The System Shell as a Construct for Mitigating the Impact...
Transcript of The System Shell as a Construct for Mitigating the Impact...
The System Shell as a Construct for Mitigating the Impact of Changing Contexts by Creating
Opportunities for Value Robustness 1st Annual IEEE Systems Conference
Honolulu, HI, April 9-12, 2007
Presented by Adam Ross, Ph.D.April 11, 2007
seari.mit.edu © 2007 Massachusetts Institute of Technology 2/21
Achieving System Success
• Goal of design is to create value (profits, usefulness, voice of the customer, etc…)
• Requirements capture a mapping of needs to specifications to guide design
seari.mit.edu © 2007 Massachusetts Institute of Technology 3/21
Deploying a “Valuable”System…
Contexts change…
seari.mit.edu © 2007 Massachusetts Institute of Technology 4/21
Achieving System Success (cont.)
• Goal of design is to create value (profits, usefulness, voice of the customer, etc…)
• Requirements capture a mapping of needs to specifications to guide design
• People change their minds; the world changes…• To continue to deliver value, systems must cope
with context change…
seari.mit.edu © 2007 Massachusetts Institute of Technology 5/21
What is Context?
• Context includes forces exogenous to system– Stakeholder expectations– Operating environment– Policy constraints– Available technologies– Competitive market– Etc…
• System success depends on system performance within a given context
• In order to ensure success, designers must consider context beyond traditional “operating environment” (classical robust design)
seari.mit.edu © 2007 Massachusetts Institute of Technology 6/21
What is System Success?
System success, Ψ, across N decision makers at time t
[ ]∑=
+≥+=ΨN
i
YCDMi
XCDMi ttYttXt DMiDMi
1)()()()()( εε
)(tXDMi )(tDMiXCε
)(tYDMi )(tDMiYCε
Decision maker i system “experience” at time t
Decision maker i system “expectation” at time t
Context effect on decision maker i “experience” at time t
Context effect on decision maker i “expectation” at time t
Nt ≤Ψ≤ )(0
System Success: Net “experience” must meet or exceed net “expectations”
Success is defined across multiple perspectives and multiple time periods
Net “experience” Net “expectations”
seari.mit.edu © 2007 Massachusetts Institute of Technology 7/21
“Context”
Dynamic System Context: Value Lenses
*“Decision”, “Experienced”, and “Remembered” Utility from (Kahneman and Tversky 2000)
**Discussion of “structural” versus “operational” metrics in (Giachetti et al. 2003)
SystemInputs Outputs
System Boundary
Value
“Experienced”
Value
“Experienced”
ConstraintsStatic ViewStructural Metrics
Delay
Dynamic ViewOperational Metrics
Value
“Decisional”
ValueValue
“Decisional”
“Expectations” “Experience”
Value
“New”
Value
“New”
Value
“New”
Value
“New”
seari.mit.edu © 2007 Massachusetts Institute of Technology 8/21
Construct: Black Box System
seari.mit.edu © 2007 Massachusetts Institute of Technology 9/21
Black Box System
The box has attributes (function and/or form) that
are value-perceived
Value
“Experienced”
Value
“Decisional”
seari.mit.edu © 2007 Massachusetts Institute of Technology 10/21
Value Robust Black Box
External Change
Box is perceivedto deliver the same value The box is robust if it can maintain its value
delivery as external “context” changes
Value
“Experienced”
Value
“Experienced”
New!Value
“Experienced”
Value
“Experienced”
New!
Value
“Experienced”
New!Value
“Experienced”
Value
“Experienced”
New!
seari.mit.edu © 2007 Massachusetts Institute of Technology 11/21
Robustness Defined
No change in perceived value
“Robust”A box can be quantified in terms of
robust in Xi to “Input” change(i.e., can Xi remain “constant” over
range of “Input”?)
