Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs)

1. Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs) Submission Title: [performance-analysis-reservation-based-low-power-function] Date Submitted: [September 18, 2007] Source: [Chang Sub Shin*, Gangja Jin*, Yuan Dao**, Yoh-Han Lee**, SeungHyun Whang**] Company [*ETRI, **SNR Inc] Address [161 Gajeong-dong Yuseong-gu, Daejeon, Korea] Voice:[+82-42-860-1668], FAX: [+82-42-869-1712], E-Mail:[shincs@etri.re.kr] Re: [] Abstract: [performance analysis for reservation-based-low-power-function in low rate WPAN mesh.] Purpose: [to discuss low power function] 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. <Chang Sub Shin>, <ETRI>

2. Performance Analysis ofReservation based Low Power Function (Low Power Support for low-rate IEEE802.15.5) ChangSub Shin*, Gangja Jin*, Yuan Dao**, Yoh-Han Lee**, Seung-Hyun Whang** * ETRI, ** SNR Inc <Chang Sub Shin>, <ETRI>

3. Contents • Overview of previous proposal • Additional issue • Time synchronization • MAC PIB value • Frame Format • Performance Analysis • Implementation result of synchronization • Implementation result of low power function • Reference of performance analysis <Chang Sub Shin>, <ETRI>

4. Overview of previous proposal - 1 • Synchronous and Reservation method for end-to-end multi-hop transmission • Consist of active period and sleep period • If a node have a packet, It transmit its REQ control packet in active period and then transmit data in sleep period • REQ control packet : inform next node to transmit and prevent other neighbor node’s transmission • RSP control packet : respond to REQ or RSP sender and inform next hop node to transmit <Chang Sub Shin>, <ETRI>

5. For one-hop unicasting Other neighbor nodes Listen Sleep Sleep Data REQ RSP Sleep Sender node (A) Ack Frame Data REQ RSP Destination node (B) Sleep RSP Listen Sleep Sleep Other node (C) Listen Sleep Other node (D) Listen Period Sleep Period Duty Cycle <Chang Sub Shin>, <ETRI>

6. For broadcasting or multicasting Listen Sleep Data Sleep Sender node (A) Data Neighbor node #1 Sleep Data Neighbor node #2 Sleep Data Sleep Neighbor node #3 Listen Period Sleep Period Duty Cycle <Chang Sub Shin>, <ETRI>

7. for multi-hop unicasting other neighbor nodes Listen Sleep Sleep Data (1) REQ RSP Sleep Sleep Sender node (A) Data Data (2) REQ RSP RSP 1st Forward node (B) Sleep Sleep Data Data (3) RSP RSP RSP Listen Sleep Listen Sleep 2nd Forward node (C) Data RSP RSP Listen Sleep Listen Sleep Destination node (D) Ack Frame Listen Period Sleep Period Duty Cycle <Chang Sub Shin>, <ETRI>

8. Multi-hop low power and minimum end-to-end latency <Chang Sub Shin>, <ETRI>

9. Contents • Overview of previous proposal • Additional issue • Time synchronization • MAC PIB value • Frame Format • Performance Analysis • Implementation result of synchronization • Implementation result of low power function • Reference of performance analysis <Chang Sub Shin>, <ETRI>

10. Time synchronization - methods • Using MAC Primitive with timestamp value • MCPS-DATA.confirm : transmitted time • MCPS-DATA.indication : received time • Methods • Method 1 : unicast 3 way message exchange • Method 2 : broadcast 3 way message exchange • Method 3 : 2 successive synch packet broadcast • Method 4 : 1 synch packet broadcast <Chang Sub Shin>, <ETRI>

11. Timestamp of IEEE802.15.4 • Timestamp of IEEE802.15.4 • Length : 3 bytes • Unit : 16 us (1 symbol) • Maximum interval : 4.5 minutes • Timestamp interrupt signal timing in PPDU frame Timestamp interrupt Signal timing <Chang Sub Shin>, <ETRI>

12. Time synchronization – method 1 A A B T4 T1 A (1) (2) (3) with T1, T4 time value B (4) Synch with A’s timer T2 T3 <Chang Sub Shin>, <ETRI>

13. Time synchronization – method 2 A B C D TA4-TC3 TA4-TD3 TA1 TA4-TB3 (5) with TA1, TA4-TB3, TA4-TC3, TA4-TD3 time value A (1) (2) B (6) Synch with A’s timer (3) TB3 TB2 C (6) Synch with A’s timer (4) TC2 TC3 D (6) Synch with A’s timer TD2 TD3 <Chang Sub Shin>, <ETRI>

14. Time synchronization – method 3 TA2 TA1 A A (1) (2) with TA1 time value B (3) Synch with TA1 and TB1 TB2 TB1 D B C C (3) Synch with TA1 and TC1 TC2 TC1 D (3) Synch with TA1 and TD1 TD2 TD1 <Chang Sub Shin>, <ETRI>

