More innovations arise from borrowing and combining than from simple invention (Fortune, 2004)
Research Scientist, Engineering Systems Division, School of Engineering, MIT and Executive Director, MIT Forum for Supply Chain Innovation
Shoumen Datta
1
I think there is a world market for maybe five computers.
Thomas Watson
Chairman, IBM
1943
There is no reason anyone would want a computer in their home.
Kenneth Olson
Founder, DEC
1970
Prediction is very difficult, especially about the future.
Niels Bohr
1920
2
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Economic History from Norman Poire
Conceptual Advances (add to the Wealth of Nations but “Adam Smith was wrong!”)
Adoption
1853
Textile
1800
1771
1853
1825
1913
Railway
1913
1886
1969
Auto
1969
1939
2025
Computer
Agents
Grid, SL
2005
1977
2061
Nanotech
Hydrogen
Fusion
2025
~1997
Technology
Introduced
1959
AI
Industrial Revolution
Atoms
3
2081
PROCESS
Knowledge Economy
Physical World Model
Bits
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
DECISION
Graphics: Forrester
Barriers to Adaptability and Death by “Clockspeed”
PLM
SCM
CRM
XSCM
Adaptable Business Network
“Clockspeed” by Charles Fine, MIT
Adaptable Business Network popularized by Bob Betts, Founder, Mainstreet Applications and co-author of “Adapt or Die”
4
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: Forrester
Data vs Noise
~10 terabytes
per second
2005
Estimate excludes real-time data
~6 terabytes
per second
2004
~3 terabytes
per second
5
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
2003
Source: Nicole DeHoratius, University of Chicago and Ananth Raman, Harvard Business School
Inventory Record Inaccuracy
% of SKUs (n=369,592)
40%
35%
35%
30%
65% of SKUs
are inaccurate
25%
22%
20%
15%
10%
10%
9%
6%
4%
5%
3%
3%
2%
2%
1%
1%
7
8
9
10
1%
0%
0
1
2
3
4
5
6
11-50
Absolute Error (units)
6
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
51-100
1 01-200
2 01-400
Convergence
Convergence
Convergence
Push Æ Pull Æ Adaptive Æ Predictive
31%
26%
Markdowns
(% of sales)
Too many wrong
products !!
21%
16%
11%
6%
1970
1980
1990
1995
“A third of customers entering a
store leave without buying. They
can’t find what they came to buy.”
Fewer right products
Source: Nicole DeHoratius, University of Chicago and Ananth Raman, Harvard Business School
7
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
http://obelia.jde.aca.mmu.ac.uk/multivar/pca_graf.htm
Data vs Information: Systems introduce Artifacts and Inaccuracies
Retain 87.5% of the information
Retain 62.5% of the information
Size = Length + Breadth
8
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Right-Time Data in Agents-integrated Adaptable Demand Network ?
Open Grid Services Architecture
SEMANTIC PORTAL
Inventory Agent
Replenishment
needs
Replenishment
needs determined
from RFID Tag info
“Intelligent Signal”
Personal Care
Consumer Tissue
Health Care
Paper
Cotton
Medical
Raw material
SUPPLIERS
Information
Agent
Future
shipping
needs
Immediate
EPC (RFID), UWB (UID)
Replenishment
needs
Customer
Info Center
Transport
real time
TLB Agent
RFID data
Store
Manufacturer
HQ
Orders
Loads
Store
Status
Confirmation
Plant
Plant
Store
DC
DC
Store
DC
DC
MANUFACTURER
Store
DC
Retailer’s DC
Cross-Docking Agent
Store
Store
Store
Store
Store
(This illustration is a modified composite from various sources including P&G, Forrester, Kimberly-Clark)
9
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Consumption
Inventory
Internet 0 Ubiquitous Infrastructure: Real-Time Data ON/OFF Control
Bits, Atoms, Decisions
Internet 0
Internet 1
Internet 2
MEMS / NEMS
Intel Motes, Crossbow
D2B / RFID / UWB
Object Oriented Hardware
Service (Value) Supply Chain
A
G
E
N
T
S
IPv6
e
ic
rv
Se
Right-Time
Analytics
dERP
GRID
y
pl
up
)S
ue
al
(V
From
an office
in Shinzen,
Real Time Data
China, you log
Streaming Data, Continuous Queries
on a SDR reader in
a warehouse in USA,
Semantic Grid
to check if your products
Web Portal
SECURITY
arrived on-time. They did.
You also get to know that
your distributor in Santiago,
Chile and retailer in Espoo,
Finland
also checked
the
delivery
Shoumen
Datta, MIT
Forum
for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
10
status, moments before you logged on. Data Interrogators as Ubiquitous Internet Appliances
C
in
ha
CONVERGENCE
Real-Time Adaptive Model
OPTIMIZE
OBJECT
DECISION
AGENTS
DATA
INFO
PROCESS
11
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
CONFLUENCE
Near Real-Time Predictive Model
OPTIMIZE
OBJECT
DECISION
AGENTS
DATA
Forecast “ Predict ”
Demand “ Pull ”
INFO
PROCESS
12
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Enabling Real-time Data at the Right-time ?
Radio Frequency IDentification
13
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
14
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
15
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
16
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
17
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Modified from: Han Pang Huang, National Taiwan University
What is ‘new’ about RFID ? Evolution of RFID
1940
RFID born out of
Radar effort (WWII)
1948
Harry Stockman
invents RFID.
Publishes paper,
“Communication
by Means of
Reflected Power”
1950
RFID crawls out
1950
D.B. Harris
patents RFID.
“Radio
transmission
systems with
modulatable
passive
responder”
1952
F.L. Vernon
“Application of
the microwave
homodyne”
1959
Identification of
Friend or Foe
(IFF) longrange
transponder
system reaches
breadboard
demonstration
stage.
1960
Theory of RFID, field
trials planned
1963-1964
R.F. Harrington
advances theory
with “Field
measurements
using active
scatterers” and
“Theory of
loaded
scatterers”
1966
Commercialization
of EAS, 1-bit
Electronic Article
Surveillance
1970
Early adopters
implement RFID
1973
Raytheon's "Raytag"
1977
RCA develops
"Electronic
identification
system"
1975
Los Alamos
National Lab (LANL)
releases RFID
research to public
sector, publishes
“Short-range radiotelemetery
for electronic
identification using
modulated
backscatter”
1976-1977
LANL RFID spinoffs Indentronix
and Amtech
1975-1978
Raytheon, Fairchild
& RCA develop RFID
18
1980
Commercial RFID
endeavors sprout
1990
Many RFID
standards emerge
1982
1991
Mikron founded; TI creates TIRIS
bought by Philips to develop and
market RFID
1987
First RFID road 1992-1995
toll collection
Multi-protocol
implemented
traffic control and
in Norway
toll collection
implemented in
Texas, Oklahoma,
and Georgia (USA)
1998
David Brock and
Sanjay Sarma of
MIT publishes an
idea: ‘Internet of
Things’
2000
RFID hype, peaks
2003
UPC and EAN
forced by US
retailers to
promote EPC
2005
Wal-Mart and
US DoD fuels
the hype curve
by demanding
suppliers use
passive RFID
and EPC.
1999
Auto ID Center
created at MIT.
Retailers drive to
standardize EPC
Vast number of RFID
companies and
‘short-sight’ enters
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
the market.
Partial Source: Shrouds of Time – The History of RFID
19
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
CHAR.
Point-of-Scanning
is potentially the
weakest link in the chain.
- UPC/EAN
- Ilv 2 of 5
3
0 0
1 2 3 4 5
6 7 8 9 0
6
- Code 128
- Code 39
PATTERN
CHAR.
1
2
3
4
5
6
7
8
9
0
A
B
C
D
E
F
M
N
O
P
Q
R
S
T
U
V
W
X
Y
Z
G
H
I
J
K
L
SPACE
PATTERN
*
$
/
+
%
- 2D Stacked
20
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
21
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
6
14141
12345
SSCC
FROM
TO Customer
DC1478
Good Time Supplier
1155 Francisco,
Battery St
5241 San Antonio Dr NE
San
Albuquerque NM 87109
94111
CARRIER
SHIP TO POST
Best Freight
B/L: 853930
(420) 87109
PO: 345-896779-0
DEPT: 092
SSCC
(00) 000521775138957172
EAN.UCC
Company
Extension Prefix
Digit
Application
Identifier
22
Check
Serial
Reference Digit
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
9
Co
m
p
SS
VA
LU
E
y
/P
ro
cu
re
m
Value Chain Management
Tr
a
M
an
u
fa
ct
ur
in
G
W oo
Tr
ar d s
ac
eh R
ki
e
o
g
ng
us ce
e ip
M t/
an P
a g i ck
em i n
en g
t
G
Re
ca
ll
D
oo el
ds ive
R e ry
ce
ip
t
Supplier Relationship Management
Distributor Retailer Management
23
CR
EA
In
Pu ven
rc
t
ha ory
si
ng
P
Co rim
ns ary
um
er
Gi
ft
Se
rv
ic
e,
W
ar
ra
Sp
nt
y
ar
e
Pa
rts
US
A
In
ve
nt
or
l
is
e
CE
M
er
lo
g
PR
O
y
Pe
M
fg
S
e-
ES
na
ng
N
So
n
et
a
ec
t ro
n
So
l
So
n
y
SN
C
/T
ra
ns
m
BU
SI
Aware Goods? Object Identification ?
ck
Co
nt
Cu
ro
l
TI
en
t
O
N
Se
r
st
om
er
/P
ro
du
c
tH
is
to
vi
ce
ry
Focus:
Customer Relationship Management
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Su
pp
or
t
“Likewise, in the past few decades most of
the companies that have created truly
extraordinary amounts of wealth have done
so by inventing great processes, not great
products (technology). Dell, Toyota and
Wal*Mart, for example, have risen to the
top of their respective industries by coming
up with amazingly efficient ways of getting
quite ordinary products into the hands of
consumers more cheaply than their rivals.”
