IEEE 802.15 <PHY Proposal>

1
Project IEEE P802.15 Working Group for Wireless
Personal Area Networks (WPANs) Submission Title
IEEE 802.15.4b High Rate Alt-PHY proposals -
Further Performance Comparison Date Submitted
10 Nov, 2004 Source Francois Chin
Company Institute for Infocomm Research,
Singapore Address 21 Heng Mui Keng Terrace,
Singapore 119613 Voice 65-6874-5687 FAX
65-6774-4990 E-Mail chinfrancois_at_i2r.a-star
.edu.sg Re Response to the call for proposal
of IEEE 802.15.4b, Doc Number 15-04-0239-00-004b
Abstract This presentation compares all
proposals for the IEEE802.15.4b PHY
standard. Purpose Proposal to IEEE 802.15.4b
Task Group 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
Background

• Main contribution of current doc is to provide
  further simulation results based on 1000 channel
  realisation, for the PHY proposals using coherent
  detection
• Previous comparison used 100 channel realisation,
  as in IEEE Doc 15-04-0507-04-004b
• Performance comparison herein done with
• 0,1,2 cyclic chip extension
• 1,2,3 RAKE fingers

3
Updates

• Corrected 3-RAKE multipath performance for all
  proposals (due to programme bug in previous
  version)
• Included PSSS performance with Precoding
• Determined RMS Delay Spread threshold below which
  cyclic chip extension is not necessary
• Include 868 MHz multipath performance with raised
  cosine filter (roll-off factor 0.2)
• Include PER performance curves
• 915MHz Transmit PSD for COBI-16 868MHz
  Transmit PSD for COBI-8 PSSS
• Stated Recommendation based on realistic channel
  RMS delay spread, achievable Transmit PSD and PER
  performance

4
Candidates for Multipath Performance Comparison
(using Coherent Chip Despreading)
Code Set E16 G16 C8 F31
Candidate for 915MHz 915MHz 868MHz 868MHz
Description Orthogonal 16-DSSS 16-chip for Coh. Chip Despreading 8-chip for Coh. Chip Despreading PSSS
Proposer Helicomm I2R I2R Dr. Wolf Assoc.
Doc 04-314 04-507 04-507 04-121
Sym-Chip mapping Orthogonal Cyclic Odd Bit Inversion Cyclic Odd Bit Inversion Multi-code
Bit/sym 4 4 4 15
Chip/Sym 16 16 8 311 cyclic extension
Bit/chip 0.25 0.25 0.50 0.47
Root Sequence N.A. 2F53 5C 08B3E375
Source 15-04-0507-04-004b
5
System Parameters for low GHz Bands
Ch 0 868MHz band Ch 0 868MHz band Ch 0 868MHz band Ch 0 868MHz band Ch 1-10 906 924 MHz Band Ch 1-10 906 924 MHz Band
Bandwidth 600 kHz 600 kHz 600 kHz 600 kHz 2 MHz 2 MHz
Code Set Candidate 8-chip COBI C8 PSSS F31 8-chip COBI C8 PSSS F31 16-chip COBI G16 DSSS E16
Chip rate 400kcps 400kcps 450kcps 450kcps 1Mcps 1Mcps
Pulse shape Raised cosine (roll off 0.25) Raised cosine (roll off 0.25) Raised cosine (roll off 0.1) Raised cosine (roll off 0.1) Half-sine Half-sine
Modulation OQPSK BPSK /ASK OQPSK BPSK /ASK Offset QPSK OffsetQPSK
Data rate 200 kbps 187.5 kbps 240 kbps 225 kbps 250 kbps 250 kbps
6
Comparison Methodology

• Multipath robustness performance
• Investigation done with
• Zero, one and two Cyclic chip(s) extension
• One, two three RAKE fingers
• Bandwidth efficiency (bps / Hz)
• RF requirement
• Memory requirement

