expected to provide better performance over
  analog implementation slide 8 of 03/0334r2

23
MBOK simulation environment

• Receiver architectures used for MBOK simulations

Analogous to the architecture proposed by
ParthusCeva in 03/0334r3, ParthusCeva employs a
single 1 bit ADC at 5.472 GHz

24
MBOK simulation parameters (Architecture 1)
150 fingers at I, Q complex are equivalent
to 300 fingers in 03/334r3 1 03/0334r3

• Degradations from packet detection, time/carrier
  tracking, front end filtering, not included in
  simulations

25
MB-OFDM simulation parameters

• Time tracking, carrier phase tracking, front end
  filtering, ADC quantization losses included in
  the simulations

26
Simulation parameters comparison

• Degradations
• MBOK simulations results are optimistic.
• Calibrated M-BOK performance (see backup slide
  62).

Degradations Multi-band OFDM MBOK
Included in the simulations Packet detection Channel estimation Time/carrier tracking ADC quantization DAC clipping Front-end filter Channel estimation ADC quantization Timing offset Channel estimate quantization
NOT included in the simulations Packet detection Time/carrier tracking Front-end filter
27
Multi-path Performance Comparison
28
Performance for 112/110 Mbps

• Multi-band OFDM outperforms MBOK with 150 finger
  RAKE by about 1 dB in multi-path channel
  environment (CM3).

29
Performance for 224/200 Mbps

• Multi-band OFDM outperforms MBOK with 150 finger
  RAKE by about 5 dB in multi-path channel
  environment (CM3).

30
Error Floor for MBOK

• The MBOK system hits an error floor in multi-path
  channel environments for data rates of 200 Mbps
  (CM3) and 448 Mbps (CM2).

Error floor for MBOK (does not reach 10-5)
31
Performance 16 finger RAKE

• MBOK performance improves marginally with 16
  finger RAKE no ADC quantization. But,
• 112/114 Mbps MBOK is 1.5 dB worse than 110 Mbps
  MB-OFDM.
• 224/448 Mbps MBOK is about 4 to 6 dB worse than
  200/480 Mbps MB-OFDM.
• 200 Mbps MBOK hits an error floor.

32
Range Comparisons
33
Range in AWGN

• Transmitter backoff, propagation loss
  calculations given in backup.
• Multi-band OFDM has better range than MBOK in an
  AWGN environment.
• 20 m (110 Mbps MB-OFDM) versus 16.8 m (112 Mbps
  MBOK)
• 14 m (200 Mbps MB-OFDM) versus 12.6 m (224 Mbps
  MBOK)
• 7.8 m (480 Mbps MB-OFDM) versus 6.8 m (448 Mbps
  MBOK)

480 Mbps
448 Mbps
224 Mbps
200 Mbps
200 Mbps
110 Mbps
114 Mbps
112 Mbps
34
Range in Multi-path

• Multi-band OFDM has significantly better range
  than the MBOK system.
• 11.6 m (110 Mbps MB-OFDM) versus 9.4 m (112 Mbps
  MBOK)
• 6.8 m (200 Mbps MB-OFDM) versus 3.9 m (224 Mbps
  MBOK)
• 2.6 m (480 Mbps MB-OFDM) versus 1.2 m (448 Mbps
  MBOK)

480 Mbps
448 Mbps
200 Mbps
224 Mbps
200 Mbps
110 Mbps
112 Mbps
114 Mbps
35
Single Tone Interferer
36
Single Tone Interferer Simulation Results

• Assumption receiver operates 6 dB above
  sensitivity (15.3a criterion)
• For MBOK need SIR 4 dB for architecture 1 and
  SIR 1 dB for architecture 2.

37
Single Tone Interferer Comparison

• For a fair comparison between two systems, we
  assume there is no analog filter notches for
  either system.
• MB-OFDM results in backup 03-268r1P802-15_TG3a-Mu
  lti-band-CFP-Document.doc
• Multi-band OFDM system out performs MBOK
  architecture 2 by 7 dB, and MBOK architecture 1
  by 12 dB.
• May be possible to use DSP techniques for MBOK to
  improve its performance, however the complexity
  of MBOK receiver will then increase.

