Towards a Smart Home
Framework
Moody Alam
Agents, Interaction & Complexity (AIC) Group,
School of Electronics and Computer Science,
University of Southampton
What is a smart home?
Visions from the past
Future Vision!
The Present
Green + Wired
The Future Home,
The Jetsons, 1962
The Smart Home,
The Jetsons, 1962
Connected devices
Home automation
Sensors
Future Homes, 1969
[A robot serving beer!]
Future Homes, 1969
[A robot serving beer!]
What is a smart home?
• No agreed
definition!
• The IBM’s vision:
1. Instrumented
2. Interconnected
3. Intelligent
Why is the Smart Home important?
home comfort


Renewable energy
lights
home automation




home care



electricity
gas
water, etc.
Zero Carbon Homes
Energy management
chronicle diseases
presence
home hospitalization
5.38 Million Smart Homes by 2015 [Berg Insight]
Smart home is an active research area
• Academia
• All top 10 Engi. & Tech Universities [Times higher Education].
• Caltech, MIT, Princeton, Cali-Berkeley, Southampton*
• Industry
• Governments – US, UK, Aus, Canada, China, EU,
• Hundreds of companies- Microsoft, IBM, British Gas..
• Numerous sub-domains: home automation, energy
conservation, elderly living.
• We are interested in those sub-domains which
require developing a software model of smart
home.
Typical workflow in such domains
Modify/
Conclude
Hypothesis
Form a
Hypothesis
Analyse
Results
Build a
Model
Simulate
/Optimise
Typical workflow in such domains
Modify/
Conclude
Hypothesis
Battery
reduces
cost
True /
False
Compare
costs
Matlab /
Java code
Minimise
cost given
battery
Form a
Hypothesis
Analyse
Results
Build a
Model
Simulate
/Optimise
Typical workflow in such domains
Modify/
Conclude
Hypothesis
Battery
reduces
cost
True /
False
Compare
costs
Matlab /
Java code
Minimise
cost given
battery
Form a
Hypothesis
Analyse
Results
Build a
Model
Simulate
/Optimise
What is the problem?
Problem:
These three phases (modelling, simulation and analysis) take up the most time.
Solution:
We propose our Smart Home Framework to speed up these phases.
We are not the only smart people to have realised this problem!
• Industry has the proprietary software toolkits.
–
–
–
–
•
Cost and Licenses!
Platform-dependency!
Limited interoperability between platforms.
Focused on the company’s business.
Academia has very few open-source toolkits:
– Focused on narrow research issues
– Models are not general and thus not extendable in other related domain
Why is SH Framework a good idea?
• Open-source and free of cost!
• SHF has three core components each focused
on a single phase:
– Model Classes Model building phase
– Optimiser  Optimisation / Simulation phase
– Visualiser  Analyse Results
Smart Home Framework
Building a Model
Optimisation
Analysis
Smart Home Framework
Building a Model
Optimisation
Analysis
SHF: Model Classes: Overview
• We take a bottom-up modelling approach:
– Smart Home is made of different components (e.g. appliances and
storage).
– We provide general models for these components.
– These components can be integrated to create a smart home.
• This general model of a smart home:
– Has an understanding of its components and how are they related
– Can be extended to specific models
• These smart homes can be connected together to form a
smart community.
Modelling a smart home
Generation
Appliances
• A collection of:
– Appliances
– Generators
– Storage
– Electric Vehicle
• Relationships:
EVs
Storage
Grid
– Between all above
– Grid (Tariff)
– Other Smart homes
Grid
Modelling a smart home
• SH Framework contains
– Interfaces
– Abstract classes
– And Implementation of abstract classes
• To model
–
–
–
–
Generation
Storage
Appliances
Appliances’ Use
SHF: Modelling Generation & Storage
• Modelling Generation Sources
– Microgeneration (e.g. Solar Panels / Wind Turbine)
– Grid
• Modelling Storage Facilities
– Electric Batteries
– EV Batteries
SHF: Appliances and their usage
• Support to model appliances (i.e. Loads):
– SHF already have implementation of common home
devices (e.g. TV, Oven)
– Abstract classes to include new appliances
• Modelling appliances’ usage (i.e. Load Events):
–
–
–
–
–
Deferrable and Non-Deferrable
Interruptible and Non-Interruptable
Critical
Baseload
Combination of above (e.g. a deferrable interruptible
critical load event)
SHF: Modelling implicit understanding
of devices and their relationships
• Consumption + Battery Charging = Generation
• Battery has a limited number of charging
cycles.
• EV battery is available only certain times a day.
Modelling is easy: Code Snippets
• Adding renewable generation and/or grid is easy:
– agent.addEnergySource(new SolarPanel(1.5kW));
– agent.addEnergySource(new WindTurbine(2kW));
– agent.addEnergySource(new Grid(tariff));
• Creating appliances and Load Events:
– TV tv = new TV(0.3kW)
– agent.addEvent(new onDeferrableLoadEvent(tv,start,end);
• Adding storage
– agent.addStorage(new Battery( 2kWh, 0.5kW, 10%loss));
Smart Home Framework
Building a Model
Optimisation
Analysis
SHF: Optimisation in a smart home
• Optimisation depends on the structure or
formation of your smart home model:
– Generally speaking, you may be solving a convex or
non-convex problem to answer your research
question.
– Your choice of optimiser will depend on the structure
of your problem.
• SHF architecture allows you to plug-in any
optimiser of your choice!
SHF comes with a default optimiser
• IBM’s CPLEX Optimiser is available as the default plug-in optimiser:
– Free of cost to academia.
– Supports LP, MIP and Convex optimisation
– Catch: License needed for commercial use.
• So if your optimisation problem falls under LP, IP, MIP or certain
convex subclasses, then you can use the default optimiser!
• This optimiser is sufficient for the common optimisation problems.
For advanced and complex optimisation problems (e.g. non-convex)
you can just plug-in a general solver of your choice.
SHF and IBM CPLEX
• An optimisation problem can be expressed as a:
– Model (variables, and constraints.)
– Objective function
• SHF already have a smart home CPLEX model (Java
code).
• Commonly used objective functions are already
implemented, e.g.
– Maximise Preference, Minimise Cost/Carbon
• If your objective function is not already
implemented, you can just write a new objective
function and use the existing CPLEX home model!
Smart Home Framework
Building a Model
Optimisation
Analysis
SHF: Analysing results
• SHF comes with a visualiser.
• Code is there to visualise common devices /
events in a smart home.
– Plots for generation, consumption, battery usage
• Visualiser is extendible, easy to include new
plots etc.
• Results available in XML, CSV formats
Smart Home Demo: Modelling,
Optimisation and Analysis
Beyond a single smart home:
Smart communities
• The framework has all the building blocks to
create a community of connected homes.
• A small community be readily modelled to test
different communal aspects:
– Energy Exchange
– Electric vehicle charging
– Battery Usage minimisation
– Coalition formation for group buying
Smart Community Demo:
Reducing the battery usage through
energy exchange
Questions??
Thank you!