7
Multipath Realisations
1000 Channel Realisations at each RMS Delay Spread
8
Multipath Realisations
1000 Channel Realisations at each RMS Delay Spread
9
Proposed Symbol-to-Chip Mapping (8-chip Code Set
C8)
Decimal Value Binary Symbol Chip Value
0 0000 0 1 0 1 1 1 0 0 (Root 5C)
1 1000 0 0 1 0 1 1 1 0
2 0100 0 0 0 1 0 1 1 1
3 1100 1 0 0 0 1 0 1 1
4 0010 1 1 0 0 0 1 0 1
5 1010 1 1 1 0 0 0 1 0
6 0110 0 1 1 1 0 0 0 1
7 1110 1 0 1 1 1 0 0 0
8 0001 0 0 0 0 1 0 0 1
9 1001 1 0 0 0 0 1 0 0
10 0101 0 1 0 0 0 0 1 0
11 1101 0 0 1 0 0 0 0 1
12 0011 1 0 0 1 0 0 0 0
13 1011 0 1 0 0 1 0 0 0
14 0111 0 0 1 0 0 1 0 0
15 1111 0 0 0 1 0 0 1 0
The sequences are related to each other through
cyclic shifts and/or conjugation (i.e., inversion
of odd-indexed chip values)
10
Other Root Sequences (8-chip C8 for Coherent
Despreading only)

• The following Root Sequences are found through
  exhaustive search with identical low cross
  correlation and autocorrelation, in base 10
• 9 18 23 29 33 36 46 58
  66 71 72 92 111 113 116 123
  132 139 142 144 163 183 184 189
  197 209 219 222 226 232 237 246

11
DSSS Sequence E16
Source doc. IEEE 802.15-04-0314-02-004b
12
PSSS Sequence F31 (15 bit/32 chip)
Source doc. IEEE 802.15-04-0121-04-004b
13
Proposed Symbol-to-Chip Mapping (16-chip Code
Set G16)
Decimal Value Binary Symbol Chip Value
0 0000 0 0 1 0 1 1 1 1 0 1 0 1 0 0 1 1 (Root - 2F53)
1 1000 1 1 0 0 1 0 1 1 1 1 0 1 0 1 0 0
2 0100 0 0 1 1 0 0 1 0 1 1 1 1 0 1 0 1
3 1100 0 1 0 0 1 1 0 0 1 0 1 1 1 1 0 1
4 0010 0 1 0 1 0 0 1 1 0 0 1 0 1 1 1 1
5 1010 1 1 0 1 0 1 0 0 1 1 0 0 1 0 1 1
6 0110 1 1 1 1 0 1 0 1 0 0 1 1 0 0 1 0
7 1110 1 0 1 1 1 1 0 1 0 1 0 0 1 1 0 0
8 0001 0 1 1 1 1 0 1 0 0 0 0 0 0 1 1 0
9 1001 1 0 0 1 1 1 1 0 1 0 0 0 0 0 0 1
10 0101 0 1 1 0 0 1 1 1 1 0 1 0 0 0 0 0
11 1101 0 0 0 1 1 0 0 1 1 1 1 0 1 0 0 0
12 0011 0 0 0 0 0 1 1 0 0 1 1 1 1 0 1 0
13 1011 1 0 0 0 0 0 0 1 1 0 0 1 1 1 1 0
14 0111 1 0 1 0 0 0 0 0 0 1 1 0 0 1 1 1
15 1111 1 1 1 0 1 0 0 0 0 0 0 1 1 0 0 1
The sequences are related to each other through
cyclic shifts and/or conjugation (i.e., inversion
of odd-indexed chip values)
14
Other Root Sequences (8-chip G16 for Coherent
Despreading only)

• The following Root Sequences are found through
  exhaustive search with identical low cross
  correlation and autocorrelation, in base 10
• 1915 3566 12115 21038
  22715 31238 34297 42820
  44497 53420 61969 63620

15
Multipath Performance (COBI 16-chip)
_at_ 1Mcps using O-QPSK For 16-chip COBI Sequence,
No cyclic chip is needed when 3 RAKE is used.
16
Multipath Performance (COBI 8-chip)
For 8-chip COBI Sequence, 1 Chip Extension is
needed even with 3-RAKE, due to weaker
despreading strength (shorter code length).
17
Multipath Performance (DSSS)
For DSSS, No cyclic chip is needed when 3 RAKE is
used.
18
Multipath Performance (PSSS)
For PSSS, best performance with 2 RAKE fingers
1 chip extension. Precoding (according to
15-04-0121-04-004b) 3rd RAKE do not seem to
help.
19
What happened to PSSS?
Neighbouring parallel sequence is using M-Seq
with 2 cyclic shifts in PSSS parallel sequence
construction
Source doc. IEEE 802.15-04-0121-04-004b
While other schemes enjoy better multipath
performance with more RAKE fingers, PSSS can only
use up to 2 fingers as the 3rd RAKE is dominated
by adjacent parallel bit sequence. PSSS is
inter-parallel sequence interference limited
20
915MHz BandsCOBI-16 vs DSSS
21
915 MHz Coherent Receiver (BER Performance)

