Download Design and Implementation of a High

High-Fidelity Building Energy
Monitoring Network
Xiaofan Jiang and David Culler
In collaboration with Stephen Dawson-Haggerty, Prabal Dutta, Minh Van Ly, Jay Taneja
LoCal Retreat 2009
Computer Science Department
University of California - Berkeley
My PG&E Statement
2

Current level of
visibility
 Delayed
 Aggregated
over time
 Aggregated over
space
 Inaccessible

Want
 Real-time
 Per-appliance
[Stern92], [Raaii83]
Aggregate is Not Enough
3
What caused the spike at 7:00AM?
What’s the effect of turning off A?
What percent is wasted by idle PCs at night?
What percent is plug-load
What’s the effect of server load on energy?
This would be nice…
4
Architecture
5

ACme application



ACme network



Standard networking tools
Python driver + DB + web
IPv6 wireless mesh
Transparent connectivity
between nodes and
applications
ACme node





Plug-through
Small form factor
High fidelity energy metering
Control
Simple API
ACme Node
6
Two Designs
7
ACme-A
ACme-B
ACme-A vs ACme-B
8
ACme-A




Resistor + direct
rectification + energy
metering chip
Real, reactive,
apparent power
(power factor)
Idle power 1W
Low CPU utilization
ACme-B




Hall-Effect + stepdown transformer +
software
Apparent power
Idle power 0.1W
Medium CPU
utilization
A tradeoff between fidelity and efficiency
ACme Node API
9
Node API function
Purpose
read() -> (energy, power)
Read current measurements
report(ip_addr, rate) -> Null
Begin sending data
switch(state) -> Null
Control the SSR

ASCII shell component running on UDP port provides
direct access to individual ACme node:
 Adjust
sampling parameter
 Debug network connection
 Over-the-air reprogramming

Separate binary UDP port for data
 Periodic
report to ip_addr at frequency rate
ACme Network
10

backhaul links
edge routers
Acme nodes
internet
data repository
app 1
app 2





IPv6 mesh routing
Each ACme is an IP router
Header compression using
6loWPAN/IPv6 (open
implementation -blip)
Modded
Meraki/OpenMesh as
“edge router”
Diagnostics using
ping6/tracert6
ACme send per-minute
digest / no in-network
aggregation
Network Performance
11




49 nodes in 5 floors
Single edge router
6 month to-date
802.11 interference
(on channel 19)
ACme Application
12

N-tier web application
 ACme
is just like any
data feed
 Python daemon
listening on UDP port
and feed to MySQL
database
 Web application
queries DB and
visualize
UDP Packets
6loWPAN
Apache
ACme Driver
MySQL DB
Python Daemon
Visualization
http://acme.cs.berkeley.edu/
13
Building Energy Monitoring
14
1.
2.
Understanding the load tree
Disaggregation


3.
Measurements
Estimations
Re-aggregation



Functional
Spatial
Individual
Understanding the Load Tree
15
Deployment
16


Edge router obtaining
IPv6 address
Ad-hoc deployment


Un-planned
Online “registration”
using ID and KEY
Meta data collection
 Security



Online for 6 month and
counting
10 million rows
Deployment
17
Raw Data
18
Additivity using Time Correlated Data
19
Multi-Resolution
20
Appliance Signature
21
Functional Re-aggregation
22
Correlate with Meta-data
23
Spatial Re-aggregation
24
Individual Re-aggregation
25
Improvements in Energy Usage
26
Reducing Desktop Idle Power
27
Discussion and Conclusion
28
Discussion






Measurement fidelity vs
coverage
Non-intrusive Load
Monitoring (NILM)
IP node level API vs
application layer gateway
Easy of deployment is key
DB design
Multiple input channel /
power strip
Conclusion


ACme is a fine-grained
AC metering network
that provides real-time
high-fidelity energy
measurement and it’s
easy to deploy
3 steps to building
energy monitoring –
understanding load
tree; disaggregation;
re-aggregation
Discussion
29



LoCal web site: http://local.cs.berkeley.edu
ACme web site: http://acme.cs.berkeley.edu
Contact: [email protected]