ADRIAN A. MARSH
ENSE 622 
Suppliment to Project Brief
19 May, 2003



CONTINUATION OF COMBAT MARKING SYSTEM 
DESIGN


























Combat Engineers marking a breach lane at the Combined Maneuver Training Center, 
Hohenfelds, Germany, April 2002.









TABLE OF CONTENTS

A.  SYSTEM REVIEW	4
1. Background	4
2. Purpose:	4
3.  Refining System Requirements:	4
4.  Refining System Structure:	4
5.  System Behavior	6
6.  Turning System Requirements into System Specifications	6
B.  DETAILED ANALYSIS (HAND CONTROL UNIT)	7
1.  Purpose:	7
2.  Revised System Structure and Behavior:	7
3.  Turning Specifications for HCU into Commercial Component Options	8
C.  TRADE-OFF ANALYSIS	9
1.  Gathering the Data:	9
2.  Decision Metrics	9
3.  Constraints	9
4.  Problem Formulation	9
5.  Running CPLEX	11
6.  First Pareto Plot Analysis	11
7.  Second Pareto Plot Optimization	12
8.  System Optimization	13
D.  COMPONENT CONFIGURATION RECOMMENDATION:	16
REFERENCES	17
APPENDIX A: Origional System Requirements
APPENDIX B: Updated System Requirements
APPENDIX C: Total System Specifications
APPENDIX D: Hand Control Unit Specification Sheet (SPEC Sheet)
APPENDIX E: SPEC sheet with vendor data
APPENDIX F: Initial Graph of Design Options
APPENDIX G: Initial Pareto Analysis
APPENDIX H: Second Graph of Design Options
APPENDIX I:  Second Pareto Analysis
APPENDIX J: CPLEX equation input
APPENDIX K: MIN MAX Life Equation Method


FIGURES AND TABLES


FIGURE 1: ORIGINAL COMBAT MARKING SYSTEM STRUCTURE	4
FIGURE 2: UPDATED COMBAT MARKING SYSTEM STRUCTURE	5
FIGURE 3: COMBAT MARKING SYSTEM BEHAVIOR DIAGRAM	6
FIGURE 4: HAND CONTROL UNIT SYSTEM STRUCTURE	7
FIGURE 5: HAND CONTROL UNIT SYSTEM BEHAVIOR DIAGRAM (FOR SYSTEM'S 
FUNCTION CHECK)	8
TABLE 1: COMMERCIAL OPTIONS FOR HCU SUB-COMPONENTS	8
TABLE 2: CPLEX RUN RESULTS	11
FIGURE 6: EXAMPLE OF PARETO POINT ANALYSIS	11
TABLE 3:  RESULT OF INITIAL PARETO ANALYSIS	12
TABLE 4:: FINAL SOLUTION OPTIONS	12
FIGURE 7: PARETO ANALYSIS OF OPTION 6	13
FIGURE 8: PARETO ANALYSIS OF OPTION 3	14
FIGURE 9: PARETO ANALYSIS OF OPTION 10	15
TABLE 5:  FINAL OPTION COMPARISON	16
TABLE 6: VENDOR SELECTION RESULTS	16





A.  SYSTEM REVIEW 
1. Background
	In ENSE 621, the need for a Combat Marking System was identified and served as the focus for 
the initial work on the modeling of this system.  Initial high-abstraction requirements were determined, 
from those requirements the system was modeled through the use of Use-Case Scenarios to refine 
requirements.  From the refined requirements high-level system structure and behavior diagrams were 
constructed to represent the system. 
2. Purpose:
	The purpose of this report is to refine the requirements created in ENSE 621, turn the requirements 
into specifications, and conduct a trade-off analysis to optimize components of the system design.
3.  Refining System Requirements:
	Close examination of the requirements identified in ENSE 621 demonstrated a need for 
refinement.  The initial requirements outlined were based on a high level abstraction and were not suitable 
for component level design.  Additionally, the initial requirements did not include Attributes and Functions 
in the requirement list, which are essential to determining component design.  Finally, the initial 
requirement numbering system needed to be updated to facilitate refined traceability needs (See 
APPENDIX A for initial system requirements).