38
ADC Requirements for an MBOK System
39
ADC requirements for MBOK architecture 2

• Multi-path simulations CM3 for 224 Mbps, CM2 for
  448 Mbps
• 3 bits required for 224 Mbps, 4 bits required for
  448 Mbps

MBOK architecture 2, 448 Mbps
MBOK architecture 2, 224 Mbps
40
ADC Requirement Comparison

• Multi-band OFDM requires a lower sampling rate
  ADC than the MBOK system.
• For rates less than 224 Mbps
• MB-OFDM requires an ADC running at 528 MHz with 4
  bits precision.
• MBOK requires an ADC running at 2736 MHz with 3
  bits precision.
• MBOK may employ chip rate sampling, but
  performance will be worse.
• For rates greater than 224 Mbps
• MB-OFDM requires an ADC running at 528 MHz with 5
  bits precision.
• MBOK requires an ADC running at 2736 MHz with 4
  bits precision.
• MBOK may employ chip rate sampling, but
  performance will be worse.
• The ADC requirements for the multi-band OFDM
  system is simpler than that required for the MBOK
  system architecture 2.

41
Analog/RF Implementation Comparison
42
Is RF sampling feasible for MBOK ?

• Proposed RF sampling architecture for MBOK in
  03/334r3.
• Two crucial issues
• Out of band interference rejection IEEE 802.11a.
• RF gain feasibility.

43
IEEE 802.11a rejection

• For an IEEE 802.11a device to operate within 1
  meter of UWB, the IEEE 802.11a rejection required
  is a total of 60 dB.
• With an off-chip filter, one can achieve 30 dB
  of rejection. Still need another 30 dB of
  rejection.
• Only other possibility Put another off-chip
  filter after LNA.
• This implies
• Higher bill of material, special components
  Higher cost.
• Increased off-chip external components Cannot
  have an integrated solution.

44
RF gain feasibility for MBOK

• The sensitivity for 110 Mbps MBOK is -80 dBm
• Even for a 1 bit ADC, an RF sampling architecture
  for MBOK will require gain amplifiers with a
  total gain of 60 dB at an RF center frequency of
  4.1 GHz and bandwidth of 1.6 GHz.
• Such wideband, high gain amplifiers at RF
  frequencies are very difficult to implement in
  practice.
• Oscillations Stability problems
• Yield Time to market
• Hence it may be very risky in practice to
  implement the RF sampling architecture proposed
  in 03/334r3

45
Mixer based architecture for MBOK System

• A mixer-based architecture for front end RF is
  feasible for the MBOK system.
• Need a 750 MHz wide low pass filter with sharp
  cutoff MB-OFDM needs 250 MHz filter
• Need a broad band GA/VGA (750 MHz) for MBOK
  MB-OFDM needs 250 MHz wide VGA
• Need 1 bit ADC at 2736 MHz for architecture 1 and
  3-4 bit ADC at 2736 MHz for architecture 2
  MB-OFDM needs 528 MHz 4-5 bits ADC.
• MB-OFDM needs to generate multiple frequencies
  while MBOK needs to generate a single frequency.

46
Comparison of RF/Analog Complexity
Architecture RF Sampling Mixer to Baseband Architecture
MBOK Architecture 1 Extremely Difficult Feasible
MBOK Architecture 2 Extremely Difficult Very Difficult

• Qualitative comparison of RF/Analog complexity
  between MB-OFDM and MBOK.
• (1) Architecture 1 1 bit ADC Quantization.
• (2) Architecture 2 3-4 bit ADC Quantization.

47
Digital Complexity Comparison
48
Digital RX Block Diagram for MBOK

• Assumption 130 nm technology at 85.5 MHz (per
  03/334r3-03/447r0).
• We estimated the complexity for the major blocks
  (shaded blocks) of the MBOK system.

The implementation complexity of shaded blocks
was calculated. The implementation complexity for
other blocks was not taken into account
49
Chip Matched Filter (CMF) Complexity

• Assumption 150 fingers with a 1-bit ADC.
• CMF needs about 225,000 gates (85.5 MHz clock)
• In 03/334r3, the estimated gate is 75,000 (85.5
  MHz).
• Difference occurs because 03/334r3 did not take
  into account that both I Q outputs (224 Mbps
  mode is QPSK) are needed from the CMF output.
• In the latest document 03-0447 it is estimated
  as 49,400 (171 MHz clock) for real CMF and 90,200
  (171 MHz clock) for complex CMF
• For a fair comparison should use the same clock
  frequency.

