y nl t O rin e P gl in r S Fo

Embedded Design / January 2007 ©2007 OpenSystems Publishing Not for distribution y nl t O rin e P gl in r S Fo

RSC#  @ www.embedded-computing.com/rsc Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution y nl t O rin e P gl in r S Fo

RSC#  @ www.embedded-computing.com/rsc Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution www.embedded-computing.com Volume 5 • Number 1 JAN U A R Y 2 0 0 7

COLUMNS FEATURES 7 Editor’s Foreword SPECIAL: Embedded industry organizations overview Processor, software improvements top trends for 2007 14 Strength in numbers: technology consortia wield collective power By Jerry Gipper By Jerry Gipper and Don Dingee 8 Embedded Perspective More batteries, Mr. Taggart? HARDWARE TECHNOLOGY: Technologyy trends in By Don Dingee embeddedl computing 10 Embedded Technology in Europe n Embedded analysis from health to travel to dirt 22 Open source opens doors for improved development By Dr. Inder M. Singh,O LynuxWorks By Hermann Strass t 11 Eclipse Perspective and News 25 The advent of COTS use in embedded systems Embedding human knowledge in software Byi Andrewn Foster, PrismTech By J. Simon Tack, Sean Melody, and Dr. Todd Griffith 29 Keyr industry movements driving designs in 2007 P By Wade Clowes, RadiSys DEPARTMENTS e SOFTWARE TECHNOLOGY: Enterprise software in 32 Editor’s Choice Products l embedded applications By Don Dingee g n 33 COTS databases for embedded systems i By Steve Graves, McObject S 38 Enterprise applications go embedded or By Collin Bruce, Hitachi America F E-CASTS Mapping Multi-function Radar Algorithms Across a VPX-REDI System January 25, 2 p.m. EST www.opensystems-publishing.com/ecast E-LETTER January: www.embedded-computing.com/eletter The rise of for the handset By Paxton Cooper, MontaVista Software COVER The intelligent embedded devices of today require enhanced databases capable of managing grow- EVENTS ing volumes of more International Consumer Electronics Show complex data, such as the OpenSystems January 8-11 • Las Vegas, NV ever-expanding directory of www.cesweb.org technology consortia. Publishing™ Bus and Board Conference OpenSystems January 15-16 • Hyatt Regency, Long Beach, CA Publishing™ www.busandboard.com WEB RESOURCES Subscribe to the magazine or E-letter: Published by: OpenSystems www.opensystems-publishing.com/subscriptions OpenSystems Industry news: Publishing™Publishing™ Read: www.embedded-computing.com/news © 2007 Embedded Computing Design Submit: www.opensystems-publishing.com/news/submit All registered brands and trademarks within Embedded Computing Design Submit new products: are property of their respective owners. www.opensystems-publishing.com/vendors/submissions/np

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution y nl t O rin e P gl in r S Fo

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution A n O p e n S y s t e m s P u b l i c a t i o n

Embedded and Test & Analysis Group n Embedded Computing Design n Embedded Computing Design E-letter n Embedded Computing Design Resource Guide n Industrial Embedded Systems n Industrial Embedded Systems E-letter n Industrial Embedded Systems Resource Guide n PXI, Test & Technology n PXI, Test & Technology E-letter n PXI, Test & Technology Resource Guide Editorial Director Jerry Gipper [email protected] Contributing Editor Don Dingee Technical Editor Chad Lumsden [email protected] Associate Editor Jennifer Hesse [email protected] European RepresentativeO Hermann Strass t [email protected] Special Projects Editor Bob Stasonis inSenior Designer Joann Toth rSenior Web Developer Konrad Witte P Graphic Specialist David Diomede OpenSystems Circulation/Office Publishing™ Manager Phyllis Thompson le [email protected]

