1. Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs)
Submission Title: Code Shift Keying for UWB PHY
Date Submitted: January 18, 2006
Source: [Serhat Erküçük (1), Dong In Kim (1), Kyung Sup Kwak (2)]
Company: [(1) Simon Fraser University, (2)UWB-ITRC, Inha University]
Address: [(1) School of Engineering Science, 8888 University Drive, Burnaby, BC
V5A 1S6, Canada (2) 253 Yonghyun-Dong, Nam-Gu, #401, Venture Bldg.
Incheon, , Korea]
Voice: [+1 (604) ], Fax: [(1) +1 (604) (2) ]
[(1) (1) (2)
Abstract: [Code shift keying randomizes burst locations without changing basic principles of the mandatory modulation format, and stabilizes the instantaneous PER performance in the presence of uncoordinated piconets]
Purpose: [To assist the group in evaluating PER performances in the presence of multiple piconets]
Notice: This document has been prepared to assist the IEEE P 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 P

2. Code Shift Keying for UWB-PHY
A Comparative Study on PER Performances in the Presence of Multiple Piconets
Code Shift Keying for UWB-PHY
Serhat Erküçük (1), Dong In Kim (1), Kyung Sup Kwak (2)
Simon Fraser University, Canada
UWB-ITRC, Inha University, Korea

3. Motivation Modulation characteristic
(2PPM+BPSK) uses fixed guard times.
Problem in the presence of multiple SOPs
Fixed guard times may cause catastrophic collisions.
Catastrophic collisions affect instantaneous BER and PER performances.
Solution
Burst location randomization using code shift keying (CSK) (2CSK+BPSK)

4. Outline (1) Highlights of mandatory modulation (2PPM + BPSK)
(2) Introduction of (2CSK + BPSK)
(3) Comparison of two modulation formats
 Average PERs
 Average uncoded/coded BERs
 PERs under different scenarios (worst collision cases)
(4) Conclusion

5. (2PPM+BPSK): Symbol Structure
Chip duration = ~2ns
Burst = 16 consecutive pulses
Peak PRF = 494 MHz [REF: 05/706r1]
Polarity scrambling
Target symbol rate = 1 MHz  32 burst durations
S = = S
1 chip ~ 2 ns
burst duration (16 chips)
symbol duration (32 burst duration)

6. (2PPM+BPSK): Modulation
TG4a draft defines 16 TH locations  Allows guard time
32 burst durations  burst locations 0 – 7: transmit a “0”
burst locations 8 – 15: guard time
burst locations 16 – 23: transmit a “1”
burst locations 24 – 31: guard time
TH sequence  3 shift register outputs (8 possible TH locations)
[REF: UWB PHY draft, TG4a draft]
TG4a draft defines 4 TH locations for length 127 preamble (Table 39i)
 No guard time
 Possibly ISI
 Consider length 31 preamble (Tables 39d-h)

7. (2PPM+BPSK): Modulation
COH.
NON COH. 00
0 - 01 10
1 - 11
PPM bit (seen by coherent and non coherent receiver)
BPSK bit (seen by coherent receiver only) S S S S
Guard Time
Guard Time 1 15 16 17 31 7 8 23 24 -S -S -S -S 1 15 16 17 31 7 8 23 24 S S S S 1 15 16 17 31 7 8 23 24 -S -S -S -S 1 15 16 17 31 7 8 23 24
burst
symbol duration

8. (2PPM+BPSK): Modulation

9. (2PPM+BPSK): 2SOPs
Effect of fixed guard times on PER for multiple SOPs
Previous studies consider average PER performances
 E.g. [REF: 05/515r1]
PER depends on collision between users
 E.g. worst case collisions if they start operating at the same time
Fixed guard times  Limited randomization in collision event
 Instantaneous BER and PER performances may degrade
 Possibly large number of packet losses
 Randomize the location of fixed guard times to avoid the worst case scenarios

10. (2PPM+BPSK): 2SOPs Case-1: Best Case Scenario No Collision
User-1 Ts
2Ts S S S S
User-2 Ts
2Ts
No Collision
Case-2: Worst Case Scenario S S S S
User-1 Ts
2Ts S S S S
User-2 Ts
2Ts
More likely
collisions
More likely
collisions