24
Source:
The Economist, April 24th, 2004
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Why use RFID ? Emperor’s New Clothes?
• Radio Frequency Identification
• Electronic Product Code (EPC)
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
RF waves transfer data (object to reader)
Re-writable secure data
Identify individual items
Line of sight not required
Stable in variety of conditions
Read through most non-metals
RFID transponders 5 cents ? ($0.25 - $150)
RFID readers: $2000 to $10 (SDR?)
Infrastructure: Profit over Physics?
RFID Interface (Real-time data) to ERP (?)
Can current RDBMS handle data flow?
Auto ID standard Global EPC at UCC.EAN
Limited spatial capacity of 1 kbpsm2
Item or Pallet
SKU
Reader
Server
ERP
Internet
RFID Tag (Active UWB)
25
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
System
What is RFID ?
•
•
•
•
•
•
•
•
CPU and Memory
Antenna
Frequencies
Active (battery)
Passive
Read only (WORM) tags
Tags
Re-writable tags
Low sophistication = Low Cost.
•
One or more RF tags
•
Two or more antennas
•
One or more interrogators
•
One or more host computers
•
Appropriate software
•
Tag memory: factory or field
programmed, partitionable
(option: permanent lock)
Bytes left unlocked can be
rewritten >100,000 times
Critical information database
•
•
26
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Types of RFID
27
ƒ
Active RFID - Longer range
ƒ Continuously powered tag
ƒ Low-level RF to the tag
ƒ High-level RF back to the reader (transmits radio signal)
ƒ Longer read ranges (>100 metres)
ƒ Multi-KB data capacity
ƒ
Passive RFID - Shorter range
ƒ Tags reflect radio signal from reader
ƒ Tag receives/stores energy to respond
ƒ Needs stronger RF signal from reader
ƒ Low RF strength from tag
ƒ Shorter range (~5 cm to ~5 metres)
ƒ May require link to database
ƒ
Semi-Passive RFID
ƒ Similar to passive
ƒ Internal power (battery) for tag circuitry
ƒ Range may be extended
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
RFID Frequencies
EPC
13.56 MHz
125 KHz
AM Radio
0.5
EPC
SW Radio
1.7
433 MHz
Garage
30
40
TV 2-6
54
FM Radio
88
TV 7-13
108 174
216
TV 14-69
470
2.45 GHz
860-930 MHz
Cordless Ph
806
902
GPS
1.2
Cell Ph
1.6 1.8
2.1
BluTh, b/g
802.11a
2.4
5.0
MHz GHz
ULTRAWIDEBAND
28
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
IC
Satel TV
5.8 10.7
12.5
RFID Frequencies
Frequency
29
Advantage
Disadvantages
125KHz and 135KHz
Free from regulation
Relatively inexpensive
Very large antenna
Slow with short range
13.56MHz
Water/Tissue penetration
Small, thinner antenna
Government Regulations
Hard to get around metal
303.8MHz, 418MHz,
433MHz, 868MHz and
915MHz
Longer range
Higher data rate
Poor water/tissue penetration
UHF spectrum crowded in US
Regulatory issues outside US
2.45GHz and 5.8GHz
Small tag/antenna size
Good range
Very high data rate
More susceptible to noise
Shared with other technologies
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
RFID Frequencies
Frequency
125-150 kHz
13.56 MHz
433 MHz
860-960 MHz
2450 MHz
30
Regulation
Unregulated
ISM band, differing
power levels and
duty cycle
Non-specific Short
Range Devices
(SRD), Location
Systems
ISM band (Increasing
use in other regions,
differing power levels
and duty cycle
ISM band (differing
power levels and
duty cycle)
Range
Data Speed
≈ 10 cm
Low
Comments
Animal identification
and factory data
collection systems
<1m
Low to
moderate
Popular frequency for
Smart Cards
1 – 100 m
Moderate
US DoD (Active)
2–5m
Moderate to
high
1–2m
High
EAN.UCC GTAG,
MH10.8.4 (RTI),
AIAG B-11 (Tires),
EPC (18000-6’)
IEEE 802.11b,
Bluetooth, CT,
AIAG B-11
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Frequencies to be Studied
Elements of the Study (each frequency, each country)
for RFID, RFDC (RFID data collection) & RFID LAN (RLAN)
Identify
Identify
Identify
Identify
Identify
Identify
Identify
Identify
Identify
Identify
Australia
Japan
China
Korea (South)
France
Russian Federation
Germany
Singapore
Hong Kong
United Kingdom
India
United States
Austria
Finland
Peru
Spain
Belgium
Israel
Philippines
Sweden
Brazil
Italy
Poland
Switzerland
Canada
Malaysia
Portugal
Taiwan
Colombia
Mexico
Romania
Thailand
Czech Republic
Netherlands
Saudi Arabia
Turkey
Denmark
New Zealand
Slovak Republic
Ukraine
Egypt
Norway
South Africa
Yugoslavia
First Focus
50 - 140 kHz
433.92 MHz
7.4 - 8.8 MHz
862 - 928 MHz
13.56 MHz
2450 MHz
315 MHz
5850 MHz
the primary user
the availability
the maximum possible output power
the maximum antenna gain
the max effective isotropic radiated power (ERIP)
the required duty cycle
the bandwidth
the channel spacing
the licensing requirements
restrictions and future plans
Argentina
Croatia
Indonesia
Qatar
Bahrain
Czech Republic
Kuwait
Slovenia
Bulgaria
Cyprus
Malta
United Arab Emirates
Chile
Hungary
Oman
Venezuela
31
Second Focus
Third Focus
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
RFID Hardware Types
•
•
•
•
•
32
Acousto-Magnetic
Inductive
Modulated Backscatter
Long Range Active
Real Time Location
–
–
–
–
–
Very Short Range
Very Short Range
Short to Medium Range/Directional
Long Range
Long Range
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Acousto-Magnetic
EAS Label
Pulses
Resonating
Signal
Transmitter
•
•
•
•
33
Receiver
Theft Prevention and Access Control
Easily deactivated
Very low frequency (50-60KHz)
Inexpensive.
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Inductive
Tag
Reader/Writer
•
•
•
•
•
34
Movement
of Tag
CPU with ferrite/air core
Short range (inches)
Low cost
<150KHz and 13.56MHz
Passive
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Inductive Passive 13.56 MHz and <135 KHz
Near Field
ASK, PSK
35
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Inductive – How it works
Traditional
Transformer
I
Power
Source
I
Primary
Coil
Secondary
Coil
Load
EM Field
Reader
RFID Tag
I
To
“System”
System”
I
< 60 cm
EM Field
36
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
EM Field UHF RFID
37
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Modulated Backscatter
• 915MHz and 2.4GHz
• Range up to 90 feet
Eye
Light
Reflective
Object
Light
To
System
RFID
Tag
Radio
RF Field
38
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Microwave RFID
Passive: ASK, PSK
Active: FSK
0.5-1.0 m (far field) 4W
Active: 15-20 meters
39
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Long-Range Active Tags
• UHF or 2.4GHz
• Range up to 600 feet
• Requires Battery
40
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Real Time Location System
• Time Differential or Signal Strength
• Range up to 1000 feet
• Requires Battery
t4
t2
Tag
t1
Reader
t3
41
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
RFID Operation
Antenna
RF Module
Tag
Reader
Host Computer
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
RFID Tag : Wireless Information System & DB
Interrogation Unit
Micro
Computer
Tx/Rx
Antenna
Tag
CPU
I/O
RAM ROM
Radio Tx/Rx
Pwr Supply
Computer Network
CPU
I/O
RAM
ROM
Radio Tx/Rx
Pwr Supply
43
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
RFID System Components
Reader
Antenna
Asset/Tag
Asset
Firmware
Tag
Insert
TCP/IP
Host
Customer’s
MIS
44
Application
Software
~
Power
API
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
RFID System Architecture
APPLICATION
INTERROGATOR
RF TAG
Tag Physical Memory
Decoder
Application Program Interface
AIR
Encoder
INTERFACE
DEVICE
COMMANDS
APPLICATION
COMMANDS
APPLICATION
RESPONSES
Command /
Response
Unit
Tag
Driver
and
Mapping
Rules
COMMANDS
RESPONSES
DEVICE
RESPONSES
Logical Memory
DATA PROTOCOL
PROCESSOR
PHYSICAL
INTERROGATOR
The Logical Memory Map in the Tag Physical Memory is given by the tag architecture and mapping rules
in the Tag Driver. All the information in the Logical Memory is represented in the Logical Memory Map.