• Even upto 1.33us RMS Delay Spread
• 1 chip extension is NOT necessary for 16-chip
  sequence (COBI-16 DSSS) if sufficient RAKE
  fingers (at least 3) are used, even in dense
  multipath environment
• General performance comparison
• COBI sequence (16 chip) gt DSSS Sequence (16 chip)

22
915 MHz Coherent Receiver (PER Performance)

• Even upto 1.33us RMS Delay Spread
• 1 chip extension is NOT necessary for 16-chip
  sequence (COBI-16 DSSS) if sufficient RAKE
  fingers (at least 3) are used, even in dense
  multipath environment
• General performance comparison
• COBI sequence (16 chip) gt DSSS Sequence (16 chip)

23
Can Non-Coherent Detection be used for COBI-16?

• The COBI are designed to give best performance
  with coherent detection receiver. Can receiver
  employs Differential Chip detection?
• Yes, given 1 PER (20 octet packet) COBI sequence
  (16 chip) can handle multipath channels with RMS
  delay spread upto 0.3us for 915MHz bands using
  1Mcps, which normally corresponds to short range
  indoor environment

24
915 MHz Band Transmit PSD (COBI-16)

• Beyond fc /- 1.2 MHz, the highest sidelobe level
  is 38 dB below the total transmit power and 30
  dB below the highest point in the PSD
• Therefore, 10 dB of margin to the -20 dBr spec.
• For a device transmitting 10 dBm, there is 8 dB
  of margin to the -20 dBm absolute spec.
• Propose to be same as existing 915MHz Mask

25
868MHz BandCOBI-8 vs PSSS
26
868 MHz Transmit PSD Achievable Chip Rate

• Transmit PSD is dependent on the truncation of
  the pulse shaping filter
• To support 450kcps, Raised cosine filter with
  roll-off 0.1 is suggested
• Lets examine this effect on raised cosine filter
  with roll-off 0.1 using truncation of 20, 10 6
  chip periods

6-chip period
12-chip period
20-chip period
27
COBI-8 Transmit PSD
truncation of the raised cosine filter (Roll-off
0.1) at 20 chip periods
28
COBI-8 Transmit PSD
truncation of the raised cosine filter (Roll-off
0.1) at 10 chip periods
29
COBI-8 Transmit PSD
truncation of the raised cosine filter (Roll-off
0.1) at 6 chip periods
30
PSSS Transmit PSD
truncation of the raised cosine filter (Roll-off
0.1) at 20 chip periods
31
PSSS Transmit PSD
truncation of the raised cosine filter (Roll-off
0.1) at 10 chip periods
32
PSSS Transmit PSD
truncation of the raised cosine filter (Roll-off
0.1) at 6 chip periods
33
868 MHz Transmit PSD Achievable Chip Rate

• Transmit PSD is dependent on the truncation of
  the pulse shaping filter
• The longer the pulse shaping filter length, the
  higher the implementation complexity, the lower
  the PSD sidelobe level
• pulse shaping filter implementation complexity
  will have significant effect on meeting transmit
  PSD Mask and thus the achievable chip rate

• Transmit PSD mask
• f-fcgt0.3MHz, relative limit lt -50dBr, abs
  limit lt-36 dBm
• _at_ 450kcps, roll-off factor 0.1,
• COBI-8 PSSS does NOT satisfy Mask

34
868 MHz Transmit PSD Achievable Chip Rate

• Lower chip rate has to be sought, such that
  transmit PSD requirement can be met with
  reasonable transmit implementation complexity
• 400kcps with raised cosine filter (roll-off 0.25)
  is proposed

35
PSSS Transmit PSD _at_ 400kcps

• Transmit PSD requirement can be met with
  reasonable transmit implementation complexity.
  Data rate 400kcps 15/32 187.5kbps

36
COBI-8 Transmit PSD _at_ 400kcps

• Transmit PSD requirement can be met with
  reasonable transmit implementation complexity.
  Data rate 400kcps 4/8 200kbps

37
Should COBI-8 use O-QPSK or BPSK?

• _at_ 400kcps, Raised Cosine roll-off factor 0.25,
  O-QPSK does give less amplitude variation across
  symbol than BPSK. Thus, O-QPSK is preferred

O-QPSK
BPSK
PAPR4dB
PAPR6dB
38
Amplitude variation for PSSS with Precoding

• _at_ 400kcps, Raised Cosine roll-off factor 0.25,
  PSSS with precoding does have larger amplitude
  variation