50
Complexity for MBOK Architecture 1

• Assumption 130 nm, 85.5 MHz clock.
• Backup slides contains calculations for MBOK
  decoder and synch block.
• To make a fair comparison with MB-OFDM system, we
  need to adjust to clock of MBOK to 132 MHz
• MBOK system requires 400K gates (_at_132 MHz).
• Multi-band OFDM system needs 295K gates (_at_132
  MHz) 03-0343.
• MBOK system requires 35 more baseband complexity
  when compared to the multi-band OFDM system.

51
Summary of Comparison Results

• Did a fair comparison of multi-band OFDM versus
  MBOK in terms of performance, complexity,
  implementation feasibility, interference, based
  upon available data on the MBOK proposal.
• The proposed direct RF sampling architecture
  (MBOK arch 1) in 03-334r3 may not be feasible
  in practice.
• MB-OFDM has a clear advantage over MBOK in terms
  of
• Significantly better range in multi-path (20
  120 increased range)
• Significantly better robustness against
  interference (7 12dB better performance).
• Simpler hardware requirements (lower rate ADCs,
  lower bandwidth VGAs)
• Lower digital complexity (25 less number of
  gates)

52
Conclusions

• Pursuance of the best technical solution has led
  to the current MB-OFDM proposal
• Current proposal is superior to all other
  proposals presented to the Task Group
• Authors from 17 affiliated companies/organizations
  and supporters from 19 others
• All these companies, which represent the vast
  majority of the industry, have spent significant
  resources to evolve to the best technical
  solution
• UWB and FCC
• Both MB-OFDM and WGN waveforms have similar
  interference properties and are less harmful than
  impulse radios, which are allowed under the FCC
  rules.
• Multi-band OFDM does not generate any more
  interference than anticipated by FCC.
• MB-OFDM is superior to MBOK based on an apples to
  apples comparison
• Multi-band OFDM is lower complexity, lower power
  consuming, more feasible, better performing and
  more robust to interference when compared to MBOK
  solution

53
Backup
54
FCCs response
Summary of Discussions with FCC Staff Concerning
IEEE 802.15 Deliberation On Standards for
Ultrawideband devices

• Over the past few weeks several parties have met
  with the staff of the Federal Communications
  Commission Office of Engineering and Technology
  to discuss how the Commissions rules for
  Ultrawideband devices might be applied for
  certain signal formats that are being considered
  by IEEE 802.15.
• OET believes it is premature to make any
  determination as to the appropriate measurement
  methods for particular signals because this
  matter is under active discussion in IEEE. In
  this regard, we have no immediate plans to
  respond to the XSI/Motorola request for a
  declaratory ruling.
• We urge that IEEE perform technical analyses to
  ensure that any UWB standard it develops will not
  cause levels of interference beyond that already
  anticipated by the rules.
• This information will be needed to support any
  necessary FCC rules interpretations or other
  appropriate action for the chosen standard.
• The FCC has had a long history of working
  cooperatively with the IEEE 802 committee in
  addressing any regulatory issues that may arise
  relative to standards. We recommend that IEEE
  proceed with its standards development process
  and that the committee address any questions to
  us at a later time when it has formed a specific
  proposal.

E-mail sent by Julius Knapp, Deputy Chief, OET,
FCC to XSI/Motorola and MB-OFDM proponents on
September 11th, 2003
55
UWB Bandwidth and Peak Radiated Emissions within
a 50 MHz BW
Radio Sample 1 Test Distance 1m Detector
PEAK RBW/VBW 3 MHz/3 MHz Meas.
Time 1 ms Emissions lt Limit UWB
BW gt 500 MHz Note Data normalized to
3m test environment and 50 MHz RBW for limit
comparison.
56
Radiated Emissions UWB
Radio Sample 2 Test Distance 1m Detector
RMS RBW/VBW 1 MHz/3 MHz Meas. Time
1 ms Emissions lt Limit Note Data
normalized to 3m test environment for limit
comparison.
57
Emissions in GPS Bands
Radio Sample 1 Test Distance
Conducted Detector RMS RBW/VBW
1 kHz/3 kHz Meas. Time 1 ms Emissions
lt Limit Note Limit line is most stringent
at 3m distance. No emissions above noise floor
in radiated or worst case conducted measurement
mode.
58
MBOK simulated data rates

• For architecture 1 simulation conditions as close
  to the receiver proposed in 03/334r3

R0.44 is concatenated ½ convolutional code with
RS(55,63) R0.50 convolutional code R0.87 is
RS(55,63)
59
Calibration of MBOK decoder

• 8-BOK BER performance matches with 03-334r3
• 4-BOK gives no performance gain over AWGN

60
Performance for 114 Mbps

• Multi-band OFDM outperforms MBOK with 150 finger
  RAKE by about 2 dB in multi-path channel
  environment (CM3).