45
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Logical
Memory
Map
RFID Technology vs Business Process
Where
UHF (400-1000 MHz)
Microwave
Approved EIRP*
Radiated Power
from
Reader
Distance
EU
0.5 Watt
0.7 metres
US & Canada
4.0 Watt
2.0 metres
US site license
30 Watt
5.5 metres
Data: AIM
Radiated Power ≈ Energy Field » Read Range
* EIRP - effective isotropic radiated power
46
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Field Strength
Transmitted Signal Inversely Proportional to Exponent of Distance
13.56 MHz Æ 1/d 6
UHF Æ 1/d 2, 1/d 3, 1/d 4 (orientation dependency)
47
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
RFID Operation - PASSIVE
Sequence of Communication
•
•
•
•
•
•
•
•
•
•
48
Host manages reader(s) and issues commands
Reader and tag communicate via RF signal
Carrier signal generated by reader (request from host application)
Carrier signal transmitted through antennas
Carrier signal reaches tag(s)
Tag receives and modifies carrier signal
Tag ‘sends back’ modulated signal (passive backscatter)
Antennas receive modulated signal and transmits to reader
Reader decodes data
Results returned to host application
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
RFID Operation - ACTIVE
Sequence of Communication
•
•
•
•
•
•
•
•
•
•
49
Host manages reader(s) and issues commands
Reader and tag communicate via RF signal
Inquiry (upon request from the host application)
‘Wake-up’ signal transmitted by interrogator to all tags within
communication range
Tags enter ‘ready state’ awaiting command from interrogator
Interrogator initiates communications; listens for response from tags
Tag communicates with interrogator based on command received
Antennas receive modulated signal and transmit to reader
Reader decodes data
Results returned to host application
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
STEPS
toward
PRIVACY
PROTECTION
50
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Portal Applications
Bill of Lading
Material Tracking
51
Number items at forklift speeds
8’ X 10’ doorways
Electronic receipt & dispatch
Wrong destination alert
Electronic marking
Pallet/container item tracking
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Conveyor / Assembly Line
Up to 450 fpm
60+ items per container
Inexpensive tunnels
Longer tunnel more items
Electronic receipt
Sorting
Electronic marking
52
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Handheld Applications
Wireless / Batch
Inventory Management
Where is it? What is it?
What is inside the box?
Where is it going? Where has it been?
Should it be here?
What have I assembled or disassembled?
How many do I have? Do I have enough?
Material Handling
Inspecting / Maintaining
53
Has this been repaired?
Is this under warranty?
Has this been inspected?
Is this complete?
What is the asset’s status?
ASN Verification
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Pet/Animal Tracking
ACCESS SECURITY
54
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
DoD In-Transit Visibility (ITV)
•
•
•
Over 350 Nodes World-Wide
Tag is Interrogated as it Passes a Node
TRANSCOM Kits for Contingency Operation.
433MHz Interrogation
ITV Nodes
ITV Server
433MHz Response with ID
and/or Data (TAV Format)
55
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: US DoD
Source: US DoD
ITV Process
Information
uploaded to ITV
server
Access to ITV
Data via Web
ITV Server
Tag Location is
Reported to Central
ITV Server
Nodes
pick-up tag id
Container contains
bar coded equipment
In Transit
Shipment Arrives
Tag Manifest is
Read
Manifest is written to
tag and uploaded to
server (TAV Format)
56
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: US DoD
Bosnian ITV Capability
All ALOC Shipments From New
Cumberland & All Containers Shipped
From USAREUR Are Tagged
GERMANY
- JTAV/LAD
Frankfurt International
AIR
TRUCK
RAIL
Ramstein
NSE
- RF Interrogators Installed
Kaiserslautern Prague
- RF Interrogators to be
Installed
AUSTRIA
xxxx
CZECH
REPUBLIC
HUNGARY
xxx
NSE
ASG
Interrogators also installed at:
• Miesau
• Germersheim
• ERF
• Baumholder
• Bad Kreuznach
• Baumholder Railhead
• Weillerbach Railhead
• Coleman Barracks Railhead
Data Passed via Phone Line to LOGSA
Within 15 Minutes of Reading Tag
57
Kaspovar
Tazar Airfield
CROATIA
ITALY
Tuzla
Tuzla APOD
123rd FSB)
QUALCOMM Provides Visibility of
Truck Convoys & Rail Movements
Data Passed to Paris Hub via Satellite
Dispatch Stations Access Paris Hub
via Modem/Phone Line
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Asset Tracking: Healthcare Example (Hospital)
Sign Post (132KHz Transmit)
Range to 12 feet
Sign Post ID
Site
Server
Tag ID and Last
Sign Post ID
(Up to 250 feet)
Tag Readers
(433MHz Receive)
Tagged Asset
Hand Held Device & Computers
With Security Client
58
Hospital LAN
802.11b WLAN
Access Points
FIPS 140-2
Firewall
Browser access to
location data from
anywhere in hospital
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Asset Tracking: Ground Services Example (Airports)
59
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Tracking
HUMANS
OBJECTS
60
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
(MIT) Auto ID Center - EPC Objective 1998
Short, simple, extensible code to uniquely identify
products and reference networked information.
61
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
RFID Evolves
Proprietary RFID
MIT Auto ID
Database
DATABASE
Object Name Server
Reader
Reader
EPC (unique ID)
Store Data on the Tag
62
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
RFID Evolution
63
Past/Present
Commercial RFID
Present/Future
Auto-ID
Data
On the tag
On the network
Applications
Closed loop
SCM-ERP systemwide
Cost
Expensive
Inexpensive
Technology
Proprietary
Open Standards
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Standards-based Auto-ID
DSS Software
ONS / PML
Reader
Antenna
PML – Physical
Markup Language
Antenna
ONS – Object
Name Service
EPC – Electronic
Product Code
UCC.EAN Alphanumeric string
64
EPC
EPC
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
EPC Objectives
Unique
EPC should be a unique numbering scheme
Reference EPC should be used primarily as an
information reference
Simple
EPC should be a simple as possible and
minimize information content
Internet
EPC should be integrally coupled to Internet
systems and protocols
Standards EPC should accommodate were possible
legacy standards, systems and codes
65
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Electronic Product Code
Naming Scheme for Physical Objects
6,000,000
560,000,000
1,000,000,000
20,000,000,000
13,000,000,000,000,000
Cars per year
Computers
Televisions
Razor blades
Grains of rice
Sub-components
Spare Parts
Assemblies
Containers
66
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
128 bit EPC structure can incorporate IPv6 numbering scheme
67
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
EPC Vision ? RFID Reality Check ?
• More than 12 separate RFID EPC Trials in US
• UK Home Office RFID Trial (2000-2003)
• Germany, Japan, Singapore, UK RFID Trials
• Gillette buys 500 million EPC RFID Alien tags
• Wal*Mart suppliers to use EPC RFID tags by 2005
• EPC Global to be managed by UCC.EAN
• US DoD urges suppliers to use RFID tags by 2005
• How much of this is marketing hype ?
68
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
P
no longer behind bars…Beyond Barcode
Electronic Product Code (EPC) 96 bits
268 million companies can each categorize 16 million different products
and each product category may contain over 68 billion individual items !!
Header ePC Manager Object Class Serial Number
01.0203D2A.916E8B.0719BAE03C
Header: 8 bits = 256
ePC Mgr: 28 bits = 268, 435,456
Object Class: 24 bits = 16,777,216
Serial Number: 36 bits = 68,719,476,736
69
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
(98) 0614141999999
Past
HEINZ
KETCHUP
Repository
01.0203D2A.916E8B.0719BAE03C
Now
70
HEINZ
KETCHUP
Repository
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
RFID tagged Gillette razors at TESCO Store (Cambridge, UK)
Elbert Hubbard, 1856-1915
One who says it can’t be done is often interrupted by someone doing it !
Source: Colin Cobain, CIO, TESCO (www.tesco.co.uk)
71
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
EAN.UCC Keys to Data
• GIAI – Global Individual Asset Identifier
• GLN – Global Location Number
• SSCC – Serialized Shipping Container Code
• GTIN – Global Trade Item Number
• and now
• Global EPC – Global Electronic Product Code
72
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
96 bit EPC
Can uniquely number …
79,228,162,514,264,337,593,543,950,336
or about 8 x 1028 individual objects
or more than 1 million times all the grains of sand on earth!!
73
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
EPC Design
If only one number is used once …
0
1
2
.
.
.
One Big Database !!
79,228,162,514,264,337,593,543,950,334
79,228,162,514,264,337,593,543,950,335
79,228,162,514,264,337,593,543,950,336
74
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
EPC Design
Two numbers …
Still somewhat large …
X.Y
75
… 281,474,976,710,656
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
EPC Design
Three numbers
X.Y.Z
4,294,967,296 …
… about right ?
76
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
EPC Design
What if we add a version number ?
