Search This Blog

Tuesday, September 27, 2011

11.2.3 Behavior Diagrams

(Apologies for the formatting problems in this posting. The editor is scrambling the input formats and won't change.)

Three of the eight behavior diagrams are discussed with examples in this section.

Activity Diagram - An activity diagram shows the flow from activity to activity within a system.  An activity shows a set of activities, the sequential of branching flow from activity to activity, and objects that act and are acted upon.  Activity digrams are used to illustrate the dynamic view of a system and are important in modeling the function of a system.  Activity digrams essentially are flow charts and emphasize the flow of control among objects.  Figure 11-4, shows a possible activity diagram for Taking a Picture in the digital camera system.

Figure 11-4 Example Activity Diagram for a Digital Camera System

Activity diagrams can also include forks to describe conditions and parallel activities.   A fork is used when multiple activities are occurring at the same time.  As shown on the diagram , all branches at some point are followed by a merge to indicate the end of the conditional behavior started by that branch. 

Activity diagrams should be used in conjunction with other modeling techniques such as sequence diagrams and state machine diagrams.  The main reason to use activity diagrams is to model the workflow behind the system being designed.  Activity Diagrams are also useful for: analyzing a use case by describing what actions need to take place and when they should occur;  describing a complicated sequential algorithm;  and modeling applications with parallel processes11-4.  Software engineers have also found them to be useful for scientific software development.

Sequence Diagram - A sequence diagram is an interaction diagram that emphasizes the time ordering of messages.  A sequence diagram shows the objects and the messages that go between those objects at a particular instance of time.  They are used to illustrate the dynamic view of a system.  An example of a possible sequence diagram for the digital camera system is hown in Figure 11-5.


Figure 11-5 Example Sequence Diagram for a Digital Camera System
As shown in the diagram, an object has veritical lines called lifelines which represent the existence of an object over a period of time.  The stick figure represents an actor, the Picture Taker.  The messages become method calls when translated into executable software.  The focus of control represents the period of time during which an object is performing an action directly or through a subordinate operation.  Other graphic elements used in a sequence diagram can be found in the UML specification at http://www.omg.org/technology/documents/modeling_spec_catalog.htm#UML
Sequence diagrams can be used at all levels of defining a software system – Use Cases down to a detailed software function.  Although sequence diagrams are typically used to describe object-oriented software systems, they are also extremely useful as system engineering tools to design system architectures.

State Machine Diagram - A state machine is a behavior that specifies the sequence of states an object goes through during its lifetime in response to events, together with its response to those events.  A state machine diagram shows the state machine, consisting of states, transitions, events, and activities.  These diagrams are used to illustrate the dynamic view of a system.  These are important when modeling the behavior of an interface, class, or collaboration and are useful when modeling reactive systems.  Figure 11-6 shows a state machine diagram for the Shutter Object.  When the button on the camera is pushed to take a picture, the Shutter object moves from the Opened state to the Closed state.  Once the button is released, the Shutter Object moves back into the opened state. 

  
Figure 11-6 Example State Machine Diagram for a Digital Camera System
Note that a transition from one state to another could be a signal, an event, a change in some condition, or the passage of time.  A state machine diagram may have a transition back to its own state.  At its simplest, an object within a software system has two states, idle or running.  When defining a software system, state diagrams are very beneficial for defining complex objects and how they behave.
Use Case Diagram – A Use Case Diagram is a diagram that shows a set of use cases and actors and their relationships.  A Use Case is a description of a system’s behavior as it responds to a request that originates from outside of that system.  It provides context to what is within a system and what interacts with a system and defines the behavior of the system when it receives external stimuli (i.e. the goals of the system).   A system is made up of multiple Use Cases to define the behavior of the overall system.  Use Cases were first defined in 1987 at the Proceedings of OOPSLA11-5 by Ivar Jacobson for use in Software Engineering to define functional requirements.  Ivar Jacobson, et al, later published a book titled Object-Oriented Software Engineering: A Use Case Driven Approach in 199211-6 based on his experiences with Use Cases while working on large telecommunications systems.  Figure 11-7 shows a Use Case Diagram for the simple Digital Camera System.


