Some web resources, references, labs, and slides for the textbook Embedded System Design: A Unified Hardware/Software Introduction by Frank Vahid and Tony Givargis (John Wiley & Sons, 2002, ISBN 0471386782). The book's companion site is at esd.cs.ucr.edu.
What the book is about
Most embedded systems courses used to be split in two: software people learned assembly and C on a microcontroller, and hardware people learned logic design. This book treats the two as one design problem. Its central idea is that a function can be built either as software on a general-purpose processor or as a custom single-purpose processor in hardware, and that a designer should choose between them by looking at design metrics such as performance, power, size, cost and time to market.
The book works through that idea from the bottom up: logic and custom processors, general-purpose processors, standard peripherals, memory, interfacing, and then a full digital camera example that shows the hardware/software trade-offs on one product. Later chapters cover state machine and concurrent process models, control systems, IC technologies and design technology.
What is on the companion site
The site gives four kinds of material for each chapter. Each link below goes to the relevant page.
- Table of contents: the full chapter and section list, useful to see the scope before buying or assigning the book.
- Lecture slides: PowerPoint slides for all eleven chapters, a ready base for a course.
- Labs: instructor-oriented lab procedures, including a simulation tutorial, VHDL synthesis on an FPGA board, a microprocessor in VHDL and 8051 microcontroller labs.
- Web resources: per-chapter links, including bus specifications and the digital camera example source files.
- References and further reading: the reading list for each chapter.
The material dates from the early 2000s, so the boards and some tools are old. The concepts, especially the hardware/software trade-off thinking, have aged well.
Who it suits
- Undergraduates taking a first embedded systems course.
- Software engineers who want to understand when a function should move into hardware.
- Hardware engineers who want a structured view of processors, peripherals and buses from the system side.
- Instructors looking for slides and labs that follow a textbook chapter by chapter.
Deep dive: the numbers and details
This part goes through the book's structure and the main technical ideas in more detail. Skip it if the overview is all you need.
Chapter by chapter
| Chapter | Topic | Key ideas |
|---|---|---|
| 1 | Introduction | Design metrics (NRE cost, unit cost, time to market, performance); processor, IC and design technologies; the design productivity gap |
| 2 | Custom single-purpose processors: hardware | Combinational and sequential logic, FSMD (finite state machine with datapath) design, optimizing the program, FSMD, datapath and FSM |
| 3 | General-purpose processors: software | Datapath, control unit and memory; pipelining, superscalar and VLIW; instruction sets, interrupts, operating systems; ASIPs such as microcontrollers and DSPs |
| 4 | Standard single-purpose processors: peripherals | Timers, counters and watchdogs, UART, PWM, LCD and keypad controllers, stepper motor controllers, ADCs, real-time clocks |
| 5 | Memory | ROM, OTP, EPROM, EEPROM, flash, SRAM, DRAM; composing memories; cache mapping, replacement and write policies; DRAM variants; MMU |
| 6 | Interfacing | Port and bus-based I/O, memory-mapped I/O, interrupts, DMA, arbitration, multi-level buses; I2C, CAN, FireWire, USB, PCI, ARM bus; IrDA, Bluetooth, IEEE 802.11 |
| 7 | Digital camera example | Specification, then several implementations compared on the design metrics |
| 8 to 11 | Models, control, IC technology, design technology | State machine and concurrent process models, control systems, IC technologies, design technology |
The design metric trade-off, worked through
Chapter 1 frames the choice between technologies with non-recurring engineering (NRE) cost and unit cost. Total cost is NRE plus unit cost times volume, so the cost per product is:
cost per product = NRE / volume + unit cost
As an illustration with made-up numbers, compare an option with NRE of 2,000 and unit cost of 100 against one with NRE of 30,000 and unit cost of 2:
- At 100 units: 2,000 / 100 + 100 = 120 for the first, against 30,000 / 100 + 2 = 302 for the second.
- At 10,000 units: 2,000 / 10,000 + 100 = 100.2, against 30,000 / 10,000 + 2 = 5.
- The two are equal where 2,000 + 100V = 30,000 + 2V, which is V of about 286 units.
Below the crossover the low-NRE option, typically a programmable part, wins; above it the high-NRE option, typically a custom chip, wins. The same reasoning still drives the choice between microcontrollers, FPGAs and ASICs today.
Single-purpose versus general-purpose processors
A general-purpose processor runs any program, so it has a large, flexible datapath and a program memory. A single-purpose processor is built for one algorithm: its datapath holds only the units that algorithm needs and its controller is a fixed state machine. The book's FSMD method takes a sequential program, turns each statement into states, then builds the datapath (registers, functional units, multiplexers) and the controller from that state machine. The result is usually much faster and lower power for its one task, at the cost of design effort and flexibility.
How to use the labs
The lab index lists the procedures, schematics and materials for an instructor or teaching assistant. Several chapters have hands-on labs, while some chapters list none or "coming later". Expect to adapt the tools: the original labs use simulators, FPGA boards and 8051 compilers of their time, and a modern course would substitute current equivalents while keeping the exercises.


Your comments will be moderated before it appears here.