Version.X.Y.Z
77
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
EPC Design
Version.X.Y.Z
Instance
Header
Serial Number
Version
Domain
Manager Number
78
Class
Object Class
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
96 bit EPC
01.0203D2A.916E8B.0719BA
E03C
Header 8 bits
256
Domain 28 bits
268,435,456
79
Instance 36 bits
68,719,476,736
Class 24 bits
16,777,216
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
EPC can embed other standards
Universal Identifier (UI)
Domain Identifiers (DI)
80
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
96 bit EPC
Object Type
4
Partition
4
Header
8
81
(Company) Prefix
37-20
EAN.UCC
Company
Extension Prefix
Digit
Application
Identifier
Item Reference
7-24
Check
Serial
Reference Digit
Serial Number
36
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
96 bit EPC Object Type Codes
Object Type
Name
0
Item/Customer Unit
1
Inner pack
2
Case/Shipping Unit
3
Load/Pallet
4
Location
5
Other
6,…, 15
Unassigned
Object type 4 bits x Partition 4 bits = 16 types
82
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
EPC Design
Partition
83
Manager
Bits
Digits
1
37
11
2
34
3
Class
Address
Bits
Digits
128 Billion
7
2
128
10
16 Billion
10
3
1024
30
9
1 Billion
14
4
16,384
4
27
8
128 Million
17
5
131,072
5
24
7
16 Million
20
6
1 Million
6
20
6
1 Million
24
7
16 Million
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Address
64 bit EPC Design
Object
Type
4
16
Header
2
4
EPC Manager
14
16,384
Item Reference
20
1,048,576
Serial Number
24
16,777,216
Item Reference is identical
to the GTIN Item Reference
84
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: AIDC
85
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Retailer
Supply (?) Chain
Local DC
Retailer
Regional DC
Retailer
Local DC
Manufacturer
Retailer
Retailer
Local DC
Retailer
Regional DC
Local DC
Retailer
Retailer
86
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Supply Chain: Material vs Information Flow
Data
Information Flow
Data
•Manual
Data Entry
•Slow
•Error prone
•Friction
Manual Check
Bar Codes
•No value add
Manual Check
Bar Codes
Material Flow
87
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Illustration: Mark Dinning, DELL Corporation
Supply Chain Optimization: Real-Time Data
Information Flow
•Automated
RFID
Reader
•High integrity
•Fast
RFID
Reader
•Frictionless
ePC
Material Flow
88
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: AIDC
MIT Auto ID Center “Town Test”
Procter
&
Gamble
CASE
PALLET
MFG
FACTORY
MFG
DISTRIBUTION
CENTER
CASE
PALLET
MFG
FACTORY
MFG DC
Chicago IL
LIQUID ALL
&
DOVE SOAP
CASE
PALLET
MFG
FACTORY
MFG DC
Krafts
Foods
CHEESE
SLICES &
MACARONI &
CHEESE
CASE
MFG
FACTORY
MFG DC
Johnson
&
Johnson
FEMININE
HYGENE
CASE
PALLET
MFG
FACTORY
MFG DC
SKUs
CASE
PALLET
SKUs
MFG
FACTORY
MFG
DC
Gillette
Unilever
MFG
89
BOUNTY
Mach III
PALLET
SAM'S CLUB
TULSA
SAM'S CLUB
KANSAS CITY
DC
WAL-MART
DC
BENTONVILLE
WAL-MART
TULSA
RETAILER
DC
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
RETAIL
STORE
MIT Auto ID Center Test : Phase One
CHEP attaches tags
for PALLET
P&G pallets
Procter
&
Gamble
Pallet back to pool
FACTORY
BOUNTY
CASE
PALLET
Cape
Girardeau Mo
MFG
SAM'S CLUB
KANSAS CITY
DC
VMI Note:
DC
SAM'S CLUB
TULSA
P&G ships high vol
products direct to WAL*MART
MFG
SKUs
CASE
PALLET
SKUs
MFG
FACTORY
MFG
DC
1
1
0
1
1
0
RETAILER
DC
0
Estimated 150
pallets per month
High volume
products in
pallets
Products with high
volume at Sam's
At manufacturer:
Tag pallets
Wire 3 doors
Wire 1 PML server
Application software
RETAILER
1
Shipped directly from
factory to retailer
At retailer wire:
6 incoming doors
2 transition doors
1 DSD door
2 shelves on retail floor
1 PML server
Application software
Tag and read pallets at mfg factory and/or DC:
400 RFID TAGS
Read pallets at Sam's Club through:
30 READERS
Incoming
2 PML SERVERS
Movement : staging area to retail floor
Pallet return to CHEP or disposal
90
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: AIDC
Source: AIDC
MIT Auto ID Center Test : Product Flow
P&G MFG FACTORY
CAPE GIRADEAU , MO
J&J DC
KRAFT FOODS DC
UNILEVER DC
WAL-MART FOOD DEPOT
CLARKVILLE, AR
GILLETTE DC
WAL-MART DEPOT
BENTONVILLE, AR.
CASES
WAL-MART STORE
TULSA, OKLAHOMA
Warehouse
Retail Floor
P&G DC
SAM'S DEPOT
KANSAS CITY
pallets
SAM'S STORE
TULSA OKLAHOMA
Staging Area
UNITS
Unit shelves on
retailer's floor
Retailer's check out
91
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Retail Floor
Source: AutoID Center
ONS on the net read via PML to translate EPC
Dumb chips with EPC, Smart net hosts ONS
92
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: EPC Global
EPC Network – The Building Blocks
Tags
The data carrier. Identity number is programmed into the
memory.
EPC
The code carried by the carrier; the globally unique
pointer for making inquiries about the item associated
with EPC.
Antenna
Connected to the chip. Could be traditional wire or coil or
could be printed antennas using conductive inks.
Reader
The data capture device; portable or fixed (installed),
connected to a Savant or network.
Savant
ONS
Servers which act as local repositories for EPCs and
associated information, and which support sophisticated,
flexible middleware for serving PML queries.
Object Name Service; the distributed resource that “knows”
where information about EPCs is held (just like DNS).
Structure to allow structured querying and reporting
EPC
Information concerning EPCs.
Service
93
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: AIDC
4
94
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: AIDC
Slow Moving Barriers ?
reduced functionality
(networking & software)
greater functionality
increased chip size
reduced chip size
(handling small chips)
95
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: AIDC
RFID: Low Cost ?
20
handling cost
die size/cost in cents
Silicon: 4 US Cents/mm2
15
10
5
time
96
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Smart Objects: Road to Ubiquitous Tagging?
(IC) Chip
Antenna
Assembly
Packaging
Total Cost
When ?
20 cents
5 cents
5 cents
20 cents
50 cents
2001
25 cents
2003 ?
2 cents
1 cent
1 cent
1 cent
5 cents
2010 ??
0.5 cents
0.1cent
0.1cent
0.3cent
1 cent
2020 ???
Plastic
97
Printed
Printed on Objects
< 0.1 cents ?
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: AIDC
Layers
Class IV tags:
Active tags with
broad-band peer-to-peer communication
Class III tags:
semi-passive RFID tags
Class II tags:
passive tags with additional
functionality
Class 0/Class I:
read-only passive tags
98
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Downward failsafe
Upward compatibility
Class V tags
Readers. Can power other Class I, II and III tags;
Communicate with Classes IV and V.
Source: AIDC
RFID Transponder : Tag
99
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: AIDC
Cheap Chip
100μm
100
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: AIDC
Cheap Chip Manufacturing
101
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: AIDC
ONS
• Redirection Service
– acts as telephone book in reverse
– principle of Domain Name Service (DNS)
102
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
MDM
103
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: AIDC
PML
• Language for describing physical objects
– classification and categorization
104
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
P
Industry Specific
105
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: AIDC
EPC OS: AIDC “Savant” (just another middleware)
“National”
Savant
“Regional”
Savant
Savant
“Store”
Savant
Savant
data readers sensors machines
106
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: AIDC
EPC OS Software
Soap Query
Soap Response
Soap Interface
Query Processor 1
Class ID
Class
Server
Query Processor 2
Task Manager
Class
Update
Schedules
Retrieve
Schedules
Query Processor n
Data Storage
107
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
EPC Network Architecture - Inside the Firewall
•Local copy of frequently
-used ONS data
•Registration for static and dynamic ONS
ONS
•Collaboration on asset tracking
Enterprise
(cache)
Application(s)
•Track and Trace Serial Items
•Referencing Business Transactions
PML Service
•Object Type Data (e.g. pallet/case/item)
(EPC Database)
Additional data
•Instance -level EPC data (e.g. expiry date)
•Fine -grained access control policy implementation
event data
Databases
(ERP..)
•Report Data
Savant
•Manage Readers
Filtered event data
(optional)
•Higher Level Filters
•Capture Events Data (tag and sensors)
Reader
•Simple Filters
Temperature,...
EPCs
•Transmit ePC data using radio frequency
•Transmit sensor data
108
Tag
Tag
Sensor
queries
updates
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
points to
provides data to
EPC Network Architecture - Outside the Firewall
Static ONS:
ONS
• converts an EPC into an
internet address to locate a
PML Service
Dynamic ONS
• provides means to locate
current and previous EPC
Custodians for the purpose of
track and trace, recall etc.
Company A
Company B
ONS (cache)
Enterprise
PML Service
(EPC Database)
Additional data
Enterprise
Applications
Business Transactions
Applications
event data
Savant
queries
updates
109
Filtered event data
(optional)
points to
provides data to
Internal
DB
(ERP)
Internal
Database
(ERP)
PML Access
Registry
•Web service interface describing
the capabilities and data
accessible through each PML
service to trading partners.
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
EPC in IMS
110
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
P
EPC in Retail
111
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
P
EPC in Theft Prevention
112
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
P
EPC in Drug Anti-Counterfeit
113
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
EPC in Healthcare Track & Trace
114
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
EPC in Waste Management
115
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
P
EPC in Patient Monitoring
116
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Telemetry : RFID + Sensors
117
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Telemetry – Shelf Life
• 76 million foodborne illness
• 1.8 million deaths worldwide
• 325,000 hospitalizations in US
• 5000 deaths in US
118
• 91 million tons of food
disposed to landfills in US
• 26% of US food supply
• 824 million ‘hungry’ per year
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Telemetry – Shelf Life
∂Q
= −k1e
∂t
⎡
Ea ⎤
⎥
⎢−
⎢⎣ Rg T ( t ) ⎥⎦
Q
n
Variables
•
•
•
•
•
•
119
Ea
k1
n
T
Q
t
Activation energy
Arrhenius constant
Order of the reaction
Temperature
Quality
Time
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Telemetry – Shelf Life
Q(t ) = Qoe
120
⎛ E ⎞ ⎞
⎛
⎜− a ⎟
⎜
⎜ Rg T ⎟ ⎟
⎠t ⎟
⎜ − k1e⎝
⎜
⎟
⎜
⎟
⎝
⎠
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Telemetry – Shelf Life
Food Quality
Name: + Activation Energy
Description: + Activation Energy
+ Arrhenius Constant
Name:
Symbol:
Ea
+
Description:
Constant
Access:
Read Name: +Arrhenius
Temperature
Symbol:
k1
ID: EPC:
010300908808BF6000000102
Description: + Temperature
Access:
Read Name: + Quality
Class: Scalar Symbol:
T
ID: EPC: 010200908238760000023877
+ Food Quality
Description:
Type: Float
Access: Read
+ Order of Reaction
Name:
Class:
Scalar
Symbol:
Q
Unit: m=2 kg=1ID:
s=-2
u=-1
EPC:
010200908238760000023877
+ Order of Reaction
Description:
Type:
Float
Access:
Write
Default:
25000.0
Class: Scalar
Symbol: n
Unit: s=-1
ID: EPC: 010200907ABC8 60000012875
Type: Float
Default: 0.002 Class: Access:
Scalar Read
Unit: k=1
ID: EPC: 01020084191000001289731
Default:Type:
286.0 Float
Class: Scalar
Unit: s=-1
Type:
Default:
100.0Int
Unit:
Default: 1
Name: + Food Quality
Description: + Food Quality based Arrhenius
Developer: + Natick Army Laboratories
ID: EPC: 010300908808BF60000000AA
Comp: + $0.25 per month
Type: Analytic
Rate: + 1 to 10,000 sec
Algorithm: +
121
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Telemetry – Shelf Life
Q
T
?
