TDC proposal to 802.15.4a

1
Project IEEE P802.15 Working Group for Wireless
Personal Area Networks (WPANs) Submission Title
Time-Domain-CFP-Response Date Submitted 4
January, 2005 Source Vern Brethour, Adrian
Jennings Company Time Domain Corp. Address
7057 Old Madison Pike Suite 250 Huntsville,
Alabama 35806 Voice Vern [phone]
Adrian [phone], E-Mail
vern.brethour_at_timedomain.com adrian.jennings_at_tim
edomain.com Re 802.15.4a CFP Abstract 802.1
5.4a CFP response from Time Domain. An impulse
radio nominally occupying 3 5 GHz with 4 ns
chip times using 40 chips/symbol and 300 ns quiet
time between symbols. Purpose Response to
WPAN-802.15.4a CFP 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 individuals or
organization. The material in this document is
subject to change in form and content after
further study. The contributors reserve the right
to add, amend or withdraw material contained
herein. Release The contributors acknowledge
and accept that this contribution becomes the
property of IEEE and may be made publicly
available by P802.15.
2
ProposalSingle Band UWB Alternate Physical
Layer for TG 802.15.4a
3
Proposal Contents

• General Overview
• Proposal Principles
• Regulatory Flexibility
• Performance
• Evaluation Matrix (in backup slides)

4
General Overview

• Impulse radio
• Single band nominally from 3 to 5 Ghz.
• 4 ns chip times
• 40 chips per symbol
• 300 ns quiet time between symbols
• Max symbol integration 64 (data)
• Max symbol integration 256 (acquisition)

5
Proposal Principles

• The most important part of a proposal is the
  signal as it appears on the air. For most signal
  definitions, there are many ways to build a
  radio, and as many corresponding performance
  results. However, as a standard, we can define a
  signal which will forever limit the systems
  ultimate performance. (For example, by not using
  all reasonably available bandwidth.)

6
The need for robust links

• There is already a 15.4a radio at 2.54GHz. We
  must be substantially better than that radio.
• This proposal provides the opportunity for
  maximum performance by occupying as much
  bandwidth as reasonable.

7
Regulatory Flexibility

• There are fundamentally two approaches to UWB
  regulatory flexibility
• 1) using multiple bands.
• 2) longer chip times.
• Using long chip times allows for filters if
  needed and does little harm if not needed.

8
Regulatory Flexibility

• This proposal occupies all of the spectrum
  between the ISM bands and the UNII bands
• This proposal allows ample (4 ns) chip time to
  accommodate spectral shaping if necessary.
• This proposal also allows a future (optional)
  band between 6 and 10 GHz.

9
Support for positioning

• There are already radios which do the low data
  rate communications job without positioning.
• Excellent positioning performance will be the key
  differentiator for 15.4a.
• Use of as much bandwidth as reasonable gives the
  best positioning performance.

10
What about the simple radio approach?

• Vocabulary is important here. Words like
  simplicity imply virtue. Words like crude
  and unsophisticated might be used by others to
  describe the same radio.
• The critical issue is that there will be other
  users of the spectrum and the 4a standard must
  use spectrum and air time efficiently and
  effectively.
• A proposed standard which we think implies a
  simple and virtuous radio might be viewed by
  others as spectrally wasteful and unworthy of
  letter ballot approval.

11
What does simple radio mean?

• A simple radio to our customers doing system
  integration, is a radio with the lowest chip
  count, the least number of passives and the most
  forgiving antenna driver.
• The integration customer does not (and should
  not) care how hard we have to work to implement
  the design inside of our chip.

12
Performance the optimistic story.

• A marketing style link budget looks very
  optimistic even for 250 Kbyte links at 100
  meters, the link budget shows over 6dB of margin.

Parameter Value Value
Information Data Rate 1 Kbps 250 Kbps
Average TX Power -12.5 dBm -12.5 dBm
Total Path Loss 83.5 dB (_at_ 100 meters) 83.5 dB (_at_ 100 meters)
Average RX Power -96 dBm -96 dBm
Noise Power Per Bit -144 dBm -120 dBm
CMOS RX Noise Figure 8 dB 8 dB
Total Noise Power -136 dBm -112 dBm
Required Eb/N0 2.25 dB 2.25 dB
Implementation Loss 6 dB 6 dB
Link Margin 31.75dB 7.75 dB
RX Sensitivity Level -128 dBm -104. dBm
Max. Range (AWGN) 3870 m 245 m
13
Link Performance A realistic story.

