Exploiting Home Automation Protocols for Load Monitoring in Smart - - PowerPoint PPT Presentation

Department of Computer Science

Exploiting Home Automation Protocols for Load Monitoring in Smart Buildings

David Irwin, Anthony Wu†, Sean Barker, Aditya Mishra, Prashant Shenoy, Jeannie Albrecht‡

University of Massachusetts Amherst Amherst College† Williams College‡

University of Massachusetts Amherst - Department of Computer Science

Why Smart Buildings?

• ~70% of grid power usage
• Smart buildings for

grid efficiency

• Economic benefits
• Environmental benefits

2

University of Massachusetts Amherst - Department of Computer Science

Demand-Side Energy Management

• Managing energy usage
• Shifting loads
• Reducing loads
• DSEM components:
• Continuous energy monitoring
• Load control

3

Peak Usage Off-Peak Usage Shiftable Load

• n
• ff

University of Massachusetts Amherst - Department of Computer Science

Energy Monitoring Systems

• Primary goals
• Inexpensive
• Non-invasive
• Examples
• ViridiScope [Ubicomp 09]
• ACme [SenSys 09]
• Single-point sensing
• ElectriSense [Ubicomp 10]
• Flick of a Switch [Ubicomp 07]
• Many others
• Monitoring ≠ Control

4

University of Massachusetts Amherst - Department of Computer Science

Exerting Control on Electrical Loads

• Home automation (HA) products
• Inexpensive and mature
• X10 (1975)
• Insteon (2001)
• Z-Wave (2005)
• Already deployed in smart grid trials

5

www.insteonsmartgrid.com

University of Massachusetts Amherst - Department of Computer Science

Combining Monitoring and Control

• Load control requires adding HA-like

hardware to devices

• Augment HA with monitoring
• Challenges:

(1) Very low bandwidth and primitive HA protocols (2) Coarse-grained events rather than fine-grained data streams (3) Mapping events to power data

6

• n / off

events power data

University of Massachusetts Amherst - Department of Computer Science

Our System: AutoMeter

• AutoMeter: our HA system for

building-wide monitoring

• Low-cost, off-the-shelf components
• Wall switches
• on/off/dim event notifications
• Power meters
• queries for outlet-level data
• Prototype deployment in a home

7

University of Massachusetts Amherst - Department of Computer Science

AutoMeter Architecture

8

Panel

AutoMeter Controller

Building/Circuit Power Switch Events

Load Disaggregation

Plug Power

University of Massachusetts Amherst - Department of Computer Science

System Components

• Using the Insteon HA protocol
• Why Insteon?
• Low cost (~$40 per device)
• More reliable than X10
• Non-proprietary
• Not solely reliant on wireless
• Complications
• Reverse engineering meter protocol
• No notifications from meters
• Usable bandwidth <180 bps

9

level changed? ...

University of Massachusetts Amherst - Department of Computer Science

Insteon Protocol Overview

• Increasing message reliability:
• Message propagation, ACKs, retransmissions
• Bandwidth limits:
• Theoretical: 2880 bps, practical: <180 bps

10

PLM Switch 1 Plug 2 Plug 1 Plug 3 Switch 2 hops = 3 h

• p

s = 3 hops = 2 hops = 3 hops = 2 h

• p

s = 1 hops = 1 hops = 2 hops = 2

Controller USB

PowerLine

"Query Plug 3"

University of Massachusetts Amherst - Department of Computer Science

Current AutoMeter Deployment

• 3 bedroom, 2 bath house, 34 wall switches
• 20 Insteon SwitchLinc relays
• 10 Insteon SwitchLinc dimmers
• 30 Insteon iMeter Solos
• TED 5000 for aggregate readings
• GuruPlug control server connected to

Insteon PowerLine Modem (PLM)

• Entire equipment budget: $3025
• 96.7% of total TED energy use accounted

for in a two-week period

11

University of Massachusetts Amherst - Department of Computer Science

Issues Encountered

• 1. Low bandwidth and message losses
• Trading off query rate and reliability
• 2. Learning switch power usage
• Proactive and reactive strategies
• 3. Tagging aggregate power variations
• Remapping power changes back to events

12

University of Massachusetts Amherst - Department of Computer Science

Problem 1: Low Data Rates

• <180 bps over power line
• No collision avoidance
• Serial meter queries
• Asynchronous switch events
• Approach: insert delay

between subsequent queries

• Delays for single meter

multiplied by # of meters!