n, k1, Ea
T
PML
122
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
RFID Temperature Sensor in US DoD MRE Simulation
123
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
RFID Monitoring Perishables (MRE Simulation)
124
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
RFID Monitoring Expiration Date (MRE Simulation)
010101
001000000100 11
010101
00 00 10001
010101000
000000
10010 00
0000
101010010000 10
10
000000000100
001000
1010 100
1001
0101010001 1 01
010100
0 0 10
1111
000100100110 11
0
101
010C0la
10ss11111
A
s
s
1 essment
ISSUE
INSPECT
DISPOSE
125
DISPOSE
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
RFID : Current Issues
•
•
•
•
•
•
•
•
126
Spatial capacity of 1 kbpsm2
Continuous wavelength
Narrow dedicated spectrum
Data corruption by frequency collision
Passive transponders in manufacturing ?
Palet size vs passive tag range ?
Metal objects: spare parts ?
Universal standards ? (915MHz, 13.56MHz, 2.45GHz, 125KHz)
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
TagArray (UC Berkeley)
PASSIVE ULTRAWIDEBAND
Solution in Search of Problems ?
Old Version: Active UWB from MSSI, Robert Fontana
disruptive technology ? de facto global standard ?
127
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
13.56 MHz
125 KHz
AM Radio
0.5
SW Radio
1.7
433 MHz
Garage
30
40
TV 2-6
54
FM Radio
88
TV 7-13
108 174
216
TV 14-69
470
2.45 GHz
860-930 MHz
Cordless Ph
806
902
GPS
1.2
Cell Ph
1.6 1.8
2.1
BluTh, b/g
802.11a
2.4
5.0
MHz GHz
ULTRAWIDEBAND
128
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
IC
Satel TV
5.8 10.7
12.5
Solution in Search of Problems ?
•
•
•
•
•
•
•
•
129
Wide spectrum (>960 MHz, 3.1-10.6 GHz, 22-29 GHz)
Spatial capacity 1000 kbpsm2
Power 200 mW (802.11b ~500mW; 802.11a ~2000mW)
Data 0.1 – 1.0 gbps2 (802.11b ~0.006gbps2 or 6mbps2 )
Contender for BlueTooth replacement
600 picosecond bursts (avoids multipath interference)
UWB+GPS+RTLS : innovative combination ?
UWB + narrow-band is catalytic for passive UWB tags
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
UWB Pulse
Source: Robert Fontana
130
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Less Power Requirement
131
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: Intel & Robert Fontana
132
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
UWB RFID Technology
Continuous wave RF modulates
data signals over carrier waves.
Ultrawide band is carrier-less RF
Amplitude
Modulation (AM)
UWB encodes information as pulse of RF energy
z Timing of pulses is used to relay information
z
t
UWB Characteristics
Low power requirements (low battery drain, lower health risks)
z Low cost transmitter design (no need for separate baseband + RF stages)
z 30 foot radius coverage at 100Mbps (longer for lower data rates)
z Demonstrated ability to support very high data rates (100Mbps and beyond)
z Immunity to interference (from other devices and multipath signals)
z Inherit security at the signal level (UWB is very difficult to detect or defeat)
z Ability to acquire accurate location information (resolutions < 1 foot)
z
133
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Frequency
Modulation (FM)
WLAN Application
134
UWB
Bluetooth
802.11A
802.11B
Throughput
~100 Mb
~700 Kb
35-54 Mb
6-11 Mb
Power
40 mw
30 mw
1-1.7W
500 mw
Range
0.01-10 km
10 m+
~100m
100 m
Resolution
~1 ft
NA
NA
NA
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: Intel
UWB as LPS Æ Indoor “GPS”
z
z
z
z
Track customer traffic flow
after they pick up an item
Track customer inspection
of items even if they don’t buy
Check activity by display
type (not just by dept)
Measure wait times by cashier
Floor Activity Simulation
Security Monitoring in Restricted Areas (Airports Operations) ?
135
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
RFID Made Difficult ?
What’s needed
Pervasive Use Cases
Ultrawide band + narrow band
Passive UWB tags
Widespread Adoption
-Software Radio (SDR) Readers
-OFDM
-Frequency agnostic readers
-Reader efficiencies
ROI
ATTH (AIT to the home)
136
Comments
Transaction cost economics
Orthogonal Frequency Division
Multiplexing (OFDM)
Coasian analysis + VAR-GARCH
Powerline data transfer
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Everything that computes also communicates and routes
E Æ merge Everything that communicates also computes and routes
Everything that routes also computes and communicates
Data
DataÆ
ÆInformation
Information
Semantic
SemanticSupraNet
SupraNet
Where
Whereare
areyou?
you?
Un-Wire
Un-Wire
Data
Data(information)
(information)
WWW,
WWW,Internet
Internet
Multi-hop/nano
Multi-hop/nanoSensors
Sensors
Agents,
Agents,Semantic
SemanticTags
Tags
dERP, Wearables
802.16, Mesh, UWB
Locate, Process, Context
WiFi
WiFi802.11b,
802.11b,RFID
RFID
BlueTooth,
BlueTooth,IPv6
IPv6
GPS,
GPS,Portals,
Portals,Voice
Voice
Browsers
Browsers
2010
137
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
log (people per computer)
Emerging Computing Trend
Number Crunching
Data Storage
Productivity
Interactive
Streaming
Information
to/from
Physical World
year
138
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
log (people per computer)
Emerging Computer Class
Mainframe
Minicomputer
Workstation
PC
Laptop
PDA
???
year
139
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: Horst Simon, LBL
Roadmap for Electronic Devices (minus CAEN)
1018
Classical Age
Quantum Age
295oK
1016
77oK
1014
1012
Chip Components
1010
108
Historical Trend
1990
106
104
102
CMOS
1980
1970
101
Size (μ)
140
o
SIA Roadmap 4 K
2010
Quantum State Switch
2005
2000
1995
100
10-1
10-2
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
10-3
“Unwired” Sensor Net
Wireless Multi-hop Broadcast Mesh
INTEL
Dot 01
Sensor
INTERNET
INTERNET
S
O
U
R
C
E
141
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
“Unwired” Sensor Net
Wireless Multi-hop Broadcast Mesh
INTEL
Dot 01
Sensor
QU
ER
Y
Database
S
O
U
R
C
E
142
in-network
processing
802.15.4
ZigBee
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Application
INFORMATION
Processed Data
Y
ER
U
Q
QUERY
INTERNET
INTERNET
MIT Project Oxygen
Emerging network nodes may be billions of embedded devices
generating exabytes of data per second but is that information?
143
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Context-Aware Computing
Environmental
Context
Sensors
• Human-centric
– “Finding” applications
• Embedded
– Sensors & Actuators
– Devices
– Monitoring & Control
Location
RFID, UWB, GPS
Processing
+
communication
Processing
+
communication
Resource
information
Network
Actuators
D2B
144
Processing
+
communication
Processing
+
communication
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: Hari Balakrishnan, MIT LCS & EECS
Source: Hari Balakrishnan, MIT LCS & EECS
Context-aware Services ÆÆÆ (Service Supply Chain)
•
•
•
•
145
Zero configuration
Context-aware, speech-driven, location-based (CRICKET location system)
Resource discovery and secure info (INS Æ Intentional Naming System)
Unconstrained, adaptive mobility (routing) to capture network context (MIGRATE)
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: Hari Balakrishnan, MIT LCS & EECS
Project Oxygen: CRICKET
Beacons
(ceiling)
θ
H21
146
B
SPACE=NE43-510
ID=34
COORD=146 272 0
http://cricket.lcs.mit.edu
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
RF module (rcv)
Ultrasonic sensor
RF module (xmit)
Antenna
MOTES
Atmel processor
RS232 i/f
Listener
Beacon
MIT CRICKET PROTOTYPE
Source: Hari Balakrishnan, MIT LCS & EECS
147
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
148
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: Hari Balakrishnan, MIT LCS & EECS
Context-aware Resource Discovery : INS
camera510.lcs.mit.edu
•
•
•
Intentional Name
[service = camera]
[building = NE43
[room = 510]
Services advertise-register resources
Consumers make queries for services
System matches services and consumers
Problem: naming systems name by (network) locations
Names should refer to what (not where)
Use expressive language (XML)
Lookup
Resolver
self-configuration
149
image
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: David Culler, University of California at Berkeley and INTEL Research Lab at UC Berkeley
Deeply Embedded Networks
•
•
•
•
•
•
•
•
•
150
# nodes >> # people
sensor/actuator data
unattended, inaccessible
prolonged deployment
energy constrained
operate in aggregate
in-network processing
dynamic functions
network programmable
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Vast Networks of Tiny Devices
•
Internet built around dedicated devices carefully configured and stable
– high-power wireless subnets, 1-1 communication between named computers
HERE ……..
