Shipping Container Monitoring
Vipul Gupta, Oracle Labs
vipul.x.gupta@oracle.com
Vail Computer Elements Workshop, IEEE Computer Society, 2012
The Internet of Things

Romkey & Hackett’s Internet Toaster, Interop ’99
http://www.savetz.com/yic/YIC11FI_6.html

2
The Internet of Things

“
’’
10 fool-proof predictions for the Internet in 2020
3. The Internet will be a network of things,
not computers.
... By 2020, it's expected that the number of
Internet-connected sensors will be orders of
magnitude larger than the number of users.

2
Smart Phones

?

• Hard to imagine using phones that
only make calls!
• Mobile phones continuously
integrating new functionality: PDA,
cameras (still/video), navigator ...
• Many built-in sensors: microphone,
camera, accelerometer, light,
proximity, GPS, compass ...
• Growing interest in connecting other
sensors/actuators:
• Android Open Accessory Dev Kit
• Redpark breakout pack for iOS

3
4.6 Billion Mobile Phone Subscribers
By 2015, Mobile Phones will Generate 6.3 Exabytes Data/month*
*Source: Cisco Systems

4
Consumer Electronics becoming smart objects

Source: Randy Krum’s blog, http://www.coolinfographics.com/blog/2010/2/15/my-digital-life-20-a-consumer-gadget-map.html

5
Utility Metering Evolution

Predict and respond
intelligently to the dynamic
behavior of all power
suppliers and consumers

Utility

HAN

Devices connect to central
control panel, through which
utility can schedule or control
devices individually

Home

Smart Grid

Smart Metering: Monthly/Daily/15-min
collection via fixed or cell network.
Two-way communication, utility can
ping meter, send “time-of-use” pricing
information or stop service remotely

Home Area
Network

One-way meter reads,
typically monthly via net,
hand-held or drive-by

AMI
"If people do not have any idea how much
energy they are using, how can you expect
them to change their behavior?"

Mechanical Meters

AMR

Jonathan Stearn, Energywatch
The Long Tail ...
Quantified Self, M2M etc.

7
Motivation
• Better efficiency -- resource management (lower cost of heating/cooling)
• Improved security and peace of mind -- accurate, responsive tracking of
valuable assets
• Timely access to information -- smart phones, traffic/environmental sensors
• Proactive health management -- both people and things (maintenance/repair)
• Comfort and entertainment -- smart environments (buildings, homes)

Empowering people with data
to make better decisions
8
Big Data

Source Proprietary and Confidential
© 2011 Oracle Corporation – : Economist Magazine

Special Report
9
The Need for Analytics
From Data to Actionable Information

What’s the best that can happen?

Optimization

What will happen next?

Competitive Advantage

Predictive Modeling
Forecasting/Extrapolation

What if these trends continue?
Why is this happening?

Statistical Analysis

What actions are needed?

Alerts
Query/drill down

Where exactly is the problem?

Ad hoc reports

How many, how often, where?
What happened?

Standard Reports

Degree of Intelligence
Source: Competing on Analytics, by T. Davenport & J. Harris

10
SPOT Device
• Basic device has three layers
• Battery
• Processor Board with Radio
• Sensor Board (application specific)
• Processor Board alone acts as a
base-station
• User programs the device entirely in
Java using standard tools, e.g.
NetBeans
• Java on bare-metal (no OS)

11
SPOT Capabilities/Technologies
Embedded Development Platform
Extremely flexible hardware and software package
Use standard Java IDE

Mobile
Built in Lithium Ion battery charged through USB

Easy to program - Java top to bottom
Java Device Drivers

Aware and Active
Able to sense and affect surroundings

Connected - Wireless Communication
Mesh Networking
Over the Air Programming

Secure
Built-in high grade ECC public key cryptography,
user-friendly key management

Built to Inspire!
(search for “spaughts” on
Flickr/YouTube)

