![Page 1: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/1.jpg)
Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu
SenSys 2014
ZiSenseTowards Interference Resilient Duty
Cycling in Wireless Sensor Networks
![Page 2: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/2.jpg)
2
Existing low-power method
• Radio: major source of energy consumption
• Duty cycling: Low Power Listening– Schedule nodes: sleep (radio off) or wake up (radio on)
Sender
Receiver
D D D
DC C
Radio on C CCA D Data packet ACK
![Page 3: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/3.jpg)
3
CCA (Clear Channel Assessment)
• Decides a node wake up or not– Energy detection by threshold– High energy on channel Busy channel Possible
transmissions Wake up nodes– Effective energy efficient method in clean
environments
Interference is ignored!
![Page 4: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/4.jpg)
4
Channel overlapping
2401.5 2480.5
Ă ĂBluetooth Channels79771 2 3
2445 2465
Microwave oven
No clean channel away from interference all the time
![Page 5: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/5.jpg)
5
Sender
Radio on
C
ReceiverCC
InterferenceI I I I I
Time out
C CCA I Interference
Impacts on LPL
• False wake-up problem– Heterogeneous interference unnecessarily wakes up the
receiver!
False wake-up, energy waste
![Page 6: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/6.jpg)
6
Adaptive Energy Detection Protocol[1]
[1] Mo Sha et al. Energy-efficient low power listening for wireless sensor networks in noisy environments, IPSN’13
No wake-up
K readings ≥ threshold wake up
RSSI
wakeup threshold
0 0 0 1 1 1 1 1 1 1 1 1 0 00
Poll CCA pin N times
RSSI
wakeup threshold
0 0 0 1 0 000 0 0 0 0 0 0 0
![Page 7: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/7.jpg)
7
Limitation
Min packet > Wake-up threshold > Max noise
RSSI
wakeup threshold
MAX noise
MIN packet
RSSI
MAX noiseMIN packet
![Page 8: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/8.jpg)
8
Key insight
• Energy detection is too simple to filter out the interference
– High energy on channel Busy channel possible ZigBee transmissions
– No matter how good the threshold is set, false wake-up problem still exists
Can we recognize ZigBee by some other information instead
of energy?
![Page 9: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/9.jpg)
9
Roadmap
• Background• Motivation• Observations • Design of ZiSense• Evaluation• Conclusion
![Page 10: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/10.jpg)
10
Recognize ZigBee
• Limited information provided by the radio hardware– RSSI (Received Signal Strength Indicator)
• Key observation:– Different technologies in 2.4GHz leave distinguishable
patterns on the time-domain RSSI sequence.
![Page 11: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/11.jpg)
11
Observations
![Page 12: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/12.jpg)
12
Feature #1: on-air time
Shorter on-air time
Valid range of on-air
time
longer on-air time
![Page 13: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/13.jpg)
13
Feature #2: packet interval
Shorter packet interval
Fixed packet interval
longer packet interval
![Page 14: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/14.jpg)
14
Feature #3: PAPR (Peak-to-Average Ratio)
Large variationFlat
sequence
![Page 15: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/15.jpg)
15
Feature #4: RSSI < noise floor
TRUEFALSE
![Page 16: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/16.jpg)
16
Roadmap
• Background• Motivation• Observations • Design of ZiSense• Evaluation• Conclusion
![Page 17: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/17.jpg)
17
ZiSense: Design
• Sense ZigBee and wake up nodes only when ZigBee signal is detected.
![Page 18: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/18.jpg)
18
ZiSense: Identify ZigBee
• Adopt the decision tree as the classification algorithm
![Page 19: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/19.jpg)
19
ZiSense: Identify ZigBee
• Rule-based identification algorithm– Simple yet effective, because features are stable– Universal to directly use in another system, without
training. • Four conditions as rules
– C1 : PAPR ≤ PAPRZigBee;– C2 : Ton ≥ Tmin; – C3 : |MPI − MPIvalid| ≤ δ; – C4 : UNF = FALSE.
• Valid conditions (C1, C2, C3, C4)– (T,T,T,T) – (F,T,T,T)– (T,F,T,T)
in strict conformance with valid ZigBee sequencedeal with some corrupted features
![Page 20: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/20.jpg)
20
ZiSense: Identify ZigBee
• Decision tree trained by C4.5
![Page 21: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/21.jpg)
21
• TP(True Positive): correctly recognize ZigBee signals
• FN(False Negative): missing valid ZigBee packets• TN(True Negative): correctly recognize non-ZigBee• FP(False Positive): false wake-ups
Identification Accuracy
Algorithm TP rate FN rate TN rate FP rate
Rule-based 97.5% 2.5% 97.6% 2.4%
Decision tree 97.3% 2.7% 99.1% 0.9%
![Page 22: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/22.jpg)
22
ZiSense: Identify ZigBee
• Comparable accuracy– Compared with specially trained decision tree
• Effective algorithm: – False positive (false wake-up) rate: 2.4%– False negative (missing packet) rate: 2.5%
• General algorithm: – Stable features which are extracted from hardware and
standard specifications– Directly used in other systems
![Page 23: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/23.jpg)
23
Roadmap
• Background• Motivation• Observations • Design of ZiSense• Evaluation• Conclusion
![Page 24: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/24.jpg)
24
Different Interference Type
• False wake-up ratios under different heterogeneous interference environments
![Page 25: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/25.jpg)
25
Different Interference Intensity• Duty cycle = radio-on time / total
time
![Page 26: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/26.jpg)
26
Integrated with Routing Protocol• Integrated with CTP
– 41 nodes deployed in a 50*100m2 office– Each method runs 24 hours
![Page 27: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/27.jpg)
27
Integrated with Routing Protocol• Integrated with CTP
– Improve energy efficiency without extra overhead
![Page 28: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/28.jpg)
28
Side effects
• NO Side effects
End-to-end ETX Routing link RSSI
![Page 29: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/29.jpg)
29
Conclusion
• ZiSense: interference-resilient duty cycling technique– Solve false wake-up problem – Recognize valid ZigBee signals by only RSSI sequence
• Keep low energy consumption, ZiSense consumes
– BoX-MAC-2: 24% (extreme case) and 38% (office)
– AEDP: 27% (extreme case) and 50% (office)
![Page 30: Xiaolong Zheng, Zhichao Cao, Jiliang Wang, Yuan He, and Yunhao Liu SenSys 2014 ZiSense Towards Interference Resilient Duty Cycling in Wireless Sensor Networks](https://reader036.vdocument.in/reader036/viewer/2022062421/56649c7e5503460f94933c5e/html5/thumbnails/30.jpg)
Thank You!