RSC#  @ www.embedded-computing.com/rsgc OpenSystems in Publishing™OpenSystems Publishing Editorial/Production office: S 16872 E. Avenue of the Fountains, Ste 203, Fountain Hills, AZ 85268 Tel: 480-967-5581 n Fax: 480-837-6466 Page/RSC# Advertiser – Product descriptionr Website: www.opensystems-publishing.com o Publishers John Black, Michael Hopper, Wayne Kristoff 23 Advantech Corporation – Stackable SBCs for Total Solutions F Vice President Editorial Rosemary Kristoff 13 Annapolis Micro Systems, Inc. – High Performance Signal Processing 31 Avionics – 5th Year Exhibition and Conferences Communications Group 24 Dataforth Corporation – Embedded Signal Conditioning Editorial Director Joe Pavlat 39 Diamond Systems Corporation – SBCs with Data Acquisition Built-in Assistant Managing Editor Anne Fisher Senior Editor (columns) Terri Thorson 30 Embedded Planet – AMCC PowerPC 405 Processor Technology Editor Curt Schwaderer 37 Embedded World – Exhibition and Conference Associate Editor Jennifer Hesse European Representative Hermann Strass 19 Emerson Network Power – MicroTCA the right solution 40 Intel – Roadmap to your future Military & Aerospace Group 5 Micro/sys, Inc. – CPU boards Group Editorial Director Chris Ciufo 21 Microsoft – Windows Embedded Assistant Editor Sharon Schnakenburg Senior Editor (columns) Terri Thorson 36 North Atlantic Industries – Multifunction I/O Cards European Representative Hermann Strass 28 Semico – 2007 Tenth Annual Summit 27 Sundance – Parallel Application from Rapid Simulation ISSN: Print 1542-6408, Online 1542-6459 9 Technobox – PMCs, Adapters, and Tools Embedded Computing Design is published 8 times a year by OpenSystems 15 Technologic Systems – 200 MHz CPU Publishing LLC., 30233 Jefferson Ave., St. Clair Shores, MI 48082. 6 TEWS Technologies LLC – Embedded I/O Solutions Subscriptions are free to persons interested in the design or promotion of embed- ded computing systems. For others inside the US and Canada, subscriptions are 16 Tri-M Systems Inc. – 100 MHz PC/104 Module $56/year. For 1st class delivery outside the US and Canada, subscriptions are $80/year (advance payment in US funds required). 34 Tri-M Systems Inc. – Serial Port Module, Fanless Embedded Controller Canada: Publication agreement number 40048627 2 VersaLogic Corp. – Product customization Return address: WDS, Station A, PO Box 54, Windsor, ON N9A 615 3 WinSystems, Inc. – EPIC Solutions for Real World Problems POSTMASTER: Send address changes to Embedded Computing Design 16872 E. Avenue of the Fountains, Ste 203, Fountain Hills, AZ 85268

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution EDITOR’S FOREWORD Processor, software improvements top Advertising/Business Office 30233 JeffersonOpenSystem Avenue s trends for 2007 St. Clair Shores, MI 48082 ™ Jerry Gipper Tel: 586-415-6500Publishing n Fax: 586-415-4882 he beginning of the year is always a great time to talk about trends. I think I can sum Vice President Marketing & Sales up the latest developments in two trends that appear to be having the biggest impact Patrick Hopper on embedded computing designs: [email protected] T Business Manager 1. Processors are more configurable Karen Layman 2. Software is easier to develop and test