Xgoal
∆Inputs
“Constant”
Level of attribute performance is function of inputs (and constraints including environment), so robustness is an insensitivity to the inputs (and constraints)
X
Inputs
seari.mit.edu © 2007 Massachusetts Institute of Technology 12/21
System Shell Illustrated
Inner Shell (The Shelter)
Change Context as seen by Box
Outer Shell(The Mask)
Change Box as seen by Context The shell protects the box from changes in
its context
The shell filters the appearance of the box to its context
seari.mit.edu © 2007 Massachusetts Institute of Technology 13/21
Examples: Shelter
Shelters “protect” system from context more cheaply than modification of system itself
+
Context: Noisy environment Context: Hostile external network
Solution: FirewallSolution: Earplugs
seari.mit.edu © 2007 Massachusetts Institute of Technology 14/21
System Mask as a Filter
A B
A
AB
B
A B
A
AB
B
System Mask customizes “experience” to meet expectations
User desires “A”
User sees “A”
User desires “AB”
User sees “AB”
User desires “B”
User sees “B”
System displays “AB”Success!
Success!
Success!
seari.mit.edu © 2007 Massachusetts Institute of Technology 15/21
Photo courtesy U.S. Department of Defense
GPS System “Appearance”
Examples: Mask
Photo courtesy Garmin
Cellular Phone “Appearance”
Images from http://www.akcessories.com/index.html?lmd=39141.647419
Masks filter “appearance” of system to context more cheaply than modification of system itself
Solution: Faceplates
Solution: Various interpretive receivers
seari.mit.edu © 2007 Massachusetts Institute of Technology 16/21
Examples: Shell
Image from Amazon.com, SamsoniteConverter/Adaptor Plug Kit with Pouch
Int'l Distribution of Voltage
90100
110115
120127
130140150160170180190200210
220230
231240
0 10 20 30 40 50 60 70 80 90 100
90
110
120
130
150
170
190
210
230
240
Volta
ge (V
)
Frequency
Int'l Distribution of Adapter Type
0
20
40
60
80
100
120
A B C D E F G H I J K L M
Adapter Type
Freq
uenc
y
Source: http://www.kropla.com/electric2.htm
Changing Context: International Electrical Power Sockets
Plug in device must “look”different to each electric socket
Device must be “protected” from varying voltage at each electric socket
Solution: Adapter (Mask) and Converter (Shelter)
(N)
(N)
seari.mit.edu © 2007 Massachusetts Institute of Technology 17/21
Examples: Software Shell
X-TOSSource: http://www.phoenix-int.com/products/modelcenter.php
Mask: standardize code appearance to ModelCenter
Shelter: provide proper code input translation from ModelCenter
Result: “black box” software code; model plug-in capability
Changing Context: Software Models to be Linked for Analysis
Plug-in software models must “look” the same to the Analyst
Plug-in software models must be “protected”from varying input/output formats
seari.mit.edu © 2007 Massachusetts Institute of Technology 18/21
…
For n = 1:3 do
…
end…CodeComponent
Expected inputActual input Component Expected output Actual output
Example: Software/SoS Shell
in out
in out
in out
in out
Component set(virtually limitless)
Sheltercustom translation code
Mask standardized format
Result: “system of system” design with standard component appearance
Appearance to SoS
in out
in out
in out
seari.mit.edu © 2007 Massachusetts Institute of Technology 19/21
Flow of Value Robustness
X
DV
DV
ContextX
U
Shelter
Robustness strategies can be pursued at various locations in value flow from context to value perception
Suh Axiomatic Design
Robustness
Taguchi Parameter Design
Robustness“Success” occurs here
Mask
seari.mit.edu © 2007 Massachusetts Institute of Technology 20/21
Discussion
• When should a shell be used?
• Implications for design– Extend operating ranges (augmented robustness)– Customized user experiences (multi-stakeholder
satisfaction)– Distribution of costs for “shell”
• System of System design• Separating changeable system parts
Benefitold_sys+shell – Costshell > Benefitnew_sys – Costnew_sys
i.e., Net Benefit of Shell > Net Benefit of New System
seari.mit.edu © 2007 Massachusetts Institute of Technology 21/21
Conclusion
• System Shell construct decouples system from context through “protection” and “filtering”
• System Shell can be part of system or layered on top
• May be a cost-effective and/or timely solution to multiple perspective, multiple time period context
Using a System Shell is an effective technique for achieving Value Robustness
Thank you for your attention!
Any questions?For further details on topic please see:
Ross, Adam M., Managing Unarticulated Value: Changeability in Multi-Attribute Tradespace Exploration.
Cambridge, MA: MIT. PhD in Engineering Systems. 2006.