61
Results for 112 Mbps (No interleaver between MBOK
and Viterbi)

• MB-OFDM outperforms MBOK 150 finger rake by 2 dB

62
Performance for 224 Mbps

• The performance of the MBOK system degrades in
  the absence of an interleaver between the MBOK
  demodulator and the Viterbi decoder
• MBOK reaches an error floor in multi-path channel
  environment with a 150 finger RAKE.

63
MBOK simulation parameters (Architecture 2)
Changes from simulations for architecture 1

• Degradations from packet detection, time/carrier
  tracking, front end filtering, not included in
  simulations

64
5 finger rake results with no ADC quantization

• 5 finger rake for 114, 112 Mbps, hence the MBOK
  atleast works in practice for these data rates,
  despite significantly bad performance ( 6 dB
  worse) with respect to MB-OFDM. For other data
  rates, 5 finger rake hits an error floor.

65
Equalizers for MBOK

• Performance for MBOK could be improved using
  equalization techniques.
• Linear equalizers
• Equalizer length would probably have to be in the
  same order as the number of rake fingers
• 150 taps for architecture 1
• 16 taps for architecture 2
• Training large number of taps requires longer
  preamble and adds to the complexity.
• Decision feedback equalizers (DFE)
• The complexity of DFEs for 64-BOK can be
  significant
• Error propagation could be significant at
  operating Eb/N0 (1 dB at the input of the
  decoder).

66
Transmitter backoff

• Need to take into account the transmitter backoff
  for range calculations
• For fair comparison, assume the same RF front end
  noise figure for MB-OFDM and MBOK

67
Range comparison

• Take into account transmitter backoff,
  propagation loss
• AWGN
• Multi-path

68
MB-OFDM results 03268r1P802-15_TG3a-Multi-band-C
FP-Document.doc

• MB-OFDM can erase tones in digital
• Required SIR gt - 8 dB

69
MBOK demodulator complexity

• Since the phase is unknown before the MBOK
  demodulation, need to do MBOK correlation for
  both CMF output I, Q with MBOK I, Q codes.

70
Synchronization/Preamble detection complexity

• The MB-OFDM system employs a preamble of 312.5 ns
  and length 128 samples long.
• A fair comparison would require similar
  synchronization performance of the correlator for
  MBOK and for MB-OFDM system.
• MBOK employs about 3X the bandwidth of the
  MB-OFDM single band, implying multi-path the MBOK
  pulses would be about 3X lower in energy compared
  to the MB-OFDM, as seen by the correlator .
• Hence in order to achieve an acquisition
  performance similar to MB-OFDM, we expect that
  the MBOK has to do a correlator of length 1
  microsecond gt 1368 chips long
• However, for reduced complexity of MBOK, assume a
  length 553 correlator only as in
  03123r6P802-15_TG3a-ParthusCeva-CFP-Presentation.
  ppt

Design challenges for very high data rate
UWB systems Somayazulu, V.S. Foerster, J.R.
Roy, S. Signals, Systems and Computers, 2002.
Conference Record of the Thirty-Sixth Asilomar
Conference on , Volume 1 , Nov. 3-6, 2002
Page(s) 717 -721
71
Synchronization/Preamble detection complexity (2)

• Assume that a 1 bit input shift register needs 4
  gates per bit
• Assume the 553 length correlator has 1 bit input.
  The adder tree for the correlator grows in the
  number of bits. Assume 6 gates/adder (optimistic
  estimate)
• Total gates of 177 K for synchronization

72
Digital complexity of Architecture 2

• Needs 16 4-bit complex multiplies Assuming 800
  gates/4-bit complex multiply gives 205K
• Assume a 1 bit synchronization similar to
  architecture 1
• Other complexity is the same as architecture 1
• 130 nm, 85.5 MHz clock