Figure 11-7 Example Use Case Diagram for a Digital Camera System
The Picture Taker and the Printer are outside the boundary of the system being defined and are called Actors.  Take Digital Picture, Print Picture, and Delete Picture are all use cases within the system (i.e. functions the Digital Camera performs).  When developing Use Cases, all Actors are described as “nouns” while the Use Cases are “verb phrases” that describe an action the system performs.  The arrow on the association line indicates initiation of interaction.  If either the Use Case or the Actor can initiate interaction, then there is no arrow on the relationship line.  There can be two types of actors depicted on a Use Case diagram:  Primary Actors which benefit from the Use Case (arrow from Actor to Use Case) and secondary actors which participate in the system but don’t get benefit (arrow from Use Case to Actor).
Use Case modeling not only includes the model, but also written text to describe how the system behaves when the external Actor stimulates the system.  Typically a Use Case template is defined so all needed information is captured on how the system behaves.  One example of a Use Case template for capturing the behavior of a system is shown in Figure 11-8.
 Figure 11-8 Example Use Case Template
When starting Use Cases, it’s advisable to keep the description at the top-level at first and add details as the Use Case is better understood. Typically a Use Case is started with the normal flow of events; alternate and exception paths are added if needed.  One of the strengths and a significant benefit of Use Cases is that it makes engineers think about how they want the system to behave under abnormal or failure conditions. 

11-5 Object-Oriented Development in an Industrial Environment, Ivar Jacobson,  Proceedings of OOPSLA´87, SIGPLAN Notices, Vol. 22, No. 12, pages 183-191, 1987
11-6 Object-Oriented Software Engineering: A Use Case Driven Approach, by Ivar Jacobson, Magnus Christerson, Patrik Jonsson, and Gunnar Ă–vergaard, Addison-Wesley, Wokingham, England, 1992. 

Tuesday, September 20, 2011

11.2 Introduction to the Unified Modeling LanguageTM(UML®)1

The Unified Modeling Language is a language used to specify, visualize, and document models of software systems, including their structure and design.  The language was formed by Grady Booch, Ivar Jacobson, and James Rumbaugh11-3 after a critical mass of ideas started forming in the 1990s from their individual work in object-oriented methods.  Collectively they defined UML for three reasons:

1.       Their methods were evolving toward each other independently
2.       To provide stability to the object-oriented modeling language
3.       To provide improvements to the language  
UML 1.0 was offered by the Object Management Group (OMG) for standardization in January 1997 with the final acceptance of version 1.1 in September 1997.  The OMG maintains the UML specification with the latest version found on the OMG website at http://www.uml.org/
In March 2003, the OMG along with the International Council on Systems Engineering (INCOSE) developed a Request for Proposal (RFP) for UML for Systems Engineering.  The RFP specified the requirements for extending UML to support the needs of the systems engineering community.  SysML™[1] is a general purpose graphical modeling language for specifying, analyzing, designing and verifying complex systems.  The system may include hardware, software, information, personnel, procedures, and facilities.  SysML represents a subset of UML.   The SysML specification was developed with the OMG announcing the adoption of the OMG SysML in July 2006 and the availability of the specification in September 2007.  The latest version of the SysML specification can be found at the OMG website at http://www.omg.org/spec/SysML/

11.2.1 Types of Diagrams in the UML- UML 2.4 See (http://www.omg.org/spec/UML/2.4defines fourteen types of diagrams, divided into two categories:

1.       Structure Diagrams, which include:
a.       Class Diagram
b.      Package Diagram (diagram same for SysML)
c.       Object Diagram
d.      Component Diagram
e.      Composite Structure Diagram
f.        Deployment Diagram
2.       Behavior Diagrams, which include:
a.       Activity Diagram
b.      Sequence Diagram (diagram same for SysML)
c.       Communication Diagram
d.      Interaction Overview Diagram
e.      Timing Diagram (optional)
f.        Interaction Tables (optional)
g.       State Machine Diagram (diagram same for SysML)
h.       Use Case Diagram (diagram same for SysML)

Structure diagrams are used to visualize, specify, construct and document the static aspects of a system i.e. no time element).  Behavior diagrams are used to visualize, specify, construct and document the dynamic aspects of a system (i.e. time is considered).
Not all diagrams in each category need to be used when defining a software system, just what is needed to communicate how the software system works; how it is built structurally, how it behaves, and what it interacts with.  Without both category representations, a software system is not fully defined and can lead to issues when the software system is built.  For example, without understanding the timing requirements for a real-time embedded system, software designers may define the software structure to meet all functional requirements of the system, but find out during testing that it does not meet the timing requirements. The software system would need to be re-architected to meet both the functional and timing requirements. 
The next sections provide simple examples of some of the UML diagram types including common diagrams with SysML.  These are simple examples based on the Digital Camera System first introduced in Chapter 6.  Note that the examples are not meant to represent a complete Digital Camera System or to be 100% accurate.  They are just representative of the UML diagrams to help understand the basic structure and information they are conveying.  There are many books that provide more detailed instructions on how to create these diagrams and on what they are used for when modeling a software system.  There are also many Web Sites that provide examples of each diagram type.

