Computers as Components: Principles of Embedded Computing System Design Xiaoming JU

© 2005 ECNU SEIPrinciples of Embedded Computing System Design2 zTeacher: Ju xiaoming ( 琚小明 ) zOffice: B219 zTel: z zPPT download: seidown download 21

© 2005 ECNU SEIPrinciples of Embedded Computing System Design3 zTextbook: W. Wolf, Computers as Components: Principles of Embedded Computing System Design,  2001, Academic Press. zReference books: 1. Berger Arnold. 吕骏 译. 嵌入式系统设计. 北京：电子工业出版社， 王田苗. 嵌入式系统设计与实例开发－－基于 ARM 微处理器与 µC/OS-II 实时操作 系统（第 2 版）. 北京：清华大学出版社， zNetwork ：

© 2005 ECNU SEIPrinciples of Embedded Computing System Design4 outline zIntroduction to Embedded Systems zModels and Architectures for Embedded System Specification zSpecification Languages for Embedded System Design zA Specification Example: Telephone Answering Machine zEmbedded System Platform zSystem-Design Methodology

© 2005 ECNU SEIPrinciples of Embedded Computing System Design5 grading zAttendance and homework 10% zPaper Reading, Final Presentation and Report 30% zFinal Exam 60%

© 2005 ECNU SEIPrinciples of Embedded Computing System Design6 Introduction zWhat are embedded systems? zChallenges in embedded computing system design. zDesign methodologies.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design7 Definition (P.1) zEmbedded system: any device that includes a programmable computer but is not itself a general-purpose computer. zTake advantage of application characteristics to optimize the design: ydon’t need all the general-purpose bells and whistles.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design8 Embedding a computer CPU mem input output analog embedded computer

© 2005 ECNU SEIPrinciples of Embedded Computing System Design9 Examples zPersonal digital assistant (PDA). zPrinter. zCell phone. zAutomobile: engine, brakes, dash, etc. zTelevision. zHousehold appliances. zPC keyboard (scans keys).

© 2005 ECNU SEIPrinciples of Embedded Computing System Design10 Early history (P.2) zLate 1940’s: MIT Whirlwind computer was designed for real-time operations. yOriginally designed to control an aircraft simulator. zFirst microprocessor was Intel 4004 in early 1970’s. zHP-35 calculator used several chips to implement a microprocessor in 1972.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design11 Early history, cont’d. zAutomobiles used microprocessor-based engine controllers starting in 1970’s. yControl fuel/air mixture, engine timing, etc. yMultiple modes of operation: warm-up, cruise, hill climbing, etc. yProvides lower emissions, better fuel efficiency.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design12 Microprocessor varieties zMicrocontroller: includes I/O devices, on- board memory. zDigital signal processor (DSP): microprocessor optimized for digital signal processing. zTypical embedded word sizes: 8-bit, 16- bit, 32-bit.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design13 Application examples (P.2) zSimple control: front panel of microwave oven, etc. zCanon EOS 3 camera has three microprocessors. yOne of 32-bit RISC CPU runs autofocus and eye control systems. zAnalog TV: channel selection, etc. zDigital TV: programmable CPUs + hardwired logic.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design14 Automotive embedded systems zToday’s high-end automobile may have 100 microprocessors: y4-bit microcontroller checks seat belt; ymicrocontrollers run dashboard devices; y16/32-bit microprocessor controls engine.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design15 BMW 850i brake and stability control system (P.3) zAnti-lock brake system (ABS): pumps brakes to reduce skidding. zAutomatic stability control (ASC+T): controls engine to improve stability. zABS and ASC+T communicate. yABS was introduced first---needed to interface to existing ABS module.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design16 BMW 850i, cont’d. brake sensor brake sensor brake sensor brake sensor ABS hydraulic pump

© 2005 ECNU SEIPrinciples of Embedded Computing System Design17 Characteristics of embedded systems (P.3) zSophisticated functionality. zReal-time operation. zLow manufacturing cost. zLow power. zDesigned to tight deadlines by small teams.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design18 Functional complexity zOften have to run sophisticated algorithms or multiple algorithms. yCell phone, laser printer. zOften provide sophisticated user interfaces.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design19 Real-time operation zMust finish operations by deadlines. yHard real time: missing deadline causes failure. ySoft real time: missing deadline results in degraded performance. zMany systems are multi-rate: must handle operations at widely varying rates.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design20 Non-functional requirements zMany embedded systems are mass- market items that must have low manufacturing costs. yLimited memory, microprocessor power, etc. zPower consumption is critical in battery- powered devices. yExcessive power consumption increases system cost even in wall-powered devices.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design21 Design teams (P.4) zOften designed by a small team of designers. zOften must meet tight deadlines. y6 month market window is common. yCan’t miss back-to-school window for calculator.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design22 Why use microprocessors? (P.4) zAlternatives: field-programmable gate arrays (FPGAs), custom logic, etc. zMicroprocessors are often very efficient: can use same logic to perform many different functions. zMicroprocessors simplify the design of families of products.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design23 The performance paradox zMicroprocessors use much more logic to implement a function than does custom logic. zBut microprocessors are often at least as fast: yheavily pipelined; ylarge design teams; yaggressive VLSI technology.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design24 Power zCustom logic is a clear winner for low power devices. zModern microprocessors offer features to help control power consumption. zSoftware design techniques can help reduce power consumption.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design25 Challenges in embedded system design (P.5) zHow much hardware do we need? yHow big is the CPU? Memory? zHow do we meet our deadlines? yFaster hardware or cleverer software? zHow do we minimize power? yTurn off unnecessary logic? Reduce memory accesses?