• Performance predicted by the link budget is
  optimistic primarily due to the use of 2 for
  the path loss exponent.
• Links inside buildings with interior walls, will
  suffer path loss exponents more like 3.

14
Link Performance

• A more realistic idea of performance is available
  by scaling the results of the simulations done
  for 802.15.3a to longer ranges and lower data
  rates.
• The 802.15.3a DS proposal uses signaling similar
  to this proposal, so I will scale from
  simulations reported in 802.15.04.0483r5
  (McLaughlin, November 2004).

15
Scaling the DS results

• The DS radio is simulating a 11.8 meter link in
  CM4 at 110 Mbit/sec with a 90 packet success
  rate.
• Going from a link distance of 11.8 meters to 100
  meters would seem to require less than 20 dB of
  additional processing gain. BUT thats with a
  path loss exponent of 2.
• A path loss exponent of 3 requires 28 dB of
  processing gain.

16
How much integration is needed for 28 dB of
processing gain?

• Each time we double the integration, we get
  another 3dB of processing gain.
• For 28 dB, we need to do 10 doublings, or an
  integration rate of 1024.
• Integration rate 1024 will take the 110 Mbit/sec
  rate down to 107 Kbit/sec.

17
Noticeable difference
Data rate Predicted range
Link Budget with Path Loss exponent 2 250 Kbit/sec 245 meters
Scaled Simulation with Path Loss exponent 3 107 Kbit/sec 118 meters
18
Even the Scaled Simulation is a very optimistic
result

• The DS radio that this prediction rests on is a
  very fancy radio
• 16 Rake taps
• 31 tap decision feedback equalizer
• Constraint length 6 convolutional code with
  Viturbi decoder
• RF front end with 6.6 dB noise figure

19
Link budgets do not address acquisition.

• Acquisition will usually be the performance
  limiter at long range.
• The Acquisition decision needs an additional 6 dB
  of processing gain over data demodulation.

20
We must acquire without benefit of a trained
equalizer.

• Equalizers are fine, but only after they have
  been trained.
• If the spacing between pulsing events is too
  short, the resulting inter symbol interference
  makes trouble for acquisition.
• This proposal uses a relatively long (300 ns)
  distance between pulse events to handle large
  channel delay spreads without an equalizer.

21
Applications need robust links.

• The applications can stand low data rates, so
  this proposal uses data symbol integration of 64
  and acquisition symbol integration 256.
• The long acquisition integration puts a burden on
  crystal tolerance (2 ppm) that not all vendors
  will want to deal with, so shorter integration
  modes will also be supported.

22
Clear Channel Assessment

• This is a hard problem for all UWB approaches.
• We should not ignore it.
• Detection of energy at the chipping rate (as in
  the 15.3a DS proposal) is doable, but not
  reliable.
• We may need relief from the MAC.

23
Simultaneously Operating Piconets

• The long symbol (40 chips) enables good
  orthagonality between piocnets.
• Different piconets use slightly different
  chipping rates like the 15.3a DS proposal.
• Bits are modulated onto symbols using only BPSK
  so all of the symbol orthogonality is used for
  piconet isolation.

24
Power control is the key to compatibility with
15.3a

• This proposal is set up for 100 meter links.
• Many applications will have shorter links.
• For shorter links, we turn down the Tx power.
• Power control gives superior compatibility with
  all services.

25
Proposal Summary

• Impulse radio
• Single band nominally from 3 to 5 Ghz.
• 4 ns chip times
• 40 chips per symbol
• 300 ns quiet time between symbols
• Max symbol integration 64 (data)
• Max symbol integration 256 (acquisition)

26
Backup Slides
27
Evaluation Matrix
28
Self Evaluation General Solution Criteria
CRITERIA Evaluation
Unit Manufacturing Cost (UMC) (no need for an equalizer)
Signal Robustness Interference (due to power control)
Signal Robustness Susceptibility (due to using max bandwidth)
Coexistence (due to power control)
Technical Feasibility Manufacturability (due to low peak Tx amplitudes)
Time To Market
Regulatory Impact (due to long chip times)
Scalability
Location Awareness (due to using max bandwidth)
29
Self Evaluation PHY Protocol Criteria
CRITERIA Evaluation
Size and Form Factor
Payload Bit Rate
Packet Overhead
PHY-SAP Throughput
Simultaneously Operation Piconets (due to the long 40 chip symbol)
Signal Acquisition (due to long symbol integration)
System Performance
Link Budget (due to using max bandwidth)
Sensitivity
Power Management Modes
Power Consumption
Antenna Practicality