11.2.2 Structure Diagrams- Examples of Class Diagrams, Package Diagrams and Deployment Diagrams are Structure Diagrams shown and discussed in this section.

Class Diagram - A class diagram is used to show a set of classes, interfaces, and collaborations and their relationships.  Class diagrams provide a static view of the system.  Using the Simple Digital Camera System, Figure 11-1 illustrates a possible class diagram based on Take a Digital Picture Use Case .





Figure 11-1 Class Diagram Example for a Simple Digital Camera System
Shutter, LCD, and Storage are all classes within the system that interact with each other and represent real world items that make-up a Digital Camera.  Each class contains attributes and operations.  Attributes are a named property of a class that describes a range of values that instances of the property may hold; using software terms, a data structure or a variable.  An attribute is some piece of information the class needs in order to perform its function.  A class may have any number of attributes or none at all.  The class Storage has an attribute of imageCount to keep track of the number of images stored and it has three operations indicated, storeImage, incrementImageCount, and isStorageAvailable. 
When first identifying classes within a system start with the requirements specification or Use Cases and identify nouns.  Alternatively, start with the real world items that make-up the system.  All of these are candidates for possible classes.  As more is understood about the system, the class diagrams will evolve until the final class diagram for the system is achieved.  In modeling a software system, the classes tend to become more abstract and may not have real world counterparts.
The example above does not contain all possible graphic elements that are possible in a class diagram.  For more information, see the UML specification at: http://www.omg.org/technology/documents/modeling_spec_catalog.htm#UML

Package Diagram- A package is a general-purpose mechanism for organizing elements into groups.  Or in other words, it indicates a subsystem or block of functionality within a system and is used to simplify the view of a system in order to understand it better.  Figure 11-2 shows a possible package diagram for the Digital Camera System.





Figure 11-2 Package Diagram Example for a Simple Digital Camera System
The package diagram also shows the interactions or relationships between each package.  Figure 11-2 illustrates the Camera Controller Package as a depdency on both the Image Acquisition Package and the Storage Device Package.  Since packages are a way of abstracting or simplfying the view of a system, each package may contain more packages, class diagrams, use cases, components,  etc. until the view is at the lowest level.  Well-designed packages group like elements that tend to change together.  Well-structured packages are loosely coupled and highly cohesive (See Section 6.6.3 for definitions of coupling and cohesion).  The graphic elements used in class diagrams to show relationships are also used in package diagrams.  Package diagrams are used to show different views of a system’s architecture.    

Deployment Diagram - A deployment diagram shows the configuration of run time processing nodes and the components that live on each node.  They are used to illustrate the static deployment view of a software system onto the hardware on which the system executes.  Names of the nodes should represent the vocabulary of the hardware in the domain of the system being developed.  Nodes represent a physical element and represent a computational resource having at least some memory and processing capability.  Figure 11-3 shows a deployment diagram for the Digital Camera System assuming there are three processors in the system.


Figure 11-3 Example Deployment Diagram for a Digital Camera System
For real-time embedded software with only one processor, the deployment diagrams look like context diagrams from the system engineering world.  They show the devices with which the processor board is interacting using what are called stereotypes to identify the hardware components.  Client/Server and Distributed software systems have more processors on their deployment diagrams and provide the designers with an understanding of where each software component resides on the physical hardware.
The relationships between nodes are typically an association that represents a physical connection (i.e. Ethernet, cPCI bus).   For more information on possible types of relationships for deployment diagrams, see the UML specification at: http://www.omg.org/technology/documents/modeling_spec_catalog.htm#UML



[1] Systems Modeling Language and SysML are either registered trademarks or trademarks of Object Management Group, Inc. in the United States and/or other countries.