•
•
•
•
•
•
151
every little node is potentially a router
work together in application specific ways
collections of data defined by attributes
connectivity is highly variable
must self-organize to manage topology, routing, etc
for power savings, radios may be off most of the time
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
NanoLetters (2004) 4 1785-1788
A Conducting Polymer Nanojunction Sensor for Glucose Detection
Erica S. Forzani, Haiqian Zhang, Larry A. Nagahara, Ishamshah Amlani, Raymond Tsui and Nongjian Tao
Department of Electrical Engineering and Center for Solid State Electronics Research
Arizona State University, Tempe, Arizona, USA
and The Microelectronics and Physical Sciences Laboratory, Motorola, Tempe, Arizona, USA
http://pubs.acs.org/cgi-bin/article.cgi/nalefd/2004/4/i09/pdf/nl049080l.pdf
152
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: David Culler, INTEL Research Lab at UC Berkeley
Sensor Network @ Work
www.greatduckisland.net
Light, Temp, Humidity,
Barometer, Passive IR
153
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: David Culler, INTEL Research Lab at UC Berkeley
Open Platform
WeC 99
“Smart Rock”
Rene 00
Mica 02
Small microcontroller
- 8 kb code, 512 B data
Simple, low-power radio
- 10 kb
Dot 01
Designed for
experimentation
EEPROM storage (32 KB)
-sensor boards
128 KB code, 4 KB data
Simple sensors
-power boards
50 KB radio, 512 KB Flash
DARPA
DARPA
INTEL
DARPA
www.tinyos.net
154
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: David Culler, INTEL Research Lab at UC Berkeley and David Tennenhouse, INTEL Research
In-network Processing in Network of “Motes”
•
•
•
•
Ad hoc sensor field of nodes
Each node knows only its own location (node id)
Neighborhood discovery (learns of “neighbors” and their locations)
Local Processing (light)
Topology
TinyDB
155
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Wide-Area Broad-Coverage Services
Embedded
Networks
156
Traditional Point-to-Point Internet
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
TinyOS Users
•
•
•
•
•
•
•
•
•
•
•
•
•
•
157
US DoD
ALTARUM
BAE SYSTEMS
VIGILANZ SYSTEMS
PHILIPS
FRANCE TELECOM
INTEL
GE
GRAVITON
HONEYWELL
HP
BOSCH
SIEMENS
XEROX
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Clustering
Neural Nets
Decision Trees
State Transitioning
Smart Maintenance
Intelligent Diagnostics
In-operation Sensor-based Monitoring for Diagnostics and Predictive Maintenance
• Trains pass sensor points at 80 mph
• Predicts ‘if’ & ‘type’ bearing failure (>97% accuracy)
2
1
0
t
0
-1
-2
158
0.01
0.02
0.03
0.04
Sensor Data
Track based
microphone
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
• Trains pass sensor points at 80 mph
• Predicts ‘if’ & ‘type’ bearing failure (>97% accuracy)
2
1
0
t
0
-1
0.01
0.02
0.03
0.04
Sensor Data
Track based
microphone
-2
159
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Product Design :
Agent-based Optimization from Sensor Data in Semiconductor Wafer Fabrication
elliptically polarized light
circular polarized light
φ1
film thickness
Film
Substrate
d
Optimize cell temperature in order to
optimize for desired refractive index.
160
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
5.5
φ0
Error
Reflected
Incident
100000
10000
1000
100
10
1
0.1
0.01
0.001
0.633
0.000
k2
n2
Optimize growing semiconductor films with ellipsometer sensor.
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
RFID Linked Biometrics & Nano-sensor Net
Blood Glucose
Nano-sensors
161
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Harvard-MIT Center for Integrated Medicine and Information Technology
Healthcare for Independent Living: Sense, then, Respond
Reducing the Cost of Old Age ?
Model
Intelligent
Real Time
Patient Specific
Action Plan Framework with
= Procedure decision support
Actions Precision Remote Controlled
Sensing
Real-time micro-status
networked, mobile
Patient
at home care
162
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Confluence of Technologies
Networking
Embedded Systems
Self-organized, power-aware
communication
Small, untethered processing,
Storage and control
Many devices monitor and
interact with physical world
Coordinate, perform
higher-level tasks
MEMS
Exploit spatial & temporal
coupling to physical world
163
Mass-produced, low-power,
short range, sensors & actuators
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
The White House said the increased technology
spending – mentioned by President Clinton
during last week's State of the Union address –
could be used, for example, to create "intelligent
agents" that roam the Internet collecting data.
AP News Service, 24 January 1999
164
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
165
Inference Technique
Key Feature
Forward Chaining
Data driven
Backward Chaining
Goal driven
Pattern Matching
Fires upon matching of set of criteria
Monitoring
Exception handling and alarms
Truth Maintenance (Retraction)
“What-if” reasoning
Dynamic Inferencing
Scenario-based business rules
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
BANKING/
FINANCE
INSURANCE
Online Mortgage
Underwriting
Point-of-Sale
Underwriting
Credit Scoring
Claims Processing
Portfolio Management
Renewal Processing
Cross Selling
Fraud Detection
Intelligent Policy
Configuration and
Pricing
Overdraft
Authorization
Eligibility
Determination
SEC Regulatory
Compliance
Cross Selling
Risk Management
166
Fraud Detection
MANUFACTURING
GOVERNMENT
Parts Selection
Welfare Eligibility
Determination
Order Configuration
Production
Planning/Routing
Production Scheduling
Regulatory
Compliance
Tax Assessment
OTHER INDUSTRIES
Transportation
Retail
Petroleum/
Oil & Gas
Maintenance and
Labor Scheduling
Entitlements and
Benefits
Determination
Health Care
Material Safety Data
Sheets
Pension Plan
Forecasting
Pharmaceutical
Distribution
Management
Worker’s
Compensation Claims
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Telecom
Utilities
Logic
Intuition
Rules
Pattern
Recognition
Inferencing
Association
Adaptive Pattern Recognition
Rules-Based Systems
•
•
•
•
•
•
167
Prediction Models
Classification – data reduction
State-transition prediction
Recipe: given an input set, predict the outcome
Quality of Models
Measures for False Positives and False Negatives
Rank importance level of each input to the outcome
Principal Component Analysis – dimension reduction
Decision tree: transform relationships into rules
Global optimization
Statistical summaries/correlations
Adaptive to changing environments
Able to deal with complex problems
Unlimited in the number of metrics that can be modeled
Accommodates both linear and non-linear relationships
Data driven – avoid human bias
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
State Transitioning
Input data
Input data
Input data
Input data
Future Event
Prediction
Input data
Input data
168
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Neuron: Non-linear Transfer Function
OUTPUT
Input data
Input data
Input data
Input data
Input data
Input data
Outcome
Outcome
Outcome
Effect
Cause
INPUT
Patterns & Relationships
169
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Training
Prediction
Training Algorithm
Input data
Input data
Input data
Input data
Input data
Input data
Effect
adjustment
•
•
•
•
•
•
•
•
Anticipate component failure
Replace part prior to failure
Preventive maintenance plan
Improve customer response
Reduce repair cycles
Support performance metrics
Better identify causes of problems
Learn to adapt to the environment
Outcome
Outcome
Outcome
Cause
170
Weight
Outcome
Outcome
Outcome
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Prediction
Training / Learning
Algorithm
Input data
Input data
Input data
Input data
Input data
Input data
Effect
adjustment
•
•
•
•
•
•
•
•
Anticipate component failure
Replace part prior to failure
Preventive maintenance plan
Improve customer response
Reduce repair cycles
Support performance metrics
Better identify causes of problems
Learn to adapt to the environment
Outcome
Outcome
Outcome
Cause
171
Weight
Outcome
Outcome
Outcome
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Collaborative Learning Agents
Smart Agents
COOPERATE
Collaborative Agents
172
AUTONOMOUS
LEARN
Interface Agents
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Advances in data routing emerging from study of Ants
Ant–based algorithms developed from swarm intelligence
X
A
Pheromone
173
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Advances in data routing emerging from study of Ants
Ant–based algorithms developed from swarm intelligence
X
A
X
A
Pheromone
174
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Adaptive ?
X
A
175
X
A
X
A
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Portal
Actual Inventory
Smart Shelf
(RFID)
Filters, Logic
Planned Consumption
& Replenishment
Replenishment
Planning
Planned &
Actual Inventory
Planned Inventory
File Sender
Low Inventory Alert
Inventory Early
Warning Agent
Low Inventory Alert
Interface Agent
Backorder
ERP
DW
Illustration: SAP AG
176
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Smart Planning with Intelligent Objects
Store
Store11
ACTION
REPLENISHMENT
X days ?
AGENT
Store
Store22
Inventory Early
Warning Agent
Plant
Plant
2 days
Distribution
Distribution
Center
Center
3 days
Distribution
Distribution
Center
Center
Store
Store33
100
EVENT
MANUFACTURER
Y days ?
50
0
AGENT
Store n
RETAILER
Information Agent
RFID Data
Inventory
Consumption
177
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Multi-Agent System
Data Agents collect ► Data
Monitoring Agent triggers ► Alert
Inventory Management Agent executes ► Substitution
M2 can be
substituted
for SKU M1
Inventory of
M2 is 2000
OOS Danger
Less chance of
a stockout with
substitution via
agent actions
(M1 & M2)
178
Source: SAP AG
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Core Engine
Encapsulations of Application Logic (OR algorithms), Agents, Data, Context, Process Semantics
179
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Adaptive ?
Semantic Tags – SL Tags
“Personal Assistant”
Agents
Agents as Intelligent
Interface Managers
A
A
A
Agent-to-Agent
Communication
Agents
“behind the scenes”
Inter-application
Communication
A
A
ERP
Web of Agents ?
180
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: Marvin Minsky, AI Lab, MIT
Difference Engines (1950)
181
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Source: From Neurons to the Brain
Basic Neural Circuits
182
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
8 corners of larger cube
8 Agents repeated 8 times
8 corners of this cube
1 corner = 1 Agent
8 Agents connected
8 Agents
= 512
183
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
512
512
512
512
Agents interconnected
512
512
184
512
8 X 512 = 4096
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Repeat this cube-on-cube pattern 10 times (10 steps).