	The initial requirements were updated in the subsequent design of the system.  Attributes and 
Functions were assigned to each object in the requirements list, additional requirements were added to 
support a lower level of abstraction, and the requirement numbering system was updated (See APPENDIX 
B for updated requirement list).
4.  Refining System Structure:
	Following the refinement of the system requirements an update of the system structure diagram 
was conducted to incorporate the changes.  The initial system structure diagram was at a high level of 
abstraction and did not sufficiently include Attributes and Functions needed for design (See Figure 1).
Figure 1: Original Combat Marking System Structure
	A new system structure diagram was required to fully capture the new system requirements 
(Appendix B).  The updated class diagram includes additional components and the Attributes and Functions 
derived from the new requirements (See Figure 2 below).

Figure 2: Updated Combat Marking System Structure 


















5.  System Behavior
The system behavior outlined in the initial analysis depicts the interaction of the system components.  For 
the purposes of the overall system, the behavior diagram was not adjusted from the original project.  The 
behavior of specific low-level abstraction components will be addressed later in the report.  The overall 
system behavior is modeled below (See Figure 3).







































FIGURE 3: Combat Marking System Behavior Diagram

6.  Turning System Requirements into System Specifications
	Once the system requirements were refined, system specifications were needed to set limits on the 
component design.  The system specifications were comprised from the Attributes and Functions outlined 
in the system requirements.  These govern the formulation of specifications by stating what the component 
must do (Functions) and also what characteristics they must have (Attributes).  The system specifications 
for this project were largely determined from analysis of similar systems already in existence, since this 
project is an innovative system (See APPENDIX C for Total System Specifications). 

	Now that the overall system has been reviewed the remainder of this report will detail the selection 
of commercial components for a sub-component of the total system.

B.  DETAILED ANALYSIS (HAND CONTROL UNIT)
1.  Purpose:  
The remainder of this report will detail the process of selecting commercially made components, 
to meet the project specifications, for the Hand Control Unit (HCU).  
2.  Revised System Structure and Behavior: 
Focussing on the HCU, the class diagram is modified to present the components of interest: the 
LCD display, the keyboard, and the processing unit (see Figure 4).  Additionally, the subsystem behavior is 
used in examining the Hand Control Unit in detail.  Figure 5 depicts the behavior of the selected system in 
conducting a systems functions check.  This will assist us in optimizing the sub-component interaction.

FIGURE 4: Hand Control Unit System Structure






FIGURE 5: Hand Control Unit System Behavior Diagram (for system's function check)

3.  Turning Specifications for HCU into Commercial Component Options

	The system requirements and specifications for the HCU were taken from the listings described in 
APPENCICES B and C and simplified into a SPEC sheet for the HCU (See APPENDIX D).  This HCU 
SPEC sheet, which quantifies the Attributes and Functions in the Requirement listing, were used to 
determine commercially available products to comprise the HCU, display, keyboard and processor.  A 
search of vendor websites provided numerous component options to meet the needs of the Hand Control 
Unit (See Table 1).  In order to optimize the system a Trade-Off Analysis is required to determine the 
proper mix of components.


COMPONENTS
POSSIBLE VENDOR OPTIONS
LCD
Pilz
PX20
Powerview
AP9215
LightView
QVGA-0076
CPU
PastDSP
Multiproc
Mango DSP
PCI 64/66
Mango PCI
FP32C
KEYPAD
Mark SYS
INC
SPEC INC
AX-50
Spectra
SYS Z21

