Lecture notes Embedded Systems

distributed embedded systems design middleware and resources and what are distributed embedded systems | pdf free download
Dr.JamesSmith Profile Pic
Dr.JamesSmith,France,Professional
Published Date:11-07-2017
Your Website URL(Optional)
Comment
Distributed Embedded Systems: Distributed Embedded Systems: An Introduction An Introduction Julián Proenza. University of the Balearic Islands. SPAIN Luís Almeida. University of Aveiro. PORTUGALJulián Proenza. UIB. Nov 2008 Luís Almeida. UA. The aim of this presentation • Introduce the concept of Distributed Embedded System • Discuss the role that the communication subsystem plays in this kind of systems • Identify the main problems to address when designing a Distributed Embedded System • Present the communication features specifically adapted to control applications NOTE: The presentation is built on the assumption that the audience is familiar with the basic concepts of data communication networks 2Julián Proenza. UIB. Nov 2008 Luís Almeida. UA. Presentation Outline 1. Distributed Embedded Systems 1. From centralization to distribution 2. Definition and advantages 2. The Communication Subsystem’s role 1. Two sources of problems: delays and unreliability 2. Two disciplines: Real-Time and Dependability 3. Networks adapted to control applications 1. The communication subsystem 2. Requirements related to traffic 3. Control networks and the OSI stack: structure and features 3Julián Proenza. UIB. Nov 2008 Luís Almeida. UA. Presentation Outline 1. Distributed Embedded Systems 1. From centralization to distribution 2. Definition and advantages 2. The Communication Subsystem’s role 1. Two sources of problems: delays and unreliability 2. Two disciplines: Real-Time and Dependability 3. Networks adapted to control applications 1. The communication subsystem 2. Requirements related to traffic 3. Control networks and the OSI stack: structure and features 4Julián Proenza. UIB. Nov 2008 Luís Almeida. UA. Centralized Embedded Systems • Control application • The controller (computer or PLC) reads some sensors and sends outputs to some actuators System under control A S Controller • The system under control (plant) requires a specific time response: the controller has to be real-time • The architecture is centralized: all sensors and actuators are connected through dedicated links. 5Julián Proenza. UIB. Nov 2008 Luís Almeida. UA. Decentralized Embedded Systems • In some applications, cabling started to be a problem: too costly (buildings) and too heavy (cars) • Decentralized solutions were proposed System under control S S A A A A S S Controller • In this kinds of systems, cabling is cheaper and simpler. Moreover addition of new s/a is easier • In fact, for sharing the comm medium some comm controller had to be added to each s/a… 6Julián Proenza. UIB. Nov 2008 Luís Almeida. UA. Distributed Embedded Systems • … so it was natural to add some extra hardware and provide several controllers each one performing part of the control tasks System under control S A A A S S … NODE 1 NODE 2 NODE N • The result is a distributed control system • They are widespread in numberless applications: factories, buildings, vehicles, etc. 7Julián Proenza. UIB. Nov 2008 Luís Almeida. UA. Presentation Outline 1. Distributed Embedded Systems 1. From centralization to distribution 2. Definition and advantages 2. The Communication Subsystem’s role 1. Two sources of problems: delays and unreliability 2. Two disciplines: Real-Time and Dependability 3. Networks adapted to control applications 1. The communication subsystem 2. Requirements related to traffic 3. Control networks and the OSI stack: structure and features 8Julián Proenza. UIB. Nov 2008 Luís Almeida. UA. But what is a Distributed System? • An accurate definition: – Multiple computers interconnected by a network that share some common state and cooperate to achieve some common goal. Essentially what was expressed by Schroeder in Mul93 • Some remarkable definitions by Leslie Lamport: – “A system is distributed if the message transmission delay is not negligible compared to the time between events in a single process.” Lam78b. – “A distributed system is the one that prevents you from working because of the failure of a machine that you had never heard of.” Leslie Lamport • Despite the differences between computing and control systems, it is clear we will have to solve some problems. 