the hitchhiker’s guide to successful residential sensing deployments

The Hitchhiker’s Guide to Successful Residential Sensing Deployments

Timothy W. Hnat, Vijay Srinivasan, Jiakang Lu, Tamim I. Sookoor, Raymond Dawson, John Stankovic, and Kamin

WhitehouseU. VirginiaSensys 11

Presenter: SY

Lesson Learned Paper

• Large scale, long-term deployment in home– 1200 sensors– Over 20 homes– Up to 1 year/home

• Experience sharing– Myths VS facts

Myths1. Sensors in homes can easily be powered using the wall sockets2. Communication in homes can be achieved with single-hop

wireless and/or power line modems3. Robust enclosures are only important for extreme outdoor

environments4. Maintenance visits are not a problem for homes5. Users can help maintain the system, and can provide

validation data through surveys or questionnaires6. Users won’t mind a few sensors around the house7. Industry has already produced a wide range of suitable

residential sensing systems

What They Learned

Diffi

culty

#Sensors

#Sen

sors

< #Out

lets

#Sen

sors

> #Out

lets

#Hom

es <

#Res

earch

ers

#Hom

es >

#Res

earch

ers

#Homes

#Day

s < ~1

mon

th

#Day

s > ~1

mon

th#Days

Outline

• Deployment• Failure Analysis• Hitchhiker’s Guide

Deployment

# Homes

Weeks

Motion

Object Use

Door Heights

Wearable Tracking

Light Switch

Power (Plugs)

Power (Circuits)

Power (Mains)

Water Mains

Custom Thermostat

Active Register

Light/Temp Humidity

A 11 1-2 25-30

12-20 - - - - - - - - - -

B 1 1 - - 12 12 - - - - - - - -

C 3 3-4 15-25 - - - - - - - - - - 12-

25

D 1 2 4 - - - - 2 - - - - - -

E 1 2 5 - - - - - 1 - - - - -

F 1 28 65 13 13 16 22 - - 1 1 1 12 86

G 1 44 54 7 31 14 22 8 37 1 1 1 12 29

H 1 39 15 7 14 - 11 4 - 1 1 - - -

I 1 32 25 10 30 - 31 3 48 1 1 - - 8

J 1 25 14 5 17 - 7 2 - 1 1 - - -

Deployments

Deployments

Deployments

Why Such Deployments• V. Srinivasan, J. Stankovic, and K. Whitehouse. Protecting your Daily

In-Home Activity Information from a Wireless Snooping Attack. In Proceedings of the 10th international conference on Ubiquitous computing, pages 202–211. ACM New York, NY, USA, 2008.

• J. Lu, T. Sookoor, V. Srinivasan, G. Gao, B. Holben, J. Stankovic, E. Field, and K. Whitehouse. The smart thermostat: using occupancy sensors to save energy in homes. ACM Sensys, 2010.

• V. Srinivasan, J. Stankovic, and K. Whitehouse. Using Height Sensors for Biometric Identification in Multi-resident Homes. In Pervasive, 2010.

• J. Lu, D. Birru, and K. Whitehouse. Using simple light sensors to achieve smart daylight harvesting. In The ACM Workshop on Embedded Sensing Systems for Energy-Efficiency in Building, 2010.

System Architecture

Sub-systems

Gateway

Failure Analysis

• Analyze the sensor down time• Method– Define a longest acceptable report interval τ• For each sensor • About 5x sample period

– Classify root cause• based on the set of simultaneous sensor failures

Classification• Wireless link loss

– single wireless sensor, less than 4 τ• Battery dead

– single battery-powered sensor, longer than 4 τ• Plug disconnected

– single plug powered sensor, longer than 4 τ• Sub-system down

– all sensors in a single sensor sub-system• Internet Down

– all sensors reliant on a broadband link• Power outage

– all sensors reliant on AC power• Gateway down

– simultaneous down time of all sensors

SummaryJanuary 1, 2011 to August 1, 2011

Fault Analysis

15

Reinstall

Hard drive failure

Sub-system failure Plug disconnections

Outline

• Deployment• Failure Analysis• Hitchhiker’s Guide

Homes are Not a Power Panacea

• Myth– Sensors in homes can easily be powered using the

wall sockets

• Fact– Wall sockets provide neither abundant nor reliable

power, especially when deploying hundreds of nodes

18

Wall Sockets– 30-40 outlets per house– Long wires– 2.3x more down time than

batteries• More maintenance calls

In-line Power & Indoor Solar

• In-line power– Wired directly into wiring– Problem• Expensive• Reboot -- rebooting the house

• Indoor Solar– Upper bound: 0.1mW– Compare to outdoor: 102mW

Homes Have Poor Connectivity

• Myth– Communication in homes can be achieved with

single-hop wireless and/or power line modems

• Fact– Homes are small but can still be challenging RF

environments, particularly for large-scale, dense, and heterogeneous networks

21

Wireless or Power-line

• Wireless connectivity

• Power line communications– Wires– 180bit/s – 5 min polling rate

Homes are Hazardous Environments

• Myth– Robust enclosures are only important for extreme

outdoor environments

• Fact– Homes are safe environments for humans but can

be hazardous for sensors, particularly when hundreds of sensors are deployed over long time durations

23

Homes are Hazardous Environments

• Children• Mobile objects• Roomba• Guests and

cleaning services

Verify Failures

• Mean time to failure (MTTF)– Deploy 500 sensors• One year MTTF means more than one sensor fail per

day

• Automated script to check

Network down Services down

Last entry time Minimum frequency

Calibration Time incorrect

Load high Space low

Timestamps incorrect

Report Failure

• By email– Too many

• Project all critical alerts on wall

Homes are Remote Environments

• Myth– Maintenance visits are not a problem for homes

• Fact– Investigators have very limited access to

deployments not in their own homes

27

Homes are Remote Environments• Minimize installation time– Scout– Lab assembly and

configuration– Checklists

• Test three time

Expect Limited User Participation

• Myth– Users can help maintain the system, and can provide

validation data through surveys or questionnaires

• Fact– A user’s ability to monitor and report activities in the

home is limited by the need to do those activities, particularly in long-duration deployments

29

Expect Limited User Participation• Button Tracking• Wearable Tracking• Self-reporting• Surveys

29

Use redundant sensing and multiple ground truth techniques

Aesthetics Matter in Homes

• Myth– Users won’t mind a few sensors around the house

• Fact– Aesthetics constrain deployments, especially at

large scale and over long time durations

31

Aesthetics Matter in Homes• Disappear into the

woodwork • Leave no trace • No LEDS at night• Noise

Simplify the Architecture

• Myth– Industry has already produced a wide range of

suitable residential sensing systems• Fact– Many COTS devices were not designed for large

scale deployments, and integration of many COTS platforms increases the possible modes of system failure

33

Simplify the Architecture

Summary

• Wall socket is neither abundant nor reliable• Communication in home still challenge• Homes are hazardous environments• Homes are remote environments• User has limited participation• Aesthetics matter• COTS are a double-edged sword

Conclusion

• Some we might already expected– But some information still useful

• Useful tips– Verify failure• Automated script• Project on wall

– Check three time– COTS are a double-edged sword