Table 1: Commercial Options for HCU Sub-Components


C.  TRADE-OFF ANALYSIS
1.  Gathering the Data:
The first step in conducting trade-off analysis was to gather the vender's performance data for their 
components and place it into our specification database (see APPENDIX E).  This structure will provide 
the tools used to formulate performance equations.
2.  Decision Metrics
	Metrics are determined, in order to optimize the system for the most important criteria.  The 
selection of HCU sub-components will be evaluated using 3 metrics.  These will be the foundation for the 
analysis; equations will be structured in order to do the following:
		1.  Maximizing Total System Life
		2.  Minimizing Total System Weight
		3.  Minimizing Total System Cost
3.  Constraints
	In addition to the metrics listed above, the sub-components will be evaluated based on additional 
performance constraints, selected to optimize the performance of the HCU.  The constraints for the system 
are as follows:
		1.  The minimum sub-component life greater than or equal to 100,000 uses
		2.  The LCD surface area greater than or equal to 10 in2
		3.  The character height on the LCD greater than or equal to 5.2 MRAD
		4.  The processing memory of the CPU greater than or equal to 500 MB
		5.  The keypad width greater than or equal to 5 in
		6.  The Total Power usage of the system less than or equal to 15W
4.  Problem Formulation
	Using the Vendor SPEC sheet, the metrics listed above and the constraints, linear equations are 
formulated to input into CPLEX, to optimize the system.  The mathematical equations based on the 
information is as follows:


THEORETICAL EQUATION BASIS:
	
OPTIMIZATION EQUATION
	  + + 

DECISION VARIABLES 
  
 

 

 

ACTUAL EQUATION FORMULATION
	By applying the vendor data (APPENDIX E), to the metrics and constraints it is possible to 
formulate equations.  These equations were placed into CPLEX to optimize for component life.  Actual 
equations are as follows:
OPTIMIZE:     
METRIC 1: Component Life=500x1+750x2+830x3+100y1+200y2+160y3+500z1+100z2+200z3  *

SUBJECT TO:
METRIC 2: Weight=850x1+750x2+650x3+300y1+450y2+250y3+900z1+750z2+800z3= K1
METRIC 3: Cost=380x1+400x2+450x3+400y1+420y2+500y3+450z1+300z2+500z3=K2

CONSTRAINT 1: 
 	 System Life LCD: 500x1+750x2+830x3>=100 *
 	System Life CPU: 100y1+200y2+160y3>=100 *
 	System Life KEYPAD: 500z1+100z2+200z3>=100 *

CONSTRAINT 2: LCD Surface Area: 11x1+15x2+18x3>=10
CONSTRAINT 3: Character Height: 5.5x1+6.0x2+6.5x3>=500
CONSTRAINT 4: Processor Mem: 535y1+576y2+768y3>=500
CONSTRAINT 5: Keypad Width: 5.0z1+5.5z2+6.0z3>=5.0
CONSTRAINT 6: Power Usage: 3.8x1+4.5x2+2.8x3+5.6y1+6.3y2+4.0y3+2.0z1+3.0z2+2.8z3<=15
CONSTRAINT 7: x1+x2+x3=1
CONSTRAINT 8: y1+y2+y3=1
CONSTRAINT 9: z1+z2+z3=1
*For the purpose of analysis, product life was scaled from the hundred thousands to hundreds place, in 
order to produce better trade-off analysis results.


*CPLEX Equation:  See APPENDIX J for actual CPLEX equation 

**Equation Note: The optimization of life in this case required determining the highest sub-component life, 
in order to maximize the total system life.  This computation is difficult to represent as a linear equation, in 
essence it would have to pick the highest component life systems, then determine the lowest life of the total 
system (which is the weakest component) then optimize to find the longest lasting of the weakest system.  
To conduct this optimization, the assumption was made to allow the sum the component lives to find the 
optimal system.  From this data a spreadsheet was used to find the weakest sub-component of the system.  
The assumption was that the highest sum value of lives would include the best component solution.  Since 
this is not completely reliable, constraints were added to the equation to further optimize life.  The life 
constraint allowed the minimal acceptable life to be used as screening criteria (See APPENDIX K for 
further explanation of method).   