© 2005 ECNU SEIPrinciples of Embedded Computing System Design26 Challenges, etc. zDoes it really work? yIs the specification correct? yDoes the implementation meet the spec? yHow do we test for real-time characteristics? yHow do we test on real data? zHow do we work on the system? yObservability, controllability? yWhat is our development platform?

© 2005 ECNU SEIPrinciples of Embedded Computing System Design27 Design methodologies (P.6) zA procedure for designing a system. zUnderstanding your methodology helps you ensure you didn’t skip anything. zCompilers, software engineering tools, computer-aided design (CAD) tools, etc., can be used to: yhelp automate methodology steps; ykeep track of the methodology itself.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design28 Design goals (P.7) zPerformance. yOverall speed, deadlines. zFunctionality and user interface. zManufacturing cost. zPower consumption. zOther requirements (physical size, etc.)

© 2005 ECNU SEIPrinciples of Embedded Computing System Design29 Levels of abstraction (P.7) requirements specification architecture component design system integration Top-down Bottom-up

© 2005 ECNU SEIPrinciples of Embedded Computing System Design30 Top-down vs. bottom-up zTop-down design: ystart from most abstract description; ywork to most detailed. zBottom-up design: ywork from small components to big system. zReal design uses both techniques.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design31 Stepwise refinement zAt each level of abstraction, we must: yanalyze the design to determine characteristics of the current state of the design; yrefine the design to add detail; yensure all design objects.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design32 Requirements (P.8) zPlain language description of what the user wants and expects to get. zMay be developed in several ways: ytalking directly to customers; ytalking to marketing representatives; yproviding prototypes to users for comment.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design33 Functional vs. non- functional requirements zFunctional requirements: youtput as a function of input. zNon-functional requirements: ytime required to compute output; ysize, weight, etc.; ypower consumption; yreliability; yetc.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design34 Our requirements form (P.9)

© 2005 ECNU SEIPrinciples of Embedded Computing System Design35 Example: GPS moving map requirements (P.10) zMoving map obtains position from GPS, paints map from local database. lat: 40 ° 13 ′ lon: 32 ° 19 ′ I-78 Scotch Road Display position Current position

© 2005 ECNU SEIPrinciples of Embedded Computing System Design36 GPS moving map needs zFunctionality: For automotive use. Show major roads and landmarks. zUser interface: At least 400 x 600 pixel screen. Three buttons max. Pop-up menu. zPerformance: Map should scroll smoothly. No more than 1 sec power-up. Lock onto GPS within 15 seconds. zCost: $500 street price = approx. $100 cost of goods sold.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design37 GPS moving map needs, cont’d. zPhysical size/weight: Should fit in hand. zPower consumption: Should run for 8 hours on four AA batteries.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design38 GPS moving map requirements form (P.10)

© 2005 ECNU SEIPrinciples of Embedded Computing System Design39 Specification (P.11) zA more precise description of the system: yshould not imply a particular architecture; yprovides input to the architecture design process. zMay include functional and non-functional elements. zMay be executable or may be in mathematical form for proofs.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design40 GPS specification zShould include: yWhat is received from GPS; ymap data; yuser interface; yoperations required to satisfy user requests; ybackground operations needed to keep the system running.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design41 Architecture design (P.11) zWhat major components go satisfying the specification? zHardware components: yCPUs, peripherals, etc. zSoftware components: ymajor programs and their operations. zMust take into account functional and non-functional specifications.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design42 GPS moving map block diagram (P.12) GPS receiver search engine renderer user interface database display

© 2005 ECNU SEIPrinciples of Embedded Computing System Design43 GPS moving map hardware architecture GPS receiver CPU panel I/O display frame buffer memory

© 2005 ECNU SEIPrinciples of Embedded Computing System Design44 GPS moving map software architecture position database search renderer timer user interface pixels

© 2005 ECNU SEIPrinciples of Embedded Computing System Design45 GPS

© 2005 ECNU SEIPrinciples of Embedded Computing System Design46 Designing hardware and software components (P.13) zMust spend time architecting the system before you start coding. zSome components are yready-made, ysome can be modified from existing designs, yothers must be designed from scratch.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design47 System integration (P.13) zPut together the components. yMany bugs appear only at this stage. zHave a plan for integrating components to uncover bugs quickly, test as much functionality as early as possible.

© 2005 ECNU SEIPrinciples of Embedded Computing System Design48 Summary zEmbedded computers are all around us. yMany systems have complex embedded hardware and software. zEmbedded systems pose many design challenges: design time, deadlines, power, etc. zDesign methodologies help us manage the design process.