12
Sun SPOT Developer’s kit
• Two full SPOTs with sensor boards and batteries
• One base-station SPOT

s
POT
S
,000 -wide
20
ld
ver
O
wor
se
in u

• Software
• Squawk VM
• Java SDK
• Netbeans
• USB cable, mounting clips
• Both hardware and software is open source
• www.sunspotworld.com (forums, SDK download, kit purchase)
• spots.dev.java.net
13
YouTube Videos

• > 200 videos
• Mike’s Flying Bike viewed >175,000 x
• Keyword SPAUGHTS

14
Shipping Containers
• Shippers want to improve logistics, cut expenses, eliminate shrinkage
• US cargo losses reported annually: > $15 Billion,
Total cost > $45 Billion
• 24M containers make ~200M trips/yr
• Average cargo value estimates (TEU):
Pharmaceuticals
Hi-tech (electronics)
Finished shirts
T-shirts
Baled cotton

~$ 12M
~$ 10M
~$115K
~$ 45K
~$ 15K

• Insurance cost: ~1% of cargo value

15
Smart Container Solution Architecture

ail

Oracle Monitoring
Service Portal

Em

Satellite Network

er
Al
ts

Logistics Personnel

CMD-Mounted
Container

Shipper's
Command & Control

16
Container Monitoring Device
• Based on the Sun SPOT platform
• Multiple sensors:
• Environmental: temperature, humidity, light
• Intrusion: light, PIR, door coil, UWB radar
• Location: GPS
• Dual-mode long-range wireless communication
• Quad-band GPRS
• Satellite communication
• Short-range low-power mesh networking:
• 802.15.4
• Secure communication
• Long-lasting battery
17
18
Networking Challenges
• Network bandwidth is limited, costly
• 340 byte/packet, $1.40/KB !!! (satellite communication)
• precludes sending raw data

• Devices are duty-cycled to preserve energy
• can not push data
• re-establishing network can take tens of seconds

• Device may not have a long-term network address
• mobility (real or perceived)
• NAT/firewall

Requires a gateway-based network architecture*
*Gupta et al., “A Network Architecture for the Web of Things”, Second Int’l Web of Things Workshop , Pervasive 2011

19
Sensor Fusion Example
A

B

C

Light
Infra-red
Proximity
PIR (motion)
Door sensor

UWB Radar

3-axis
Accelerometer

{

{
20
Device v/s Back-end Alerting
• Intrusion notifications sent with higher priority than periodic reports
• Both include location. Out-of-order processing at portal causes
unexpected alerts behavior
• Device-side alerting can be smarter, e.g. adaptive sampling

Waypoint

21
Sensor Recalibration
• Sensor readings vary across devices, impacted by environmental
variations:

A metal detector based
on pulse-induction, the
same technology used
in our door sensor.

22
Story of the GPS Module

9.5mA
12mA

13.1mA

3.3mA

•
•
•
•

GPS low-power mode, device off but retains almanac/ephemeris data
Allows valid “fix” in as little as 6 sec after power-on
Early “fix” quality is poor, can be improved in subsequent readings
How long to stay on?
23
Story of the GPS Module

9.5mA
12mA

13.1mA

3.3mA

•
•
•
•

GPS low-power mode, device off but retains almanac/ephemeris data
Allows valid “fix” in as little as 6 sec after power-on
Early “fix” quality is poor, can be improved in subsequent readings
How long to stay on?
23
Other Challenges

24
Cloud-based Monitoring Portal
• Stores and analyzes sensor data and alerts received from all CMDs
• Dashboard and timeline views show current data and history
• Allows users to commission and decommission trips
• Lets users define geofences of interest
• Generates notifications for registered events

25
Summary
• The Internet of Things represents the future of data and is a key area
of interest at Oracle
• A flexible sensor platform emphasizing ease-of-use and rapid
prototyping
• End-to-end integrated solutions (hardware, software and services)
for specific markets

