Scribd
Upload
56 views23 pages

Machine Learning in IoT

This document summarizes a presentation about using machine learning and business intelligence with IoT data. It discusses how IoT data generation will increase dramatically. Machine learning can be used to analyze IoT data to find patterns, clusters, and classes. As an example, the presentation describes how a university analyzes WiFi access point usage data through machine learning and business intelligence to optimize energy usage and costs by determining when access points can be safely turned off without disrupting services. This data analysis could provide insights for facilities management, marketing, and security.

Uploaded by

Liberito Santos
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
56 views23 pages

Machine Learning in IoT

This document summarizes a presentation about using machine learning and business intelligence with IoT data. It discusses how IoT data generation will increase dramatically. Machine learning can be used to analyze IoT data to find patterns, clusters, and classes. As an example, the presentation describes how a university analyzes WiFi access point usage data through machine learning and business intelligence to optimize energy usage and costs by determining when access points can be safely turned off without disrupting services. This data analysis could provide insights for facilities management, marketing, and security.

Uploaded by

Liberito Santos
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 23

Machine Learning and Business

Intelligence in IoT
Mukhammad Andri Setiawan, PhD ([email protected])
Chief Information Officer - Universitas Islam Indonesia

Presented at Cisco Netacad Conference 2019


Universitas Muhammadiyah Malang, 17 October 2019
Hello!
Little bit of my background

• CIO, Universitas Islam Indonesia, since 2018


• Cisco Netacad Instructor, since 2004
• Cisco Instructor Trainer, ITC UGM, since
2016
• Tim Infrastruktur Indonesia Research
Education Network (IdREN), since 2019
• Main contact and administrator of eduroam
Indonesia, since 2017
• Training and Community, IDNIC (Indonesia
Network Information Center) – The National
Internet Registry of the Republic of Indonesia,
since 2018

Department of Informatics, Universitas Islam Indonesia 2


IoT as the future
3
The IoT landscape, today!

More Smart Cities IoT security IoT in Traditional


Industry

Department of Informatics, Universitas Islam Indonesia 4


IoT as Value-Added Services

Example: The ubiquitous deployment of wireless sensors is


extending digital connectivity to tasks, processes, and machine
and service operations

M. Iansiti, K.R. Lakhani, Digital ubiquity: how connections, sensors, and data are revolutionizing business, Harvard
Bus. Rev. 92 (11) (2014) 91–99.

Department of Informatics, Universitas Islam Indonesia 5


Why Do Companies use IoT?

Improving service Increasing visibility Enabling new Creating new product


operations into operations business models and service offerings.

Department of Informatics, Universitas Islam Indonesia 6


Why Machine Learning?

• 49K photos in Instagram


• 3.9M Google searches
• 4.3M YouTube watch
• 473K Twitter tweets
• 12.9M text sent
• 750K Spotify stream
• 156M emails sent
• 15K GIFs sent via Facebook

Department of Informatics, Universitas Islam Indonesia 7


How Big the Data from
IoT is?
Around 4.4 trillion GB of data will be generated by the year
2020 through the Internet of Things (IDC, 2014)

Department of Informatics, Universitas Islam Indonesia 8


• Large amount of data will be generated with
3V characteristics, volume, velocity, and
variety
Processing • Big data files stored the data
IoT data • BI tools/ML tools such as Hadoop
MapReduce or Spark is used
• Generate the analysis report

Department of Informatics, Universitas Islam Indonesia 9


Values of
Data
• Patterns
• Clusters
• Classes
• Principal Component

Department of Informatics, Universitas Islam Indonesia 10


Machine Learning
explained (very shortly)
A hypothesis typically can be of the following
forms.
• mathematical formula
• description
𝑇
• graphical illustration
 
𝑦 =𝜃 𝑥+𝑏
Learning here, is to find the best hypothesis,
out of those possibilities.
Machine learning is about algorithms for
learning.

Department of Informatics, Universitas Islam Indonesia 11


Machine Learning

Machine learning is used for making hypotheses to our real-


world problems

AND/OR

evaluating hypotheses.

Department of Informatics, Universitas Islam Indonesia 12


How do we make use of data in our
campus?
• UII has almost 800 WiFi Access Points installed across
campuses, and keep growing
• Each AP can consume 20 Watt
• A day usage of each AP = 20 * 24 = 0,48 kWh
• Roughly each AP will cost us = Rp 725 per day
• In a day, UII spent Rp 580.000,00
• In a month, UII roughly spent Rp 17.400.000,00, only for Access
Points’ electricity (excluding switches, routers, WLC, etc.)
Department of Informatics, Universitas Islam Indonesia 13
How do we make use of data in our
campus? (cont.)
• Example of Access Points usage

Department of Informatics, Universitas Islam Indonesia 14


How do we make use of data in our
campus? (cont.)
• How can we make a greener and sustainable approach for our
WiFi?
• Can we shut down some access points without disrupting the
services (at all)?
• What kind of approach that we can use to exactly pinpoint when
to shut them off and when to turn it back on? Linear regression?
Clustering?

Department of Informatics, Universitas Islam Indonesia 15


What time shall we turn off the AP? Which month?

Determining daily schedule, when to turn on when to turn off the Determine (classify) on which month shall we
Access Points turn off the access points

Department of Informatics, Universitas Islam Indonesia 16


Potential future work, cross
referencing
• To enrich the dataset, we can make use some of these data:
• Tracking human activities (walking, speaking, sitting) in surrounding
access point that create noise ambience – as measured by IoT sensors
- and then cross reference them with the status of the access point 
minimizing the service disruption

Department of Informatics, Universitas Islam Indonesia 17


Business Intelligence

• Def: turning data into actionable information


for leadership, management, organization
and decision making
• BI creates effective and timely decisions
• In our case, we analyse time series data of
WiFi usage

Department of Informatics, Universitas Islam Indonesia 18


Eliciting knowledge from WiFi data

Meneses and Moreira,


Sevstuk et. al., 2009; Prentow et. al., 2015
2012;
• there is a lack of • analysing eduroam • assumption of
understanding about SSID, however it’s smartphone holder
how non-laptop limited when an within the premise is
users exploit the access point can span 59% according to
campus (MIT) across more than one estimated
• unclear conclusions floor of departments percentage of
about the impact of • in the future, plan to smartphone
WiFi on people’s analyse human penetration in the
spatial preferences movements country

Department of Informatics, Universitas Islam Indonesia 19


Data Collection
Approach

• Users Data, to indicate who is using access point


now
• Access Point Data, to explain detail of each
access point (physical location, MAC address, AP
identifier)
• Timestamp Data, to record time association with
users
• All data are collected into a data warehouse system
for further actions (Extract Transform Load)

Department of Informatics, Universitas Islam Indonesia 20


What analytics we can offer to
management?
• Seamless attendance system
• The dashboard helps management to
intervene, to plan, and to appraise
employee

Department of Informatics, Universitas Islam Indonesia 21


What analytics we can offer to
management? (cont.)
• Calculating ROI • Crowd Measurement

Who can use it? Property and Facilities, Marketing,


Security, etc.

Department of Informatics, Universitas Islam Indonesia 22


Thank You

Q&A
[email protected]

Department of Informatics, Universitas Islam Indonesia 23

You might also like