23
Robust Topology Control for Indoor Wireless Sensor Networks Greg Hackmann, Octav Chipara, and Chenyang Lu SenSys 2009 S Slides from Greg Hackmann at Washington University in Saint Louis

Robust Topology Control for Indoor Wireless Sensor Networks

  • Upload
    savea

  • View
    48

  • Download
    0

Embed Size (px)

DESCRIPTION

Robust Topology Control for Indoor Wireless Sensor Networks. Greg Hackmann, Octav Chipara , and Chenyang Lu SenSys 2009 S S lides from Greg Hackmann at Washington University in Saint Louis. Motivation. Goal: reducing transmission power while maintaining satisfactory link quality - PowerPoint PPT Presentation

Citation preview

Page 1: Robust Topology Control for Indoor Wireless Sensor Networks

Robust Topology Control for Indoor Wireless Sensor Networks

Greg Hackmann, Octav Chipara, and Chenyang LuSenSys 2009 S

Slides from Greg Hackmann at Washington University in Saint Louis

Page 2: Robust Topology Control for Indoor Wireless Sensor Networks

2

Motivation

Goal: reducing transmission power while maintaining satisfactory link quality

But it’s challenging: Links have irregular and probabilistic properties Link quality can vary significantly over time Human activity and multi-path effects in indoor networks

Page 3: Robust Topology Control for Indoor Wireless Sensor Networks

3

Outline

Empirical study Algorithm Implementation and evaluation Conclusion

Page 4: Robust Topology Control for Indoor Wireless Sensor Networks

4

Existing Empirical Studies

Many studies explore link performance at a fixed transmission power [Srinivasan 2006], [Woo 2003], [Reijers 2004], [Zhou 2004], [Lai 2003]

[Son 2004] evaluates older Chipcon CC1000 radios [Lin 2006] uses a indoor environment with line-of-sight

Our study considers 802.15.4-compliant CC2420 radios in an office environment

Page 5: Robust Topology Control for Indoor Wireless Sensor Networks

5

Related Topology Control Schemes

Many schemes based on theoretical radio properties [Li 2001], [Li 2005], [Blough 2006], [Gao 2008]

Power Control with Blacklisting (PCBL) [Son 2004] Collect extensive PRR (packet reception rate) data over all links at

all power levels during bootstrapping phase Statically reduce TX power on high-quality links Blacklist low-quality links

Adaptive Transmission Power Control (ATPC) [Lin 2006] Offline, compute correlation between RSSI/LQI and PRR Online, model impact of TX power at sender on RSSI/LQI at

receiver

Page 6: Robust Topology Control for Indoor Wireless Sensor Networks

6

RSSI & LQI

RSSI (Received Signal Strength Indicator) Remaining energy in the received signal

LQI (Link Quality Indicator) CC2420 radio provides LQI following 802.15.4 standard Can accurately reflect the packet error rate (PER) of a channel, if

measured for 100 packets

Page 7: Robust Topology Control for Indoor Wireless Sensor Networks

7

Advantages of Topology Control Testbed Topology

0 dBm-15 dBm-25 dBm

Page 8: Robust Topology Control for Indoor Wireless Sensor Networks

8

Is Per-Link Topology Control Beneficial?

Impact of TX power on PRR

3 of 4 links fail @ -10 dBm ...

... but have modest performance @ -5 dBmInsight 1: Transmission power should be set on a per-

link basis to improve link quality and save energy.

Page 9: Robust Topology Control for Indoor Wireless Sensor Networks

9

What is the Impact of Transmission Power on Contention?

Highcontention

Low signal strength

Insight 2: Robust topology control algorithms must avoid increasing contention under heavy network load.

Page 10: Robust Topology Control for Indoor Wireless Sensor Networks

10

Is Dynamic Power Adaptation Necessary?

Link 110 -> 139

Page 11: Robust Topology Control for Indoor Wireless Sensor Networks

11

Can Link Stability Be Predicted? Long-Term Link Stability

Insight 3: Robust topology control algorithms must adapt their transmission power in order to maintain

good link quality and save energy.

Page 12: Robust Topology Control for Indoor Wireless Sensor Networks

12

Are Link Indicators Robust Indoors?

Two instantaneous metrics are often proposed as indicators of link reliability: Received Signal Strength Indicator (RSSI) Link Quality Indicator (LQI)

Can you pick an RSSI or LQI threshold that predicts whether a link has high PRR or not?

