 Applied Architectures and Styles Chapter 11, Part 2 Service-Oriented Architectures and Web Services Architectures from Specific Domains Robotics Wireless Sensor Networks

Service-Oriented Architectures and Web Services  Supports business enterprises on the internet  Independence is fundamental  Conceptually part of the decentralized design space  Significant complexity

Service-Oriented Architectures and Web Services  Independent Components  Interface describing what operations they provide  Have their own thread of control  Can be described using WSDL  Web Services Description Language  XML  APIs written in languages such as Java, PHP, Ruby

Service-Oriented Architectures and Web Services  Connectors of SOA  Asynchronous Event Notification  Sends structured XML document to service provider  Variety of protocols: HTML, , etc.  No obligation for provider to return anything to requestor  Different from remote procedure calls  No object persistence  Absence of programming semantics

Service-Oriented Architectures and Web Services  Requires Architecture Description Language  Formality and complete semantics required  Similar to a scripting language  Other Architectural Styles and Concepts  Publish/Subscribe Mechanisms for discover of new services or service providers

Service-Oriented Architectures and Web Services  Significant Complexity  Interoperability across platforms  Open, decentralized system  Difficult to change services once announced  Compares to Java and changing a public method  Due to these issues, developers often choose REST instead of SOAP  REST – focuses on the exchange of representations/data  SOAP – focuses on the invocations of functions with arguments  Amazon provides both methods

Service-Oriented Architectures and Web Services

Architectures from Specific Domains  Robotics  Sense-Plan-Act  Subsumption  Three-Layer  Reuse-Oriented  Wireless Sensor Networks

Architectures from Specific Domains  Robotics Examples  Mobile tele-operated systems  Full or partial control of the robot’s systems is handled by a human operator  Industrial automation systems  Autonomous  Environment conforms to robot  Mobile autonomous robots  Autonomous  Navigate varying environments

Architectures from Specific Domains  Robotics Challenges  Physical platforms and devices  Unpredictable nature of the environment  Robotic Software Qualities  Robustness  Performance  Reusability  Adaptability

Architectures from Specific Domains  Sense-Plan-Act (SPA)  Unidirectional  Sense – Interface with the world  Plan – Behavior  Reconciles actual state with internal model  Internal model repeatedly updated in response to newly acquired sensor information  Objective it to keep the model consistent with actual environmental conditions

Architectures from Specific Domains  Sense-Plan-Act (SPA) – Drawbacks  Sensor information must be integrated (known as sensor-fusion) and incorporated into planning models at each step  These operations are very time consuming and typically cannot keep up with environmental changes  Therefore this model does not scale well as robot capabilities and goals expand

Architectures from Specific Domains  Subsumption  Inhibition and Suppression  No world model or plan  Independent components, each encapsulating specific behavior or skill  Reactive in nature  Excellent performance  Drawbacks  Lack of coherent architectural plan  Components inserted into data flow depending on task without position necessarily being related to the layer with in which they are located

 Three-Layer Architecture (3L)  Hybrid of SPA and Subsumption  Three Layers  Reactive  Sequencing  Planning  Challenges  Deciding how to separate functionality into layers  Reconciliation of world and behavior models Architectures from Specific Domains

 Reuse-Oriented  Component Reuse  Object Oriented  Middleware Frameworks  Few Frameworks  Player (  Orca (  Microsoft Robots Studio (  Lego Mindstorms (

Architectures from Specific Domains  Wireless Sensor Networks (WSN)  Sensor nets  Monitor the environment  Widely used  Medical systems  Navigation  Industrial automation  Low installation costs  Easy reconfiguration

Architectures from Specific Domains  WSN Challenges  Integration with legacy systems, other embedded devices, mobile networks  PDAs, cell phones, etc.  Typical Constraints  Power  Bandwidth  Range  Processing capacity  Falt-tolerance  Availability  Scalability

Architectures from Specific Domains  Bosch Research and Technology Center and University of Southern California  Three separate architectural styles to achieve all goals  Peer-to-peer  Publish-subscribe  Service-oriented