13

‘query meter’ ‘switch off’

X

University of Massachusetts Amherst - Department of Computer Science

Meter Query Losses

• Approximate query duration: 1.0333 sec
• Reliability vs. global query rate
• Much lower reliability with lower interarrival times

14

20 40 60 80 100 1 2 3 4 5 6 7 8 9 10

% Queries Received Interarrival Time (sec)

Home Deployment Isolation Model (no retransmissions)

1.0333s

University of Massachusetts Amherst - Department of Computer Science

Wall Switch Event Losses

• Cannot control when event messages occur
• Increase interarrival time to reduce collisions
• Round-robin queries every 10s (5 min / device)
• <5% switch loss probability

15

10 20 30 40 50 60 70 80 90 100 1 2 3 4 5 6 7 8 9 10

% Events Lost Interarrival Time (sec)

Home Deployment Model (no retransmissions)

University of Massachusetts Amherst - Department of Computer Science

Smart Polling

• Idea: How much energy could we miss between

queries to a device?

• Cap amount of unaccounted energy
• Per-device query rate
• power usage and typical duration

16

vs.

• n or off?
• n or off?

slow queries fast queries

University of Massachusetts Amherst - Department of Computer Science

Problem 2: Learning Switch Power

• Switches only report on/off/dim
• Goal: learn switch power
• Use aggregate TED data
• Simple proactive approach
• Programmatically disable all loads
• Turn device on, record delta
• Repeat for each device
• 93% accurate, but requires cycling

17

‘power: 100W’ ‘ s w i t c h

• n

’ ‘switch usage: 100W’

University of Massachusetts Amherst - Department of Computer Science

Reactive Approach: Learning on-the-fly

• Learning power values on-the-fly
• Problems encountered
• Delayed data points
• Simultaneous events
• Bad readings
• Record deltas around events
• ‘Bin’ them based on delta size
• Avg most common bin (e.g., 55-65W

deltas) as energy value

• Intuition: over many events,

bins will reveal true value

18

‘last power: 432W, new power: 493W’ ‘ s w i t c h

• n

’ ‘delta: 61W, record in 55-65W bin’

University of Massachusetts Amherst - Department of Computer Science

Reactive Approach: Binning

• Wide range of energy deltas around events
• Bins usually identify true delta
• But...need enough data points
• And highly correlated events are bad

19

5 10 15 20 25 30 35 40 15- 25W 25- 35W 35- 45W 45- 55W 55- 65W 65- 75W 75- 85W 85- 95W 95- 105W 105- 115W 115- 125W 125- 135W 135- 145W 145- 155W 155- 165W 165- 175W 175- 185W

Number Events

Watt Bins

guestbath:overheadlight guestbath:sinklight masterbath:sinklight

University of Massachusetts Amherst - Department of Computer Science

Problem 3: Tagging Power Variations

• Objective: tag aggregate data with specific events
• Problem: errors in aggregate data
• Reading errors (2% TED error)
• Timing errors (missed readings, 5-minute frequencies)
• Rampup errors (TED readings change gradually)
• Many events missed even with high deltas

20

20 40 60 80 100

100 200 300 400 500 600 700 800 900

Percentage Threshold (W)

Individual Events:Building Events

Figure 6. We use AutoMeter’s switch, plug, and circuit

University of Massachusetts Amherst - Department of Computer Science

Conclusions

• HA protocols show promise for providing

monitoring capabilities

• Smart polling
• Accurate building data
• Other types of data – circuits, topologies, ...
• Issues encountered
• Switch power: learn proactively or reactively over time
• Outlet power: cope with limitations with intelligent polling
• Time and cost not a significant barrier for

complete HA instrumentation

21

Department of Computer Science

Questions?