26
Questions?
27

Real-time, Sensor-based Monitoring of Shipping Containers

  • 1.
    Shipping Container Monitoring VipulGupta, Oracle Labs vipul.x.gupta@oracle.com Vail Computer Elements Workshop, IEEE Computer Society, 2012
  • 2.
    The Internet ofThings Romkey & Hackett’s Internet Toaster, Interop ’99 http://www.savetz.com/yic/YIC11FI_6.html 2
  • 3.
    The Internet ofThings “ ’’ 10 fool-proof predictions for the Internet in 2020 3. The Internet will be a network of things, not computers. ... By 2020, it's expected that the number of Internet-connected sensors will be orders of magnitude larger than the number of users. 2
  • 4.
    Smart Phones ? • Hardto imagine using phones that only make calls! • Mobile phones continuously integrating new functionality: PDA, cameras (still/video), navigator ... • Many built-in sensors: microphone, camera, accelerometer, light, proximity, GPS, compass ... • Growing interest in connecting other sensors/actuators: • Android Open Accessory Dev Kit • Redpark breakout pack for iOS 3
  • 5.
    4.6 Billion MobilePhone Subscribers By 2015, Mobile Phones will Generate 6.3 Exabytes Data/month* *Source: Cisco Systems 4
  • 6.
    Consumer Electronics becomingsmart objects Source: Randy Krum’s blog, http://www.coolinfographics.com/blog/2010/2/15/my-digital-life-20-a-consumer-gadget-map.html 5
  • 7.
    Utility Metering Evolution Predictand respond intelligently to the dynamic behavior of all power suppliers and consumers Utility HAN Devices connect to central control panel, through which utility can schedule or control devices individually Home Smart Grid Smart Metering: Monthly/Daily/15-min collection via fixed or cell network. Two-way communication, utility can ping meter, send “time-of-use” pricing information or stop service remotely Home Area Network One-way meter reads, typically monthly via net, hand-held or drive-by AMI "If people do not have any idea how much energy they are using, how can you expect them to change their behavior?" Mechanical Meters AMR Jonathan Stearn, Energywatch
  • 8.
    The Long Tail... Quantified Self, M2M etc. 7
  • 9.
    Motivation • Better efficiency-- resource management (lower cost of heating/cooling) • Improved security and peace of mind -- accurate, responsive tracking of valuable assets • Timely access to information -- smart phones, traffic/environmental sensors • Proactive health management -- both people and things (maintenance/repair) • Comfort and entertainment -- smart environments (buildings, homes) Empowering people with data to make better decisions 8
  • 10.
    Big Data Source Proprietaryand Confidential © 2011 Oracle Corporation – : Economist Magazine Special Report 9
  • 11.
    The Need forAnalytics From Data to Actionable Information What’s the best that can happen? Optimization What will happen next? Competitive Advantage Predictive Modeling Forecasting/Extrapolation What if these trends continue? Why is this happening? Statistical Analysis What actions are needed? Alerts Query/drill down Where exactly is the problem? Ad hoc reports How many, how often, where? What happened? Standard Reports Degree of Intelligence Source: Competing on Analytics, by T. Davenport & J. Harris 10
  • 12.
    SPOT Device • Basicdevice has three layers • Battery • Processor Board with Radio • Sensor Board (application specific) • Processor Board alone acts as a base-station • User programs the device entirely in Java using standard tools, e.g. NetBeans • Java on bare-metal (no OS) 11
  • 13.
    SPOT Capabilities/Technologies Embedded DevelopmentPlatform Extremely flexible hardware and software package Use standard Java IDE Mobile Built in Lithium Ion battery charged through USB Easy to program - Java top to bottom Java Device Drivers Aware and Active Able to sense and affect surroundings Connected - Wireless Communication Mesh Networking Over the Air Programming Secure Built-in high grade ECC public key cryptography, user-friendly key management Built to Inspire! (search for “spaughts” on Flickr/YouTube) 12