5.  Running CPLEX
	Once the equations were determined, CPLEX was used to optimize the system.  The variables K1 
and K2 listed above were varied 12 times and run into CPLEX.  CPLEX optimizes the solution and 
provides sub-component selections by allocating the number 1 by the optimal solution.  The results of the 
CPLEX runs are listed below (Table 2):
 




LCD 1
LCD 2
LCD 3
CPU 1
CPU 2
CPU 3
KP 1
KP 2
KP 3
Option
Cost 
(K2)
Life
Weight 
(K1)
x1
x2
x3
y1
y2
y3
z1
z2
z3
1
2200
200
2300


1

1

1


2
1250
100
1650


1

1


1

3
2000
200
2000


1

1

1


4
1080
100
2200
1


1



1

5
1450
100
1650


1


1

1

6
1600
160
1750


1


1


1
7
1300
100
2000


1
1


1


8
1080
100
1900
1


1



1

9
1280
100
1900
1


1



1

10
1450
160
1800


1


1
1


11
1100
100
1800

1

1



1

12
1300
100
1850


1
1


1


Table 2: CPLEX run results
6.  First Pareto Plot Analysis
	Using the data in Table 2, graphs were created in order to identify Pareto, metric trade-off, points.  
Analysis of these points produced the first set of optimal solutions.  The Pareto analysis produced the 
graphs in APPENDIX F.  The actual Pareto points and discussion are presented in APPENDIX G.  These 
points were selected because they were trade-off optimization points.  As indicated in Figure 6, the Pareto 
points are the points where no gain of one metric is achieved without the loss of another metric.


















FIGURE 6: Example of Pareto Point Analysis 

	Following the analysis of the first Trade-off graph Pareto Points the six of the optimization 
solutions were eliminated resulting the in the following solution set:





LCD 1
LCD 2
LCD 3
CPU 1
CPU 2
CPU 3
KP 1
KP 2
KP 3
Option
Cost
Life
Weight
x1
x2
x3
y1
y2
y3
z1
z2
z3
2
1250
100
1650


1

1


1

3
2000
200
2000


1

1

1


6
1600
160
1750


1


1


1
8
1080
100
1900
1


1



1

10
1450
160
1800


1


1
1


11
1100
100
1800

1

1



1

Table 3:  Result of initial Pareto Analysis


7.  Second Pareto Plot Optimization
	Using the results remaining from the first analysis (see Table 3), additional graphs were 
constructed to further analyze the solutions.  The graphs are displayed in APPPENDIX H.  Analysis of 
these graphs indicated a wide range of possible solutions; however, typical Pareto points were not selected 
from these results in their entirety.  All of the points indicating life solutions of 100,000 uses were thrown 
out due to the availability of solutions ranging in the 160,000 to 200,000 life range.  This was done due to 
the prioritization of the metrics with life being first, followed by weight, then cost (see APPENDIX I).
	
Following the second analysis, the possible solutions were reduced to the following:





LC
D 1
LC
D 2
LC
D 3
CP
U 1
CP
U 2
CP
U 3
KP 
1
KP 
2
KP 
3
Option
Cost
Life
Weight
x1
x2
x3
y1
y2
y3
z1
z2
z3
3
2000
200
2000


1

1

1


6
1600
160
1750


1


1


1
10
1450
160
1800


1


1
1


Table 4:: Final solution options



8.  System Optimization
	Having narrowed the options to three remaining points, again graphs were constructed to assist in 
the final optimization of the system.  We will now examine each remaining option in detail.

COMPONENT 1: OPTION 3 ANALYSIS:

	Analysis of Option 3 shows that its component combination maximizes system life with a value of 
200,000 uses before failure.  However, this combination also maximizes cost by bringing the total system 
cost to $2,000 per HCU.  Additionally, this option maximizes the weight of the system at2000g. (See 
Figure 7).


FIGURE 7: Pareto Analysis of Option 6 










SOLUTION 2: OPTION 6 COMPONENTS

	Option 6 provides a relatively high value for life with 160,000 uses before failure.  Of the three 
remaining options it is the lowest weight, weighing 50g less than the next closest option.  However, it is the 
second highest costing option with a value of $1,600 per HCU system (see Figure 8).