Sales Group What do I mean by these statements? I have written several times about mass cus- tomization and what that means for the embedded computing market. Designers want Dennis Doyle components that can be quickly and effectively customized for their application, yet Senior Account Manager y they also want to take advantage of the price points and commonl architectures of mass- [email protected] produced parts. Processors are the most advanced in availabilityn for configurable varia- Tom Varcie tions. Technology improvements make it possible to get a customized processor either Account Manager through ASCIs and FPGAs with a standard processorO core and custom peripherals, or [email protected] for high unit volume applications, a processort modified to your specifications from a Doug Cordier major processor supplier. n Account Manager ri [email protected] Software is going through its own changes. Commercially available packages for op- erating systems have been around for several years, but now a great selection of mid- Barbara Quinlan P Account Manager dleware, databases, drivers, and other foundation code is available. This reduces the amount of efforte needed to get the basic functionality in place for an embedded design. [email protected] l Instead, the designer can focus on the application. Andrea Stabile g Advertising/Marketing Coordinator nWhile more packages are available, a new wave of tools for testing and simulation [email protected] i is simplifying the entire software development process. Simulation tools enable a tre- Christine Long Smendous amount of software development to be completed, evaluated, and tested prior E-marketing Manager r to the arrival of first hardware. At the same time, new tools make it easier to manage [email protected] multiple configurations of the finished product that meet the requirements of many dif- F ferent customers. Regional Sales Managers This year promises to be an exciting one for Embedded Computing Design. The edito- Jane Hayward – California rial calendar for 2007 has been structured to include a special feature written by our [email protected] editorial staff for each issue. These will be in-depth articles examining key topics facing embedded computing systems designers and managers. Embedded Computing Design Phil Arndt – East Coast will also have hardware and software feature tracks. The hardware track will cover [email protected] technologies and trends of interest with expanded coverage of processors, boards, and Richard Ayer – West Coast systems. The software track will discuss software topics ranging across the entire design [email protected] cycle from requirements definition to test, plus trends and new developments in firm- Ron Taylor – Mid Atlantic/Midwest ware, operating systems, and middleware. [email protected] The rampant proliferation of embedded electronics is one of the driving factors for the changes we’ve made to Embedded Computing Design for 2007. We will still cover International Sales technology used in communications, automotive, telematics, transportation, medical, Stefan Baginski simulation, and test and measurement applications, plus broaden the coverage to in- European Bureau Chief clude electronics used in consumer devices. [email protected] Dan Aronovic The Embedded Computing Design team and I look forward to covering all of your great Account Manager – Israel stories for 2007! [email protected]

Reprints and PDFs Becky Mullaney: 717-399-1900 ext 166 [email protected] Jerry Gipper, Editorial Director [email protected]

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution By Don Dingee More batteries, Mr. Taggart?

n a recent trip to visit my daughter, we found the electronic “We’re trying to maximize the usable hydrogen storage capac- deadbolt to her apartment was – well, dead. Fortunately a ity of borohydride in order to make a fuel cell power source last roommate was inside and able to open the door, and after- longer,” says study leader Dr. Don Gervasio, an associate research wards we replaced the four AA batteries in the unit. professor within the institute’s Center for Applied NanoBiosci- O ence. “That could lead to the most powerful power source ever I’m imagining that many of you are fresh off similar experiences produced for portable electronics.” with the electronic devices you got over the holidays. For in- stance, I understand the Nintendo Wii Remote eats two AA bat- Borohydride isn’t new for fuel cells,l buty using novel chemical teries every 30 hours with the MEMS accelerometer and pointer additives Gervasio’s team is workingn to increase hydrogen capac- device on. ity of their solution two to three times that of simple aqueous so- dium borohydride solutions,O while preventing the solution from I’m thinking of rewriting that scene in Blazing Saddles for a bat- solidifying. t tery commercial. Instead of a plea for dimes to feed the absurd toll booth in the middle of nowhere, Mr. Taggart (in that Slim Pickens in drawl) utters: “Somebody’s gotta go back and get a whole mess Thisr borohydride technology can be … of AA batteries.” P Charge it on my USB e commercialized within a few years, powering The primary AA battery and other batteries as we know theml are not going away soon, but other ideas are hittingn the market.g portable electronic devices three to five A new twist on the secondary AA batteryi is the USBCELL by times longer than conventional batteries Moixa Energy Ltd. (www.usbcell.com,S see Figure 1). These are standard AA size rechargeable NiMH of the same size and weight. cells (1.2 V, with a 1,300 mAhr capac- ity) but with a handy featureo – when you flip the end Fcap, a USB connector The ASU prototype fuel cell system houses the improved boro- is revealed. Plug the battery into any hydride solution in a tiny gas generator containing a ruthenium USB port, and the battery recharges metal catalyst. In the presence of the catalyst, the borohydride enough for temporary use in a few min- solution reacts to form hydrogen gas. The hydrogen moves across utes and completely in a few hours. a special gas/liquid separating membrane to the fuel cell com- ponent, where it combines with oxygen to generate electricity. Two pairs of USBCELLs might be Because the hydrogen generated is consumed immediately and worth the price for Wii addicts out there produces only a nontoxic watery by-product, the fuel cell is safe. – one pair in use, one pair charging in the Wii console. These would also be Although the fuel cell itself is reusable and doesn’t need recharg- ing, the borohydride fuel eventually is depleted and replacing handy in an office environment for Figure 1 things like the DLP projector remote, borohydride cartridges is required. The fuel cell generates some laser pointer, or other device that seems to be dead just when most heat, but researchers find it generally doesn’t get any higher than needed. A couple minutes with the USBCELLs in a USB port, and body temperature. that online match or meeting can proceed. While the prototype fuel cell is the size of a shoebox, Gervasio Better chemistry, longer power says this borohydride technology can be developed into a compact Rechargeable NiMH is today’s technology, but fuel cells are in fuel cell package and commercialized within a few years, pow- the foreseeable future as a portable power solution. Last year, we ering portable electronic devices three to five times longer than noted UltraCell (www.ultracellpower.com) launched its XX25 conventional batteries of the same size and weight. reformed methanol fuel cell for laptop systems. Inventors are continuing to push the envelope seeking denser, longer-lasting Creative portable power? solutions for portable electronics power. I’d say I’ve had enough; it’s time for some better portable power ideas like these. Here’s a question for designers: What do you Chemists at Arizona State University’s (ASU’s) Biodesign Insti- see as creative portable power solutions for the devices you’re tute have created a tiny hydrogen-gas generator using an improved designing today? As always, e-mail your thoughts and ideas to solution containing borohydride. Unlike other fuel sources, boro- [email protected] – before those AA batter- hydride works at room temperature to liberate hydrogen. ies we all depend on give out again.