15. MAC PIB value • Default MAC PIB during Listen period • macMaxCSMABackoffs = 4 • macMinBE = 3 • Temporal MAC PIB during Sleep period • macMaxCSMABackoffs = 2 • macMinBE = 0 • using MLME-SET.request primitive to change value <Chang Sub Shin>, <ETRI>

16. Frame Format <Chang Sub Shin>, <ETRI>

17. Contents • Overview of previous proposal • Additional issue • Time synchronization • MAC PIB value • Frame Format • Performance Analysis • Implementation result of synchronization • Implementation result of low power function • Reference to performance analysis <Chang Sub Shin>, <ETRI>

18. Test environment of time synch - 1 (1) (2) (3) (4) 0 1 2 3 4 Coordinator Level 1 Level 2 Level 3 Level 4 Time synch event node Time synch event node <Chang Sub Shin>, <ETRI>

19. Test environment of time synch - 2 <Chang Sub Shin>, <ETRI>

20. Test environment of time synch - 3 • Hardware platform • MCU : MSP430 • RF : CC2420 • Oscillator max skew : 20 ppm • Synch packet interval • every 8 second test • every 3 minute test • test packet interval • Every 1.1 second test • test result graph • y-axis is sync error in microsecond unit, • x-axis is in 1.1 second unit <Chang Sub Shin>, <ETRI>

21. Test scenario • Multi-hop time synch procedure • Coordinator send synch packet to level 1 node. • Level 1 node receiving synch packet will forward it to next level • Subsequently, every each level nodes receiving synch packet will forward it to next level. • Synch error measurement • Periodical multi-hop time synch procedure • Event node broadcast event packet to simulate an event happen at a certain time(every 8s and every 4 min) • All nodes including coordinator receive the broadcast packet at the same time, and send receiving timestamp back to event node. • Event node prints out the timestamps via RS232 for synch error analysis <Chang Sub Shin>, <ETRI>

22. synchronization error (8s interval) <Chang Sub Shin>, <ETRI>

23. absolute synchronization error (8s interval) <Chang Sub Shin>, <ETRI>

24. synch error analysis (8second interval) • error unit is us • synch packets are sent every 8 seconds, and event samples are taken every 1.1 second. • Theoretically, average synch error should be zero, but the skew difference will cause the average synch error deviate from zero. • the standard deviation of synch error which reflects the jitter tells more about the synch accuracy. It shows the synch error is accumulated every hop. <Chang Sub Shin>, <ETRI>

25. synchronization error (4minutes interval) <Chang Sub Shin>, <ETRI>

26. synchronization error (4minutes interval) <Chang Sub Shin>, <ETRI>

27. synch error analysis (4minutes interval) • The difference on clock skew will slowly take effect when synch interval increases. • Once synch packet is sent, synch errors reduce to zero. When time passes, each local timers of the nodes in each level increases at a slightly different rate. This rate uncertainty is caused by the frequency uncertainty of crystal oscillator which is 20 ppm according to the datasheet. • In the following picture, node3 is slower than the coordinator, and other nodes are faster than coordinator at different speed. • It’s possible to use skew compensation to reduce the long term drift effect caused by skew difference. But it requires extra computation overhead.  • However, if the synchronization interval is very small, the synch error caused by the skew uncertainty can be negligible <Chang Sub Shin>, <ETRI>

28. Test environment of low power function • Test value • Chain topology based 4 nodes test • 5% Duty Cycle • Period : 500, 1000, 2000, 5000 msec (4 cases) •  2.5% Duty Cycle • Period = 1000, 2000, 5000 msec (3 cases) • procedure of multi-hop low power function • Periodical multi-hop time synch procedure • All nodes make synch • Source node generate packet • measurement • Every node have timer for power consumption measurement • Calculate active time and sleep time • Result : end-to-end delay, power consumption <Chang Sub Shin>, <ETRI>

29. Implementation result of low power function - 1 <Chang Sub Shin>, <ETRI>

30. Implementation result of low power function -2 <Chang Sub Shin>, <ETRI>

31. Reference of performance analysis - 1 • Refer to “Li, J., Blake, C., Couto, D., Lee, H.I., Morris, R.: Capacity of ad hoc wireless networks. In: ACM MobiCom (2001)” • MAC interference among a chain of nodes. The solid-line circle denotes a node’s valid transmission range. The dotted-line circle denotes a node’s interference range. Node 4’s transmission will corrupt node 1’s transmissions at node 2. • Effective capacity of a forwarding chain topology becomes just 1/3 of the effective capacity of a single-hop connection <Chang Sub Shin>, <ETRI>

32. Reference of performance analysis - 2 • Refer to “Tony Sun, Ling-Jyh Chen, Chih-Chieh Han, Guang Yang and Mario Gerla: Measuring effective capacity of IEEE 802.15.4 beaconless mode” • Results on Multihop forwarding chain testbed • effective end-to-end capacity of a chain topology decreases as the hop length increases <Chang Sub Shin>, <ETRI>