  • 14.
    Sun SPOT Developer’skit • Two full SPOTs with sensor boards and batteries • One base-station SPOT s POT S ,000 -wide 20 ld ver O wor se in u • Software • Squawk VM • Java SDK • Netbeans • USB cable, mounting clips • Both hardware and software is open source • www.sunspotworld.com (forums, SDK download, kit purchase) • spots.dev.java.net 13
  • 15.
    YouTube Videos • >200 videos • Mike’s Flying Bike viewed >175,000 x • Keyword SPAUGHTS 14
  • 16.
    Shipping Containers • Shipperswant to improve logistics, cut expenses, eliminate shrinkage • US cargo losses reported annually: > $15 Billion, Total cost > $45 Billion • 24M containers make ~200M trips/yr • Average cargo value estimates (TEU): Pharmaceuticals Hi-tech (electronics) Finished shirts T-shirts Baled cotton ~$ 12M ~$ 10M ~$115K ~$ 45K ~$ 15K • Insurance cost: ~1% of cargo value 15
  • 17.
    Smart Container SolutionArchitecture ail Oracle Monitoring Service Portal Em Satellite Network er Al ts Logistics Personnel CMD-Mounted Container Shipper's Command & Control 16
  • 18.
    Container Monitoring Device •Based on the Sun SPOT platform • Multiple sensors: • Environmental: temperature, humidity, light • Intrusion: light, PIR, door coil, UWB radar • Location: GPS • Dual-mode long-range wireless communication • Quad-band GPRS • Satellite communication • Short-range low-power mesh networking: • 802.15.4 • Secure communication • Long-lasting battery 17
  • 19.
  • 20.
    Networking Challenges • Networkbandwidth is limited, costly • 340 byte/packet, $1.40/KB !!! (satellite communication) • precludes sending raw data • Devices are duty-cycled to preserve energy • can not push data • re-establishing network can take tens of seconds • Device may not have a long-term network address • mobility (real or perceived) • NAT/firewall Requires a gateway-based network architecture* *Gupta et al., “A Network Architecture for the Web of Things”, Second Int’l Web of Things Workshop , Pervasive 2011 19
  • 21.
    Sensor Fusion Example A B C Light Infra-red Proximity PIR(motion) Door sensor UWB Radar 3-axis Accelerometer { { 20
  • 22.
    Device v/s Back-endAlerting • Intrusion notifications sent with higher priority than periodic reports • Both include location. Out-of-order processing at portal causes unexpected alerts behavior • Device-side alerting can be smarter, e.g. adaptive sampling Waypoint 21
  • 23.
    Sensor Recalibration • Sensorreadings vary across devices, impacted by environmental variations: A metal detector based on pulse-induction, the same technology used in our door sensor. 22
  • 24.
    Story of theGPS Module 9.5mA 12mA 13.1mA 3.3mA • • • • GPS low-power mode, device off but retains almanac/ephemeris data Allows valid “fix” in as little as 6 sec after power-on Early “fix” quality is poor, can be improved in subsequent readings How long to stay on? 23
  • 25.
    Story of theGPS Module 9.5mA 12mA 13.1mA 3.3mA • • • • GPS low-power mode, device off but retains almanac/ephemeris data Allows valid “fix” in as little as 6 sec after power-on Early “fix” quality is poor, can be improved in subsequent readings How long to stay on? 23
  • 26.
  • 27.
    Cloud-based Monitoring Portal •Stores and analyzes sensor data and alerts received from all CMDs • Dashboard and timeline views show current data and history • Allows users to commission and decommission trips • Lets users define geofences of interest • Generates notifications for registered events 25
  • 28.
    Summary • The Internetof Things represents the future of data and is a key area of interest at Oracle • A flexible sensor platform emphasizing ease-of-use and rapid prototyping • End-to-end integrated solutions (hardware, software and services) for specific markets 26
  • 29.