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution y nl t O rin e P gl in r S Fo

RSC#  @ www.embedded-computing.com/rsc Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution Embedded analysis from health to travel to dirt By Hermann Strass Telemedication The German association of electrical/electronics engineers pres- mechanisms to protect stored information and data exchange ents a major conference every two years. This year’s venue was with authorized reading devices. The U.S. DoD is also using these held at Aachen, Germany. Reports about telemedical services chips for ID cards. once again comprised a significant part of the conference pro- gram, including the following applications. Archeological detection Archeologists now can analyze terrain without digging into Intelligent shoe sole the earth, thanks to the Institute for Physical High-Technology Exercise and simple body movement (walking, jogging, skiing, (IPHT) at the University of Jena, Germany,ly which has developed a and so on) are vital in preventing blood circulation problems. Superconducting Quantum Interference Detector (SQUID) to de- tect historically interesting archeologicaln sites worldwide. Teams Coordinated exercise, mostly walking, lowers weight and reduces high blood pressure. Sendsor GmbH, Germany, has developed recently used this technologyO to analyze ground pictures near Nasca an intelligent shoe sole that reads personal movement parameters in Peru, as illustratedt by Figure 1, courtesy of IPHT. SQUIDs are and sends that information to the walker’s mobile phone. Sen- also used to analyze electrical currents in human brains. sors in a matrix generate a detailed pressure profile of the sole. in Steps are counted and walking habits are detected. The walker can r download training programs from the Internet to a mobile phone P and receive feedback from the fitness program supervisor. e Asthma analysis l Asthma is a widespread disease. Thorough analysisg requires two weeks’ worth of lung function (peak-flow) measurements using a spirometer. Few doctors own such ani instrument.n A project team at the Technical University of Munich (TUM) and Sendsor GmbH have developed a simple, low-cost,S fully automatic spi- rometer system that can monitorr about 40 parameters. It contains no moving parts and onlyo one start button. Controls are not need- ed. The spirometerF device from Vitalograph, United Kingdom, was integrated into this Telemetry Personal Health Monitoring (TPHM) project. The spirometer works in combination with an inhalator, which documents medicinal applications. Doctors and health insurance companies can supervise compliance with the prescriptions.