Wednesday, September 14, 2011

11 Introduction to Model Based Systems Engineering



11.0 Introduction
The advantages of using labeled graphical models, diagrams, tables of data and similar non prose descriptions compared to natural language or prose descriptions have been discussed several times. Now we make a distinction between two types of models. One type is, as stated, a non-prose description of something. The second type is analysis models; either static models that predict performance or dynamic models referred to as simulations. Static analysis models may be strictly analytical or may be machine readable and executable. Modern simulations are typically machine readable and executable. This is an arbitrary distinction as the DoD defines a model as a physical, mathematical, or otherwise logical representation of a system, entity, phenomenon, or process. (DoD 5000.59 -M 1998)
In Chapter 5 it was stated that PBSE is model based but includes prose documents as well. The models used in PBSE can be either the first type or the second type. Now we want to introduce a different approach to using models for systems engineering. This approach is called Model Based System Engineering (MBSE) and it strives to accomplish system engineering with models that are machine readable, executable or operative. An INCOSE paper11-1 defines MBSE as an approach to engineering that uses models as an integral part of the technical baseline that includes the requirements, analysis, design, implementation, and verification of a capability, system, and/or product throughout the acquisition life cycle.
This chapter is an introduction to MBSE; no attempt is made to review or even summarize the extensive literature on MBSE. MBSE is rapidly evolving, facilitated both by development of commercial tools and by an INCOSE effort to extend the maturity and capability of MBSE over the decade from 2010 to 2020. Whereas we attempt to describe how MBSE offers benefits compared to traditional prose based systems engineering it isn’t claimed that pure MBSE is superior or inferior to methodologies that mix MBSE, PBSE and traditional methods. The intent is to provide an introduction that enables readers to assess how MBSE can be beneficial to their work and to point the way toward further study.
Traditional systems engineering is a mix of prose based material, typically requirements and plans, and models such as functional diagrams, physical diagrams and mode diagrams. Eventually design documentation ends in drawings, which are models. MBSE can be thought of as replacing the prose documents that define or describe a system, such as requirements documents, with models. We are not concerned as much with plans although plans like test plans are greatly improved by including many diagrams, photos and other models with a minimum of prose.
To some it may seem difficult to replace requirements documents with models. However, QFD can be stand-alone systems engineering process and QFD is a type of MBSE. Although it does not attempt to heavily employ machine readable and executable models, QFD is an example of defining requirements in the form of models. Another way to think about requirements is that mathematically requirements are graphs and can therefore be represented by models. A third way to think about requirements as models is as tree structures. Each requirement may have parent requirements and daughter requirements and just as no leaf of a tree can exist without connection to twigs, twigs to limbs, and limbs to the trunk no requirement can stand alone. Trees can be represented by diagrams so requirements can all be represented in a diagram.
Throughout this book there is an emphasis on representing design information as models in order to reduce ambiguity and the likelihood of misinterpretation of text based design information. There is also an emphasis on using analysis models and simulations as much as possible throughout the life cycle of a system development. The use of models and simulations improves functional analysis, design quality, system testing and system maintenance. Think of MBSE as combining these two principles; then it becomes clear why MBSE is desirable. Another way to look at traditional systems engineering vs. MBSE is for traditional systems engineering engineers write documents and then models are developed from the documents. In MBSE the approach is to model what is to be built from the beginning.
Model based design has been standard practice for many engineering specialties since the 1980s. Structural analysis, thermal analysis, electrical circuit analysis, optical design analysis and aerodynamics are a few examples of the use of Computer Aided Design (CAD) or model based design analysis. It is systems engineering that been slow to transition from non-model based methods, with the exception of performance modeling and simulation. To achieve the benefits of MBSE systems engineers need to embrace requirements diagrams, Use Case analysis and other MBSE tools along with performance modeling and simulation.
11.1 Definitions of Models As Applied to MBSE
Models have been referred to throughout this material without providing a formal definition or defining the types of models typically used in systems engineering. Formally, a model is a representation of something, as described in the DoD definition given above. For our purposes a model is a representation of a design element of a system. Types of models of interest to MBSE include11-2:
Schematic Models: A chart or diagram showing relationships, structure or time sequencing of objects. For MBSE schematic models should have a machine-readable representation. Examples include FFBDs, interface diagrams and network diagrams.
Performance Model: An executable representation that provides outputs of design elements in response to inputs. If the outputs are dynamic then the model is called a simulation.
Design Model: A machine interpretable version of the detailed design of a design element. Design models are usually represented by CAD drawings, VHDL, C, etc.
Physical model: A physical representation that is used to experimentally provide outputs in response to inputs. A breadboard or brass board circuit is an example.
Achieving machine readable and executable models means that the models must be developed using software. Useful languages used by software and systems engineers for such models are the Unified Modeling LanguageTM (UML®) and its derivative SysML™. A brief introduction to these languages is presented here along with references for further study.