PAPR gt 10dB
39
868 MHz COBI-8 vs PSSS(PER Performance
Comparison)

• Raised cosine filter
• (roll-off factor 0.25)
• _at_ 400 kcps
• Even upto 1.33us RMS Delay Spread
• Similar performance between COBI-8 and PSSS at 1
  PER

40
Can Non-Coherent Detection be used for COBI-8?

• The COBI are designed to give best performance
  with coherent detection receiver. Can receiver
  employs Differential Chip detection?
• Yes, given 1 PER (20 octet packet), COBI
  sequence (8 chip) can handle multipath channels
  with RMS delay spread upto 0.3us for 868MHz band
  using both 400kcps (roll-off factor 0.25), at
  even shorter range

41
868 MHz COBI-8 vs PSSS
PSSS COBI-8
Transmit amplitude distribution Larger range Lower range
Multipath performance Similar at low RMS delay spread Better in large RMS delay spread esp. at lt 1 PER
Bandwidth efficiency 15/32 0.47 0.50
42
Multipath Performance Summary (Coherent Chip
Despreading)

• To combat inter-chip interference due to channel
  delay spread with RMS delay spread upto 1.33us
  (e.g. industry application space)
• COBI 16-chip (O-QPSK with half-sine pulse
  shaping) is recommended for 915MHz bands
• COBI 8-chip (O-QPSK with raised cosine pulse
  shaping roll-off 0.25) is recommended for 868MHz
  bands.
• RAKE combining (with at least 3 fingers) is
  necessary in receiver to combine path diversity
  (this does not affect standard)
• Few RAKE fingers can be used in realistic
  channels with lower delay spread
• Differential chip despreading can also be used in
  shorter transmission range environment,e.g.
  residential space, where multipath channel RMS
  delay spread is upto 0.3us

43
Summary of Comparsion
Code Set E16 G16 C8 F31
Candidate for 915MHz 915MHz 868MHz 868MHz
Description Orthogonal 16-DSSS 16-chip for Coh. Chip Despreading 8-chip for Coh. Chip Despreading PSSS
Proposer Helicomm I2R I2R Dr. Wolf Assoc.
Doc 04-314 04-507 04-507 04-121
Sym-Chip mapping Orthogonal Cyclic Odd Bit Inversion Cyclic Odd Bit Inversion Multi-code
Bit/sym 4 4 4 15
Chip/Sym 16 16 8 311 cyclic extension
Bit/chip 0.25 0.25 0.50 15/32 0.47
Multipath performance Best Best Better Good
Memory requirement High 16 sequence Low Single sequence Low Single sequence Low Single sequence
RF linearity requirement Low Low Moderate High
Note Red - desirable
44
System Parameters for low GHz Bands
Recommended
Ch 0 868MHz band Ch 0 868MHz band Ch 0 868MHz band Ch 0 868MHz band Ch 1-10 906 924 MHz Band Ch 1-10 906 924 MHz Band
Bandwidth 600 kHz 600 kHz 600 kHz 600 kHz 2 MHz 2 MHz
Code Set Candidate 8-chip COBI C8 PSSS F31 8-chip COBI C8 PSSS F31 16-chip COBI G16 DSSS E16
Chip rate 400kcps 400kcps 450kcps 450kcps 1Mcps 1Mcps
Pulse shape Raised cosine (roll off 0.25) Raised cosine (roll off 0.25) Raised cosine (roll off 0.1) Raised cosine (roll off 0.1) Half-sine Half-sine
Modulation OQPSK BPSK /ASK OQPSK BPSK /ASK Offset QPSK OffsetQPSK
Data rate 200 kbps 187.5 kbps 240 kbps 225 kbps 250 kbps 250 kbps
Does not meet Transmit PSD Mask
45
Supporting Materials
46
AWGN Performance
47
Flat Fading Performance
48
Coherent Receiver Multipath Performance
What leads to Multipath robustness? Frequency
selectivity leads to Inter-chip interference, and
that is the killer. To overcome, code must have
good autocorrelation properties, i.e. low
sidelodes
49
How these codes achieve Multipath robustness?

• COBI, maintain constant module, can at best
  achieve zero auto-correlation within 2 chips
  from cor. Peak that is good enough to handle ICI
  of upto 2 chip periods
• DSSS, comprising Walsh sequences, is not designed
  with auto-correlation sidelodes in mind
• PSSS, uses flexibility in amplitude to achieve
  low (zero?) auto-correlation throughout for each
  parallel sequence. However, it is inter-parallel
  sequence interference limited