Blood pressure control Also in conjunction with the TPHM project, the Heinz Nixdorf Institute at the TUM and Omron have developed a blood pressure Figure 1 measuring device that sends patients’ readings via the Internet. An integrated position sensor supervises correct positioning of the The SQUID consists of a SQUID chip and an embedded computer reading meter at the wrist. The blood pressure sensor transfers paired with a cryogenic device that cools the chip to extremely the readings to a device like a mobile phone or PDA. Patients low temperatures to enable superconductivity. The 10-layer chip can monitor their blood pressure at any time on the device dis- (LTS SQUID gradiometer), a magnetic field sensor developed at play. Doctors can check blood pressure more often and for longer the IPHT Department of Quantum Physics, handles Josephson time periods without personal contact with the patients. This helps currents up to 200 A. The complete system with separately sus- make drug dosage more precise and identifies deviations earlier, pended wheels for smooth movement can be moved by foot or as well as opening the way to new forms of therapy. hauled by an all-terrain vehicle. The antennas for position acqui- sition and radio reception of the reference station’s signals are Monitoring travelers mounted on top of the system. The SQUID’s real position is re- The United States is improving the next generation of pass- calculated from the GPS antenna position utilizing data from an ports by embedding security chips from Infineon, Germany, into inertial system, resulting in a position resolution in the centimeter American citizens’ passports. Next year the U.S. government will range. All data is available online, viewed on a laptop, and regis- issue 15 million new passports as part of the world’s largest na- tered in a data logger. tional passport project. A computer chip will be implanted into the back cover of the passport and protected by shielding material. For more information, contact Hermann at: The security and ID chips use more than 50 individual security [email protected].

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution PERSPECTIVE AND NEWS Embedding human knowledge in software

By J. Simon Tack, Sean Melody, and Dr. Todd Griffith he challenge of knowledge acqui- sition is legendary. The world’s by interviewing experts to discern rules ly experts – scientists, researchers, of thumb or collecting information by These knowledgen engineering tools are engineers, military personnel, and observing experts in the field, but rather developed using the Eclipse open devel- manyT others – are in increasingly short enabling experts to sketch and articulate opmentO platform. Eclipse tools provide supply. their expertise. This way, expertise is suc-t a robust, low-cost, mainstream develop- cessively refined as it is applied to data ment environment. One of the reasons These experts have built up strategic and integrated with systems. Thei expertsn development teams decided to migrate to knowledge over decades of experience. retain ownership of their expertiser and are the Eclipse platform was the broad range They can effectively analyze most situa- motivated to participate and create models of plugins available for Eclipse. This tions through the lens of their experience that help solve new problems.P speaks of both the widespread adoption to quickly determine the actions needed e of the platform and the vibrancy of the to solve problems. Unfortunately, novices Softwarel tools step in community. typically do not acquire their expertise To surmountg these challenges, the Exper- readily because they cannot effectively tise Encoding and Execution Workshop To show the merits of this technology for n3 3 obtain expertise through observation, and i(E W) has been developed. E W is novel embedded systems, a demonstration sys- experts usually have little time to train software that enables experts to easily tem is being constructed using an RLW S 2 others. What is needed is a way to rap- computerize their own strategies for solv- Inc. S NAP wireless embedded data con- idly capture their expertise perhapsr in two ing complex problems. Expertise encoded verter and multiprocessor. The SxNAP ac- weeks or less and integrateo it into existing as Discovery Machine (DM) models may quisition processing devices are designed systems. F be deployed as DM Gears – customized, to answer the “missing inch” between stand-alone, executable process models machinery and enterprise software in The creation of systems that use expertise that run on the lightweight DM Runtime Condition-Based Maintenance solutions. has been a goal since the advent of com- Environment (DMRE) on top of the Java The availability of DM Gears for the puters. From the early 1960s on, research- Runtime Environment (see Figure 1). S2NAP sensors/processors will enhance ers have been working in areas of artificial intelligence, expert systems, cognitive sys- tems, and intelligent systems to capture expertise. Each attempt was accompanied by promises of computers as intelligent as human beings.

