Security & Privacy in Smart Grido Detect data sources, read data streams & build

usage pattern.

o Send “constructed” data & determine internal

state by analyzing output.

o Impersonate the system & get all users’ data

stream; Send bogus responses back to the

consumers.

o Connect to the system & add/change queries

on application’s behalf.

Historical Storage Data Models

APIs Semantic Processing/

Machine Learning

In House Apps (e.g. Demand Response)

Third Party Apps

Admin

ORG-X

ORG-Y Distributed Complex Stream

Processing

Machine Learning for Predicting Energy Usage Cloud Computing for Scalable Info Management

SOFT

WA

RE A

RC

HIT

EC

TU

RE

Power Utility

Residential Customer

Industrial Customer

Commercial Customer

Historic Data

Utility Data Consumer

Data

Demand Response

Engine Consumer Policy

Utility Policy

Current Load Current Supply Current Usage

Event Stream

o Predict peak demand on Utility

o Predict usage for new customers

o Cluster customers into sub-groups

o Provide users with individual usage

data & analysis

o Data mining for fault detection

o Energy Monitoring tools

o Means for sharing & comparing

usage data with other parties

o Track consumption change with

change in appliances/equipment

o Provide appliance-level

consumption details

o Explain unusual usage activity

o Learn from historical data to

predict energy use patterns

Y. Simmhan, S. Aman, B. Cao, M. Giakkoupis, A. Kumbhare, Q. Zhou, K. Gomadam & V. K. Prasanna

Semantic Event Processing & Information Integration

o Large-scale, high-frequency data/metadata collections

• Power consumption in residential, commercial area

• Power production data, Weather data

o Data storage challenge: GB’s of data per day

accumalating from 1000’s of sources

o Computation challenge • Real-time analysis of streaming data at scale • Historic data pattern matching for usage prediction

o Cloud storage for historic data

o Tailor Cloud VMs to various roles

• Pattern matching on streams • “Hub-Spoke” VMs for response

propagation to consumer AMIs • VMs to inform Utilities of

power demand predictions o Research

• Mapping DR apps to compute • Optimize VM usage for cost

o Integrated Smart Grid Information

Model (iSGIM) – modular & extensible

domain ontologies.

o Provide common semantics for Smart

Grid data and concepts.

o Support intelligent applications using

heterogeneous information sources

• Smart meters, Household appliances,

weather forecast service

o Production data accessible only to ORG-

X’s processing system & In-house apps.

o Consumer devices & usage pattern not

to be disclosed.

o Adapt to on-demand change in privacy

policies.

o Control access to specific objects –

streams, attributes, operators – based

on admin/consumer policies.

o Control access to APIs based on the

contracts with the applications.

o Semantic Complex Event Processing for Smart Grid

o Abstract complex events & processing operations as

queries and rules on top of iSGIM

o Provide platform-independent and scalable event

processing

o Identify meaningful events within the information cloud

o Analyze their impact & take subsequent realtime actions

Power Utility

Power Distribution Supply,

Regeneration

Weather Station

Billing & Usage

Policy

System Load

SOFTWARE INFRASTRUCTURE

Data & Metadata Management Appliance Database

Historic Usage & Supply Data

Data Processing

Data Analysis Data

Prediction Pattern Mining

Semantic Event Integration

Current Load, Use, Supply

Event Stream Domain

Ontology

Comm. & Policy Management Demand

Response Consumer & Utility

Policy

Privacy Policy Security

Static Utility & Consumer Data

Residential Building

Policy

Solar Gen

Home Area N/W Smart Meter

Load Control Device

Industry

Commercial Building

Policy

Consumer Portal

Metering

Communication

Data Acquisition

INTE

RFA

CE

Policy Feedback