Page 13: Robust Topology Control for Indoor Wireless Sensor Networks

13

Are Link Indicators Robust Indoors? Links 106 -> 129 and 104 -> 105

RSSI threshold = -85 dBm, PRR threshold = 0.9

4% false positive rate62% false negative rate

RSSI threshold = -84 dBm, PRR threshold = 0.9

66% false positive rate6% false negative rateInsight 4: Instantaneous LQI and RSSI are not robust

estimators of link quality in all environments.

Page 14: Robust Topology Control for Indoor Wireless Sensor Networks

14

Summary of Insights

1. Set transmission power on a per-link basis2. Avoid increasing contention under heavy network load3. Adapt transmission power online4. LQI and RSSI are not robust estimators of link quality in all

environments

Page 15: Robust Topology Control for Indoor Wireless Sensor Networks

15

Adaptive and Robust Topology control (ART)

ART:1. Adjusts each link’s power individually 2. Detects and avoids contention at the sender3. Tracks link qualities in a sliding window, adjusting

transmission power at a per-packet granularity4. Does not rely on LQI or RSSI as link quality estimators

5. Is simple and lightweight by design 392 bytes of RAM, 1582 bytes of ROM, often zero network

overhead

Page 16: Robust Topology Control for Indoor Wireless Sensor Networks

16

ART Algorithm Outline

TRIAL

Trying a lowerpower

INITIALIZING

Packet windowfilling

STEADY

Link quality withinthreshold

Packet window full

Link quality abovethreshold

Link quality goes back below threshold

Link quality fallsbelow threshold

Link quality

stays

at/above threshold

Page 17: Robust Topology Control for Indoor Wireless Sensor Networks

17

Avoiding Contention

Naïve policy: When link quality falls below threshold, then increase power level

But what if this makes things worse? Remember, higher power → more contention

Initially increase power when link quality is too low, but remember how many failures were recorded in window

If # of failures is worse than last time, then flip direction and decrease power instead

Page 18: Robust Topology Control for Indoor Wireless Sensor Networks

18

Implementation Details

Implemented using TinyOS 2.1 CVS on top of the MAC Layer Architecture [Klues 07]

Sits below routing layer -> has been tested with Collection Tree Protocol [Gnawali 2008]

Deployed on Jolley Hall testbed for three experiments: Link-level High contention Data collection (not presented here)

Page 19: Robust Topology Control for Indoor Wireless Sensor Networks

19

Link-Level Performance

Selected 29 links at random from 524 detected in empirical study

Transmitted packets round-robin over each link in batches of 100, cycled for 24 hours (15000 packets/link)

PRR Avg. Current

Max Power 56.7% (σ = 2.5%) 17.4 mA (σ = 0)ART 58.3% (σ = 2.1%) 14.9 mA (σ =

0.32)

Page 20: Robust Topology Control for Indoor Wireless Sensor Networks

20

Handling High Contention

Select 10 links at random from testbed Send packets over all 10 links simultaneously as possible

(batches of 200 packets for 30 min.)

Compare again against PCBL and max-power Also run ART without “gradient” optimization to isolate its

effect on PRR

Page 21: Robust Topology Control for Indoor Wireless Sensor Networks

21

Handling High Contention Packet Reception Rate and Energy Efficiency

Max-Power PCBL ART ART (w/o gradient)0

0.2

0.4

0.6

0.8Pa

cket

Rec

eptio

n Ra

te

Max-Power PCBL ART ART (w/o gradient)0

0.2

0.4

0.6

0.8

1

Nor

mal

ized

Cos

t of

One

Goo

d Pa

cket

50.9% moreenergy efficientthan max-power

40.0% moreenergy efficientthan max-power

Page 22: Robust Topology Control for Indoor Wireless Sensor Networks

22

Handling High Contention Distribution of PRR

0 0.050.10.150.20.250.30.350.40.450.50.550.60.650.70.750.80.850.90.95 10

0.10.20.30.40.50.60.70.80.9

1

Max-Power PCBLART ART (w/o gradient)

Packet Reception Rate

CDF(

Pack

et R

ecep

tion

Rate

)

Page 23: Robust Topology Control for Indoor Wireless Sensor Networks

23

Conclusions

Our empirical study shows important new negative results: RSSI and LQI are not always robust indicators of link quality

indoors Profiling links even for several hours is insufficient for identifying

good links Inherent assumptions of existing protocols!

ART is a new topology control algorithm which is robust in complex indoor environments

ART achieves better energy efficiency than max-power without bootstrapping or link starvation