Although many interesting prototypes ex- ist, the promise of artificial intelligence has never crossed the chasm into main- stream commercial use. Getting human knowledge into software is extremely difficult. Those who know how to build intelligent systems rarely have expertise in other areas. Encoding expertise has been challenging, time consuming, and obscure; human expertise cannot be re- duced to a general method such as search- ing through a problem space. There are as many strategies as there are problems.

A new approach is needed for experts to introspect on their problem solving, not Figure 1

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution the capabilities of mobile, low-power In addition to EDMRE, development of systems by providing an intelligent low- an Embedded Expertise Encoding and power data interpretation system (see Execution Workshop (E4W) will allow Figure 2). 4 users to encode, test, and deploy their E W will allow knowledge strategies as DM Gears that Transmitting data in a typical sensor uses can run on EDMRE. E4W will be built a large percentage of power compared based on the existing architecture, which to the power used computing the data to users to encode, provides the software components essen- be transmitted. Enhancing the expertise tial to knowledge capture environments as of the SxNAP processors will reduce the test, and deploy an API, optimizing reuse and streamlining power demand by trading computational maintenance of customized applications. bandwidth for transmission bandwidth. The sensor will transmit less yet higher- their knowledge The availability of an open development level data based on its inputs. platform such as Eclipse provides not only an able environment for developing To help developers deploy expertise, the strategies as applications such as this, but also a large Insert Gear plugin, an Eclipse plugin de- developer community that can provide and veloped in Java using the Eclipse Plugin be provided with tools. Using the Eclipse Development Environment, inserts Java DM Gears... development platform helps Discovery code that executes expertise in the form of Machine technology achieve its primary a DM Gear. The plugin reads the Gear file goal – deliveringly human knowledge to and provides a list of the Gear inputs and software. outputs. The generated code executes the fashions, will benefit greatly from this n Gear either synchronously or asynchro- technology. O nously. This closes any potential gaps in t J. Simon Tack is a software engineer at the knowledge-capture process by allow- Commercialization plan Discovery Machine. He has more than ing software engineers to utilize subject Next in the development plan i is n bring- 10 years of professional application matter expertise, and in turn, simplifies the ing these powerful knowledge-enablingr development experience using C++ and deployment of strategic expertise. tools to a range of embedded systems Java. Simon holds a BS in Electrical such as robots, sensors, P appliances, cel- Engineering from Drexel University. With tools like these, developers can di- lular telephones, and PDAs. The embed- rectly embed the knowledge of experts in ded systemsl ande robotics domain needs Sean Melody has been a software robotics technologies such as: knowledgeg capture and deployment tools. engineer at Discovery Machine since The technology platform is not domain graduating from Northwestern University n Perception inspecific and will have applications across with a BS in Computer Science in 2001. n Vision and sensing many industries and applications. n Data fusion S Dr. Todd Griffith is the founder and CTO n Mapping and navigation r An even lighter Embedded Discovery Ma- of Discovery Machine. Todd holds a PhD n Positioning o chine Runtime Environment (EDMRE) and an MS in Computer Science from the n Collaborative/cooperativeF behavior under development runs on top of the Java Georgia Institute of Technology. He also n New concept prototypes Micro Edition (Java ME). This will cap- holds a BS in Computer Engineering and ture expertise in the form of Discovery a BA in Philosophy, both from Bucknell These fields, which are currently trying Machine Gears to be deployed on any em- University. to encode expert knowledge in traditional bedded system that supports Java ME. To learn more, contact the authors at:

Discovery Machine, Inc. 454 Pine Street Williamsport, PA 17701 570-329-5661

[email protected] [email protected] [email protected] www.discoverymachine.com

For more information on RLW and S2NAP, contact:

Derek Stott, CFO RLW Inc. 2029 Cato Avenue State College, PA 16801 814-867-5122 Figure 2 www.rlwinc.com

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution y nl t O rin e P gl in r S Fo

RSC# 13 @ www.embedded-computing.com/rsc Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution Embedded industry organizations overview