Supercube (810 = 1, 073,741,824) will contain over 1 billion Agents.
Each Agent in the original smallest cube (of 8 Agents) can
communicate with 1 billion Agents (sources, variables) in 10 steps.
Link each Agent to 50 other Agents:
Each Agent communicates with >15 billion Agents in 6 steps (506).
CocaCola can monitor nearly each RFID tagged unit case of its
product. Real-time data can be collected by an Agent (Agency) in
mere 6 steps for analysis (inventory, distribution, storage, transit,
temperature). In 2004, CocaCola produced 19.8 billion unit cases.
185
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Distributed Agent Based Models
Model
186
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Distributed Agent Based Models
187
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
188
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Where Artificial Intelligence Meets Natural Stupidity !!
189
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Adaptive Æ Autonomic
Autonomic Agent Architecture
Source: IBM Systems Journal 41 368 (2002)
190
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Languages and Open Standards
XML eXtensions
XML Core
191
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Languages
2003 Ontology Working Language (OWL) DAML + OIL
DARPA Agent Markup Language + Ontology Inference Layer
1999 XML-based Physical Markup Language (PML)
RFID Object Description Language (AIDC, MIT)
1998 eXtensible Markup Language (XML)
World Wide Web Consortium (W3C)
1996 eXtensible Markup Language (XML)
World Wide Web Consortium (W3C) Initiative
1993 HTML Browser Mosaic - Marc Andreessen
National Center for Supercomputing Applications (NCSA) University of Illinois
1989 HyperText Markup Language (HTML) - Tim Berners-Lee, CERN
1986 SGML - International Organization for Standardization (ISO)
1983 SGML Computer Graphics Association (CGA)
1978 Standard General Markup Language (SGML) ANSI Initiative
1975 Document Composition Facility (DCF)
1971 Document Type Definition (DTD)
1969 General Markup Language (GML) - Charles Goldfarb, Ed Mosher, Ray Lorie
Compiled by: David Brock
192
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Development
• Individuals
• Academia
• Corporations
• Industry Consortia
• Government
• International Organizations
193
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Compiled by: David Brock
Explosion ……
4ML
AML
AML
AML
AML
AML
AML
ABML
ABML
ACML
ACML
ACAP
ACS X12
ADML
AECM
AFML
AGML
AHML
AIML
AIML
AIF
AL3
ANML
ANNOTEA
ANATML
APML
APPML
AQL
APPEL
ARML
ARML
ASML
ASML
ASTM
ARML
ARML
ASML
194
ARML
ARML
ASML
ASML
ASTM
ATML
ATML
ATML
ATML
AWML
AXML
AXML
AXML
AXML
BML
BML
BML
BML
BML
BML
BannerML
BCXML
BEEP
BGML
BHTML
BIBLIOML
BIOML
BIPS
BizCodes
BLM XML
BPML
BRML
BSML
BCXML
BEEP
BGML
BHTML
BiblioML
BCXML
BEEP
BGML
BHTML
BIBLIOML
BIOML
BIPS
BizCodes
BLM XML
BPML
BRML
BSML
CML
xCML
CaXML
CaseXML
xCBL
CBML
CDA
CDF
CDISC
CELLML
ChessGML
ChordML
ChordQL
CIM
CIML
CIDS
CIDX
xCIL
CLT
CNRP
ComicsML
CIM
CIML
CIDS
CIDX
xCIL
CLT
CNRP
ComicsML
Covad xLink
CPL
CP eXchange
CSS
CVML
CWMI
CycML
DML
DAML
DaliML
DaqXML
DAS
DASL
DCMI
DOI
DeltaV
DIG35
DLML
DMML
DocBook
DocScope
DoD XML
DPRL
DRI
DSML
DSD
DXS
EML
EML
DLML
EAD
ebXML
eBIS-XML
ECML
eCo
EcoKnow
edaXML
EMSA
eosML
ESML
ETD-ML
FieldML
FINML
FITS
FIXML
FLBC
FLOWML
FPML
FSML
GML
GML
GML
GXML
GAME
GBXML
GDML
GEML
GEDML
GEN
GeoLang
GIML
GXD
GXL
Hy XM
HITIS
HR-XML
HRMML
HTML
HTTPL
HTTP-DRP
HumanML
HyTime
IML
ICML
IDE
IDML
IDWG
IEEE DTD
IFX
IMPP
IMS Global
InTML
IOTP
IRML
IXML
IXRetail
JabberXML
JDF
JDox
JECMM
JLife
JSML
JSML
JScoreML
KBML
LACITO
LandXML
LEDES
LegalXML
Life Data
LitML
LMML
LogML
LogML
LTSC XML
MAML
MatML
MathML
MBAM
MISML
MCF
MDDL
MDSI-XML
Metarule
MFDX
MIX
MMLL
MML
MML
MML
MoDL
MOS
MPML
MPXML
MRML
MSAML
MTML
MTML
MusicXML
NAML
xNAL
NAA Ads
Navy DTD
NewsML
NML
NISO DTB
NITF
NLMXML
NVML
OAGIS
OBI
OCF
ODF
ODRL
OeBPS
OFX
OIL
OIM
OLifE
OML
ONIX DTD
OOPML
OPML
OpenMath
Office XML
OPML
OPX
OSD
OTA
PML
PML
PML
PML
PML
PML
PML
PML
P3P
PDML
PDX
PEF XML
PetroML
PGML
PhysicsML
PICS
PMML
PNML
PNML
PNG
PrintML
PrintTalk
ProductionML
PSL
PSI
QML
QAML
QuickData
RBAC
RDDl
RDF
RDL
RecipeML
RELAX
RELAX NG
REXML
REPML
ResumeXML
RETML
RFML
RightsLang
RIXML
RoadmOPS
RosettaNet PIP
RSS
RuleML
SML
SML
SML
SML
SAML
SABLE
SAE J2008
SBML
Schemtron
SDML
SearchDM-XML
SGML
SHOE
SIF
SMML
SMBXML
SMDL
SDML
SMIL
SOAP
SODL
SOX
SPML
SpeechML
SSML
STML
STEP
STEPML
SVG
SWAP
SWMS
SyncML
TML
TML
TML
TalkML
TaxML
TDL
TDML
TEI
ThML
TIM
TIM
TMML
TMX
TP
TPAML
TREX
TxLife
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
UML
UBL
UCLP
UDDI
UDEF
UIML
ULF
UMLS
UPnP
URI/URL
UXF
VML
vCalendar
vCard
VCML
VHG
VIML
VISA XML
VMML
VocML
VoiceXML
VRML
WAP
WDDX
WebML
WebDAV
WellML
WeldingXML
Wf-XML
WIDL
WITSML
WorldOS
WSML
WSIA
XML
XML Court
XML EDI
XML F
XML Key
XMLife
XML MP
XML News
XML RPC
XML Schema
XML Sign
XML Query
XML P7C
XML TP
XMLVoc
XML XCI
XAML
XACML
XBL
XSBEL
XBN
XBRL
XCFF
XCES
Xchart
Xdelta
XDF
XForms
XGF
XGL
XGMML
XHTML
XIOP
XLF
XLIFF
XLink
XMI
XMSG
XMTP
XNS
Systems 2000
DATA MOBILITY
TOOLS
CONVERGENCE
OR and Game Theory
Languages
Distributed Artificial Intelligence
Autonomous Agents
Semantic Web
Grid Computing
Simulation
Streaming Database
Clockspeed
802.11b / WiFi
802.11a, 802.11g, 802.16
BlueTooth
Mesh Networks
Ultrawideband (UWB)
Sensors (MEMS ÆNEMS)
GPS / RTLS
IPv6, 4G
Radio Frequency Identification
Adaptive Mobile eXtended Decision Systems 2020
195
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
SEMANTICS
Adaptive ?
Process
Context
• Vendor Managed Inventory (P&G)
• Just-in-Time Distribution (Barilla)
Process names, context of words and their meanings, usage,
differs with country, industry and host of other factors that
may not be standardized reflecting one universal description.
VMI ≈ JTID
196
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Relationships
197
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Tim Berners-Lee
Semantic Web
198
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Semantic Web Bus
www.w3c.org
199
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Simulation •DML
•ADL
•DMP
•ADP
ADL
ADP
Data Modeling Language
Automated Decision Language
Data Modeling Protocol
Automated Decision Protocol
ADP
Host
DMP
DMP
DML
200
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Applications
201
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Applications
202
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Applications
203
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Past
Network
204
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Present
Network + Data
205
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Future ?
Network + Data + Models
206
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Distributed Interactive Simulation
“A Template for Distributed Modeling”
207
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Data Modeling Language (DML)
Mode
l
Input
208
Output
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Data Modeling Language (DML)
209
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Data Modeling Protocol (DMP)
DMP
210
DMP
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Automated Decision Language (ADL)
Input
Command
211
Output
Command
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Automated Decision Protocol (ADP)
212
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Automated Decision Protocol (ADP)
ADP
ADP
213
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Extract Intelligence from Real-Time Data to Feed (information) Processes
DATA
Architecture
214
Standards
Models
Hardware
Software
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Applications
Data Models
Model
215
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Data Models
Model
216
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Data Models
217
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Standards
need for standard
The more complex the network, the greater the need for standards.
value network complexity
218
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Standards: Data Project
Data Modeling Languages and Protocols
DML – Data Modeling Language
DMP – Data Modeling Protocol
Data Control Languages and Protocols
ADL – Automatic Decision Language
ADP – Automatic Decision Protocol
219
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
Where is the ROI ?