9Julián Proenza. UIB. Nov 2008 Luís Almeida. UA. Advantages of distribution • Processing closer to data source / sink – Intelligent sensors and actuators • Dependability – Independence of failures among nodes – Error-containment within nodes • Composability – System composition by integrating subsystems • Scalability – Easy addition of new nodes with new or replicated functionality • Maintainability – Modularity and easy node replacement – Simplification of the cabling 10Julián Proenza. UIB. Nov 2008 Luís Almeida. UA. Presentation Outline 1. Distributed Embedded Systems 1. From centralization to distribution 2. Definition and advantages 2. The Communication Subsystem’s role 1. Two sources of problems: delays and unreliability 2. Two disciplines: Real-Time and Dependability 3. Networks adapted to control applications 1. The communication subsystem 2. Requirements related to traffic 3. Control networks and the OSI stack: structure and features 11Julián Proenza. UIB. Nov 2008 Luís Almeida. UA. The communication subsystem’s role • The communication subsystem is the backbone • Lamport’s definitions pointed at two key problems: – Communication introduces delay • More delays to consider in order for the control to be timely • Inexact global state (In a system in which the events occur faster than the communications among computers, it is impossible to obtain an exact global state of the system from the information available at any computer in any circumstance). – Communication is unreliable (compared with the intra comms in a computer) and is a means for error propagation • A network is more prone to errors • If the network fails, the whole system fails • Errors in one computer may propagate to the other computers • This subsystem has to be adapted to control 12Julián Proenza. UIB. Nov 2008 Luís Almeida. UA. Overcoming the two key problems • In order to solve the two key problems of extra delays and unreliability it is necessary to use concepts and techniques related to two disciplines: – Real-Time Systems and – Dependable Systems • It is possible to provide a non-exhaustive list of the specific issues that each discipline has to deal with • Aspects to take into account: – The set of specific issues to solve depends on the application – The properties achieved can/should be extended to the system as a whole 13Julián Proenza. UIB. Nov 2008 Luís Almeida. UA. Issues solved by each discipline • Issues solved with Real-Time techniques – Jeopardized system timeliness: Provide guarantees that the network is going to exchange the messages on time 14Julián Proenza. UIB. Nov 2008 Luís Almeida. UA. Issues solved by each discipline • Issues solved with Real-Time techniques – Jeopardized system timeliness: Provide guarantees that the network is going to exchange the messages on time • Issues solved with Dependability techniques – Hostile environments (EMI, vibrations, dust…): Robustness 15Julián Proenza. UIB. Nov 2008 Luís Almeida. UA. Issues solved by each discipline • Issues solved with Real-Time techniques – Jeopardized system timeliness: Provide guarantees that the network is going to exchange the messages on time • Issues solved with Dependability techniques – Hostile environments (EMI, vibrations, dust…): Robustness – Single point of failure: … NODE 1 NODE 2 NODE N Bus 16Julián Proenza. UIB. Nov 2008 Luís Almeida. UA. Issues solved by each discipline • Issues solved with Real-Time techniques – Jeopardized system timeliness: Provide guarantees that the network is going to exchange the messages on time • Issues solved with Dependability techniques – Hostile environments (EMI, vibrations, dust…): Robustness – Single point of failure: Add redundant channels or ensure high reliability … NODE 1 NODE 2 NODE N Bus (replicated) 17Julián Proenza. UIB. Nov 2008 Luís Almeida. UA. Issues solved by each discipline • Issues solved with Real-Time techniques – Jeopardized system timeliness: Provide guarantees that the network is going to exchange the messages on time • Issues solved with Dependability techniques – Hostile environments (EMI, vibrations, dust…): Robustness – Single point of failure: Add redundant channels or ensure high reliability – Error propagation … NODE 1 NODE 2 NODE N Bus 18Julián Proenza. UIB. Nov 2008 Luís Almeida. UA. Issues solved by each discipline • Issues solved with Real-Time techniques – Jeopardized system timeliness: Provide guarantees that the network is going to exchange the messages on time • Issues solved with Dependability techniques – Hostile environments (EMI, vibrations, dust…): Robustness – Single point of failure: Add redundant channels or ensure high reliability – Error propagation (even if redundancy is present): Ensure error containment … NODE 1 NODE 2 NODE N Bus (replicated) 19Julián Proenza. UIB. Nov 2008 Luís Almeida. UA. Issues solved by each discipline • Issues solved with Real-Time techniques – Jeopardized system timeliness: Provide guarantees that the network is going to exchange the messages on time • Issues solved with Dependability techniques – Hostile environments (EMI, vibrations, dust…): Robustness – Single point of failure: Add redundant channels or ensure high reliability – Error propagation (even if redundancy is present): Ensure error containment – Inconsistent global state: Enforce consistency • Consistent comms are a means for, among others, replica determinism; and replica determinism is a particular case of consistent view of system state. • Redundant channels may also cause inconsistencies: Need for consistent management of redundancy 20

Advise: Why You Wasting Money in Costly SEO Tools, Use World's Best Free SEO Tool Ubersuggest.