Strength in numbers: technology consortia wield collective powerly On By Jerry Gipper and Don Dingee t rin he world of high technology is replete with organizations P gathering industry members together to advance common technology causes. It isl note unusual for a corporation or individual to be a member Tof several of these organizations, andg designers and managers should consider participating in at least one if they don’t already do so. Thisi articlen reviews the types of organizations and their benefits, presents a handy list of technology consortia, S and highlights issues they cited as important to their efforts. or Many types of consortiaF are in existence to- two structures. Some have both individual and corporate memberships available. day, ranging from formal standards bodies New standards are developed through the interaction of multiple corporations and that define the next generations of technol- individuals. Examples of standards organizations include IEEE, IEC, and ANSI. ogy to loosely organized ecosystems with n Ecosystem alliances primarily focus on marketing activities. They are often led by a marketing focus. Each of these organiza- one or two of the primary technology providers and are structured to promote that tions fills a specific niche and helps gener- technology. Membership fees are not usually charged and no standards are devel- ate support and interest in their respective oped, though technical specifications may be produced. Membership is typically at technologies. The number of members can the corporate level and by invitation only to companies that adopt the technology. range from a small handful of companies These organizations tend to be the most fluid with a much shorter life span and to thousands of individual and corporate changing membership to match the life cycle of the focus technology. members. Some organizations have a long n Trade associations (sometimes called Special Interest Groups or SIGs) are the life and rich tradition while others have a middle ground between a full-blown standards organization and an ecosystem alli- short life span that serves its purpose only ance. Technical specification development is high on the list of activities. Some trade through its technology foundation. associations are affiliated with a standards organization to allow them to develop accredited standards. A strong marketing effort often creates awareness for the orga- While there may be shades of gray in nization and its members. Most trade associations have some form of membership between, technology consortia can be fee, and sometimes fees are multitiered to enable a governance structure. grouped into three major categories: Working together to create mutual benefits n Standards organizations concen- The “strength in numbers” philosophy apparently applies to consortia; the collective trate on establishing well-defined and power derives from multiple members working together. Collaborating with a consortium structured standards that are properly almost always produces results in the marketplace faster than working alone. The deci- vetted and administered. Many have a sion to join a consortium depends on several factors, including the objectives and benefits long and established history with high associated with the organization, the types of technology involved, the composition of technology. The range of technology members, and the life-cycle position of the technology. covered by any individual standards organization is much broader than Table 1 summarizes the generic benefits of membership. Advantages in working with a an organization in either of the other specific technology change as the technology progresses through its life cycle, and an

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution organization may have several technologies at various points of the life cycle under its watch.

The following benefits are emphasized in the introduction portion of the technology life cycle:

n Access to specifications and standards and participation in their development is available to members. Members can play several roles in the development of future specifications and standards: very active and influential, passive observation, or something in between. Membership lets participants choose the most Strength in numbers: appropriate role. n Early adoption of new and emerging technologies can be fueled through sponsor- ship by and endorsement from industry organizations. An organization backing a technology brings credibility individual members may lack. Members’ participation in the development of early marketing messages helps strengthen and foster owner- ship of them. Consortia involvement gives members the chance to influence the technology consortia marketing message before it goes public. n Sound engineering practices are reflected in the collective work of the teams developing specifications and standards. Multiple sources can review the work, influence changes, and improve the quality of the finished standards product. Subject wield collective power matter experts that may not have been available to individual members can be used ly collectively. In the end, the finished product reflects the best work of the industry technology experts. On Once a technology is launched and early adopters have accepted it, consortia memberst group together backing efforts to help the technology grow into the mainstream. The benefits of growth phase participation include: rin n Reduced market risks because multiple members are testing the waters and sharing the risk. The number of members usually increases after a successful P introduction. Designers and managers can often judge the success of a technology by the size and constituents of the membership roster and how those membersle introduce products using the technology. Nonparticipants risk missingg out on market share attained by members working together. n Reduced development time and cost cani nbe gained from the knowledge learned in participation as a specification or standard is developed and finalized. Direct access to industry experts can provide insightS that reduces experimentation and trial and error compared to developmentr done individually. Often, intellectual property becomes available too help design and development proceed on a larger scale. n Branding andF awareness drive broader adoption of a new technology. Consortia help create and support the perception of momentum, highlighting proof points of usage and floating new application ideas. This is often difficult to accomplish as a small or individual company working alone and takes large amounts of financial resources or many voices. The collective voices of consortia members can generate more awareness faster.