11. Randomized collisions Randomized collisions
(2CSK+BPSK): Random Positions
(2CSK+BPSK) randomizes the location of fixed guard times
 Each user has 2 user-specific TH codes
 TH codes span the whole symbol range
 Guard times redefined in an adaptive manner
 Collision event randomized
Code1: the code can choose one of 32 locations
Randomized collisions
Code1: the code can choose one of 32 locations
Randomized collisions S S
User-1 Ts
2Ts
Code2: the code can choose one of 32 locations
Code2: the code can choose one of 32 locations
Code1: the code can choose one of 32 locations
Code1: the code can choose one of 32 locations S S
User-2 Ts
2Ts
Code2: the code can choose one of 32 locations
Code2: the code can choose one of 32 locations
asynchronism

12. 2CSK vs. 2PPM: Collision Event
Probability of collisions between two SOPs for different time asynchronism values
 2CSK: independent of time asynchronism
 2PPM: strictly depends on time asynchronism

13. (2CSK+BPSK): Code Design
Constraint: The duration of guard time used for 2PPM kept the same
Step-1: Randomly choose one of the 32 locations for the first entry of code-1 S Ts
2Ts
Step-2: Determine the guard band and possible locations for the first entry of code-2 S Ts
2Ts
Guard band
Possible locations
Step-3: Randomly choose the location of the first entry of code-2 within the green area S S Ts
2Ts
Guard band

14. (2CSK+BPSK): Code Design
Step-4: Determine the guard band and possible locations for the second entry of code-1 S S Ts
2Ts
Guard band
Step-5: Randomly choose the location of the second entry of code-1 within the green area S S S Ts
2Ts
Guard band
Step-6: Determine the guard band and possible locations for the second entry of code-2 S S S Ts
2Ts
Guard band
Guard band

15. (2CSK+BPSK): Code Design
Step-7: Randomly choose the location of the second entry of code-2 within the green area S S S S Ts
2Ts
Guard band
Guard band
Step-8: Determine the guard band and possible locations for the third entry of code-1 S S S S Ts
2Ts
Guard band
Guard band
Step-8+: Repeat the same procedure until both codes are determined for the length
of one packet (32 octets).
For each user, repeat the same code design procedure.

16. Receivers for 2PPM and 2CSK
Coherent/non-coherent receivers: Collect energy from two PPM locations
 E.g.: TH code is “3” for the first symbol S S Ts
2Ts 3 19
2CSK
Coherent/non-coherent receivers: Collect energy from two code shift locations
 E.g.: Code locations of the 1st symbol from Slide-13 S S Ts
2Ts

17. Simulation Parameters
Sampling rate = 16 GHz
Monocycle: Root raised cosine with beta=0.6
Channel realizations - CM1
2-SOP & 3-SOP performance – time asynchronism uniformly distributed over one symbol period for average PER
1500 packets (32 octets) for each SIR value
½ rate Viterbi decoder (k=3)
Types of receiver - Coherent Receiver and Non-coherent Receiver
Acquisition assumed for both receivers
Coherent receiver – 4-tap RAKE fingers
Non-coherent receiver – (integration time = 1 burst duration)

18. Simulation Results-(1a)

19. Simulation Results-(1b)

20. Simulation Results-(1c)

21. Simulation Results-(1d)

22. Simulation Results-(2a)

23. Simulation Results-(2b)

24. Simulation Results-(3a)

25. Simulation Results-(3b)

26. Observations and Conclusion
(2CSK+BPSK)  An optional implementation to (2PPM+BPSK)
Main difference  It randomizes the location of the transmitted bursts.
Main advantage  Collision event independent of timing asynchronism
 PER stable for various collision scenarios
(2CSK+BPSK) and (2PPM+BPSK) have similar average PER values.
(2PPM+BPSK)  Worse instantaneous PER performance
 Less uniform distribution among packets
Summary: Code shift keying randomizes burst locations without changing basic principles of the mandatory modulation format, and stabilizes the instantaneous PER performance in the presence of uncoordinated piconets.