•
An Analogy from Quantum Physics
•
•
•
•
•
•
•
•
•
•
220
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
221
•
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
•
•
•
•
•
•
••
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
222
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
••
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
• •
•
•
•
•
•
•
•
•
•
•
••
•
•
•
•
•
•
•
•
•
•
•
• •
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
••
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
• •
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
• •
•
•
•
•
•
•
••
•
• •
•
•
•
•
•
•
•
•
•
•
•
•
••
•
•
•
•
•
•
•
•
•
• •
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
•
•
•
•
•
•
•
•
•
• • •
•
•
• •
•
•
•
•
•
•
•
•
•
• •
•
•
•
•
•
•
• •
• • •
• •
• •
•
•
•
•
•
•
•
•
•
•
• •
• • • • •• • •
•
• •
•
•
•
•
•
•
•
• •• • •
•
• •
•
•
• •
•
•
•
•
• •
•
•
• • •
•
•
•
•
•
• •
• • • •
•
•
•
•
•
•
•
• • •• ••
•• •
• • •
••
• • •• • • •
•
• •
•
••
•
•
•
•
•
•
•
•
•
•
•
•
•
• •
•
•
• •
••
• • • ••
•
• ••
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
• • •
• ••• ••
•
•
•
• • • •
••
•
•
• • •
• ••
• •
•
•
•
•
•
•
•
•
•
••
•
•
• •
• •
• • •
• ••• • • •
•
••
•
•
•
• •• • ••
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
• • •
• •
•
• • •• •
• ••
•• •
•
••
• •
•
•
•
•
•
•
•
•
•
• •
• • • • • •• • • •
•
• • •
••
•
• •
•
•
•
•
• •
•
• ••
•
•
•
•• •
•
•
•
••
••
• •
•
• • • • • •• •
•
•
•
•
• •• • •
••
• •• •• •• •
•
•
•
•
•
•
•
•
• •
•
•
• •
••
• • • ••
•
•
•
•
•
•
•
•
•
•
••
•
• • •
•
•
• •
•• • • •
• •
•
• •
•
•
•• •• •
••
•• • •
•
•
•
•
•
•
•
•
•
•
•
• •• • •
• •
• • •
•
••
•
•
•
• • • • •• •
•
••
•
• •
• •
• •
•
•• • • • •
•• •
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•• •
•
•
•
•
•
• • •
•• • • • • • •• • • • •
•
•
•
•
• •
•
••
• •••
•
• •••
•
• •
• • • • • •• •
•
• • • •• •
••
•
•
•
•
•
•
•
•
•
•
•• •
•
•
• •
•
•
•
•
•
• • •
•
••
•
•
•
•
•
•
•
•
•
•
•
•
•
•
• •
•
•• • •
•
•
• • • • • ••
•
•
•
•
•
•
• •
•
•• •
• • • • • • •• •
•
•
• • •
• •• •
•
•
• •• •
• •• •
•
• • •
•
•
•
•
•
•
•
•
• • • • • •• • • •• •
••
• •
•
•
•
•
•
•
•
•
•
•
•
••
••
• • •• • •
•
•
•
••
•
•
•
•• •
•
•
• •
•
••
• •
•
•
•
•
•
••
•• •
•
•
•
•
•
•
•
•
•
•
• •
•
•
•
•
•
•
•
•
•
••
• •
•• • •
•• • • • • •
• • ••
• • •• ••
• • •
•
•
•
•
•
• •
•
• ••
•
• •• •
•
•
• • ••• • •
•
•
• • •• • •
•
••
•
•
•
•
•
•
••
•
•
•
•
• •
•
• • • •• • •
• •
•
•
•
•
• •
•
•
• •• • •• • • •
• • •• ••
•
•
•
•
•
• • •• •
•
•
•
•
•
•
•
•
•
•
•
•
•
• •• •
•
•
• • • •
• •
•
• •
••
•
•
•
•
••• •••
•
• • •
•
•
•
•
•
• •• • ••
•
•
••
•
•
•
•
•
•
• •
••
•
• •
••
• •
•
•
• •• ••
•
•
••
•
•
•
•
•
• • • • • • •
•
•
•
•
•
•
•
•
•
•
•
• •
•
•
•
•
•
•
• •
•
• •
•
•
•• ••
• •
•
•
•••
•
• •
•
••
• •
• •
Shoumen Datta, MIT
Forum
for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
•
•
•
•
•
•
• •
•
•
•
•
•
•
•
• •
•
•
•
•
•
••
•
•
•
•
•
•
•
•
•
•
•• ••
•
•
•
•
• •• •
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
••
•
•
•
•
•
•
•
•
•
• •
•
•
•
•
•
•
•
•
•
•
•
•
•
•
• ••
•
•• ••
• •
•
•
•
•
•
•
•
•
•
•
•
• ••
•
••
•
•
•
•
•• ••
•
• •
•
• •• •
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
••
• •
•
•
•
•
•
••
•
• •
•
•
•
•
•
•
•
•
•
•
••
•
•
•
•
•
• •
• •
•
•
•
•
•
•
•
•
•
••
•
•
•
•
•
•
•
•
•
•
• •
•
•
• •
•
• •• •
•
•
•
•
•
•
•
•
•
•
•
••
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
• •
• • •
•
•
•
•
•
•
•
•
•
•
••
•
•
•
•
223
Young’s Double Slit Experiment
Young’s Double Slit Experiment with Electrons
Dr. Akira Tonomura, Hitachi Research Laboratories
•
••
••
• • •
• • • •• •
• •
•
•
•
•
• ••
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•• ••
•
•
•
•
•
•
• •
•
•
•
•
• ••
•
• •
•
•
• • • •• • • •• •
•
• •
•
•
• •
•
•
•
••
•
• • • • • ••
• •
•
• ••
• •• •
• •
•
• •
• • •
• • • •• • •
••
•
•
•
•
•
•
•
•
•
•
•
• •
•
•
•
•
•
••
•
•
•
•
•
•
•
•
•
•
•
• •• •
•
•
•
•
•
•
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>
•
•
••
•
•
• •
•
•
•
•
•
•
•
•
•
•
•
•
•
• •
••
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
••
•
• •
••
•
•
•
•
•
•
•
•
•
•
•
•
•
••
•
•
••
•
•
• •
••
•
•
•
•
•
•
•
•
•
•
••
•
•
•
•
•
•
•
•
• •
•
• •
•
•
•
•
• •
• •
•
• •
•
• • •
•
•
• •
•
•
•
•
•
• •
•
•
•
•
•
• •
•
• •
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
• •
•
•• •• •
•
•
••
••
••
••
•
•
•
•
••
•
• •
•
•
•
•
•
•
•
•
•
• •
•
•
•
•
• •
•
•
• •
••
••
••
••
•
•
•
••
•
• •
•
•
•
• •
•
•
•
•
•
•
•
•
•
•
• •
•
•
•
•
• • •
•
•
• • •
•
•• •
•
•
•
•
•
•
• •
•
•
•
• •
• • •
• ••
•
•
•
•
•
224
•
•
•
•
•
•
•
•
• •
•
•
•
•• •
•• •
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
• •
•
•
•
• •
• • •
• •
• •
•
•
• •••• ••
•
•
•
•
•
• • • • •
•
••
• • ••
•
• ••
•
•
• •
•
• • •
•
• • ••
••
•
•
••
• •
•
• •
•
•
•
• •• • • •
• •• • • •• • •• • • • • • • •
• •
• • •
• ••
•
•
• •• • • • • • •
•
••
•
•
• • • •• •
• •
•
•
• • • •• •
•
••
• •• •
•• • • •• •
• •
•
• • • • • • •
•
• •
•
•• • • •
•
• • • •
••
•
•
•
•
• • • • • •• • • •
•
•
• •
• •
•
•• • • • •
•
•
•
•
• • • • • • •• • • • •
•
• •• • • • • •
• •• •
• • • •• •
• • •
•
• •
•• •
•
•
•
•
•
•
• •
•
•
•
•
• •
•
• • • • •• •
•
•
•
•
•
••••
• •
•
•
•
•
•
•
•
•
• • • • • •• •• • ••• •
• •
• •
• • •• • •
•
•
•
•• •
•
•
•
• •
• ••• •
•
•
•
•• • •
•• • • • • •
•
•
• • ••
•
• •• •
•• •
•• • • • • •
• ••
•
• •
•
• • •• •• • •
• •• ••• • • •
• • •• ••
•
• • • •• ••
• •
•
•• • • • •
• •
•
•••
•
•
•
•
•
•• •
• •
•
• •
•
••
• •
•• •• •
•
•
• • • • • • •
• • • •
• • ••
• •
• •
•
•• • • •
•
•
• •
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
• •
•
•
•
•
•
• •
•
•
•
• •
• •
••
•
••
•
• •• • •
•
•
•
• •
•
•
•
•
• •
•
• • • •
• • •• • • •
•
•
• • •
• •• • • • ••
•
•
•
•
•
•
• ••• •
• • •
•
• •
•
• • •
•
•
•
•
• • •• •• ••
•
•
•
•
••
•
••
• •
•
•
•
•
•
••
•
•
• •• • •
•
•
•
•
•
•
•
• •
•
•• •
• •
•
• • • • ••
•
• • •
•
• •
••
•
•• •
• • •
•
•
••
•
• ••
•
••
•
•
•
•
•
• •
••
•
•
•
••
• • • • • ••
•
•
•
• •• •
•
• • •
• • •
•
•
•
• •
•
•
•
•
•
• •
•
•• • ••
•
•
•
• •
•
••
• • ••
•• • •
•
•
• ••
• • •••••
•
• •
•
•
•
• •
•
•
•
•••
•
• •• •
•
•
••
•
• • •
•
•
• •• •
•
•
•
•
•
• ••
•
•
••
••• ••
•
•
•
•
•
•
225
Shoumen Datta, MIT Forum for Supply Chain Innovation, School of Engineering <shoumen@mit.edu>