FIGURE 8: Pareto Analysis of Option 3


















SOLUTION 3: OPTION 10 COMPONENTS

	Option 10 offers a relatively long product life of 160,000 uses before failure.  It does have the 
second highest weight, weighing 50g more than option 6.  However, it is the lowest costing option, offering 
$160 savings per HCU when compared to it's closest competitor  (see Figure 9).



FIGURE 9: Pareto analysis of Option 10
















D.  COMPONENT CONFIGURATION RECOMMENDATION:

	Final component selection requires quantifiable and subjective analysis of the trade-off data.  The 
chart below summarizes the trade-off results (Table 5).





LCD 3
CPU 2
CPU 3
KP 1
KP 3
ANALYSIS
Option
Cost
Life
Weight
x3
y2
y3
z1
z3

3
2000
200
2000
1
1

1

PRO:  Maximizes Life (by 40,000 
uses)                          
CON:  Increases WT (by 200g)                                                         
CON: Highest cost ($400/sys)
6
1600
160
1750
1

1

1
PRO:  Long Life (160,000 uses)                                      
PRO: Lowest WT of options by 
(50g)    
CON:  2d Highest Cost 
10
1450
160
1800
1

1
1

PRO:  Long Life (160,000 uses)                                      
PRO: Lowest cost of remaining 
options by ($150/sys)                                                       
CON:  Slightly higher WT than 
option 6 (50g) 
Table 5:  Final option comparison


 		Based on the data presented in Table 5, option 10 offers the optimum component configuration for 
the system.  Although this option does not maximize component life, it does offer a life expectancy of 
160,000 uses, which is acceptable when taking into account the extremely high cost and weight offered by 
the longest lasting configuration in option 3.  Option 10 is not the lightest configuration, outweighing 
option 6 by 50g.  This again is acceptable since the weight increase of 50g is relatively insignificant when 
considering cost gains.  Option 10 is the least costly option saving $150.00 per unit, when compared to 
option 6, which offers only slight gains by decreasing weight.  
Option 10 offers an acceptable life expectancy, with moderate weight at a low cost, making it the 
optimal solution.  This selection results in the following commercial component selection (Table 6):

COMPONENT
VENDOR
LCD
LCD:  LightView QVGA-0076
CPU
Mango PCI FP32C
KEYPAD
Mark SYS INC
Table 6: Vendor Selection Results



REFERENCES

Mark Austin: Lecture Notes for ENSE 621.  Institute for Systems Research, University of 
Maryland, 2002.

John Baras, Lecture Notes ENSE 622,  Institute for Systems Research, University of 
Maryland Spring 2003.

John Baras, Integrated Product and Process Design for Electromechanical Products, 
University of Maryland Institure for systems Research, 8 April, 1998.

Mingjia Dai, MP3 Player System Project Final Report, ENSE 623 Project, University of 
Maryland, 19 DEC, 2002 http://www.isr.umd.edu/Courses/ENSE622/index.html

FM 101-5.  STAFF ORGANIZATION AND OPERATIONS. U.S. Army Publishing, MAY 
1997.

Martin Fowler,  Scott, Kendell, UML Distilled, Second Edition. Addison-Wesley-
Longman, 2000.

Elizabeth Hull, Jackson, Ken, Dick, Jeremy, Requirements Engineering, Springer, 
London, 2002.

Adrian Marsh, Combat Marking System Design, ENSE 621 Project, University of 
Maryland, 19 DEC, 2002.

Vimal Mayank, Saraf, Deep, Modeling and Design of a Fast Food Restaurant, 
ENSE 623 Project, University of Maryland, 19 DEC, 2002 
http://www.isr.umd.edu/Courses/ENSE622/index.html

Practical UML a Hands On Introduction for Developers.  
www.togethersoft.com/services/practical_guides/umlonlinecourse/

    Adrian A. Marsh, Combat Marking System Design, ENSE 621 Project, DEC 02.
17

2