Finally, as the technology moves into maturity, consortia benefits shift:

n Decreased trading costs and lowered trade barriers can be gained when standards are developed in a community, especially considering the increased pressure to license technology fairly to members. In fact, several efforts are underway to im- prove the patent process as it relates to standards development. The trading costs and barriers are also lowered when technology is more widely used. The increased unit

RSC# 15 @ www.embedded-computing.com/rsc

Table 1

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution Burning issues for consortia Embedded Computing Design informally surveyed the consortia in our list about issues they face as they advance their cause. Not surprisingly, the organizations share a number of concerns. A summary of the most pressing topics consortia representatives expressed follows.

Interoperability is considered crucial because it constitutes a basic benefit of having a standard in the first place. How do consortia ensure their work will interoperate with other related efforts? This is one of the most common issues across the consortia and has even led to the formation of a new class of organizations referred to as an alliance of alliances. These are super-level organizations that take the work by-products of several related organizations and attempt to ensure some level of interoperability. CAN in Automation (CiA) states their biggest concern is “to enable automation islands to grow together,” which hits at the heart of interoperability.

Many de facto company-specific specifications compete with open specifications and standards developed by the com- munity, and interoperability issues often must be addressed. Getting everyone on the same page can be very difficult, especially in cases where developers are trying to create differentiation and a competitive edge. Is it better to go it alone, or be involved in a co-opetition environment where everyone is working together to create a bigger market?

“Interoperability of devices used to sense, measure, and control industrial manufacturing is a key issue both at the device level and in the interface between the plant floor and the business enterprise.” – Robert Renner, executive director of ISA

Awareness is a primary goal of most of the organizations. Educating the target audience to make them aware of the technology is an ongoing process. The organizations use multiple venues to achieve awareness. Tradeshows, confer- ences, seminars, webcasts, mailings, and grassroots efforts are the most common toolsy for creating awareness. Several organizations are contemplating adding additional worldwide exposure and user groupl affiliations to create a larger voice for their technology. n “The ZigBee Alliance and its members are facing a familiar challenge found with the emergence of any new standard: growing the value chain, educating O the market, and boosting global market recognition.” t in – Bob Heile, chairman, ZigBee Alliance Innovation applies mostly to organizationsr that develop specifications and standards. Innovation drives new growth, pulls in new members, and helps create awareness. Coming up with new innovation is a challenge for many of the organizations. Innovation thatP is relevant and useful to the industry and members is difficult but critical to success. New marketsle are important as many organizations are working hard to expand the use of their technology. Growth ninto gnew markets further establishes a technology and helps stimulate innovation. ivolumes and market acceptance drive requires application of resources, mostly in S down the cost models. membership fees and time of key person- n Increased product quality and nel. Larger companies are often involved r safety is secured from the extensive in many consortia to stay “in the loop,” o vetting process that occurs when but strategically work with a core set. F several companies are developing Good participation forms the perception with the same technology and sharing of a company being an industry insider. results through technical committees. These committees can make changes The list to the specifications or standards that Participation in technology consortia can will improve the quality, safety, and render many benefits. Designers and man- other aspects of the technology. agers should study the industry their com- n Protection against obsolescence pany operates in and become involved in is provided when many members the organizations that best fit with their participate and the market reaches strategic intent. critical mass. Strong incentives help retain existing users of the technology To aid in that search, Embedded Comput- and give them a technology roadmap ing Design has identified and verified a list that will ensure members’ success. As of trade associations and special interest standards are developed and ratified, groups that those involved in embedded the likelihood for changes to instantly computing technologies should consider. make a product obsolete decreases. Many of these consortia are connected in ways that may affect the decision process. In general, the advantages of participation They work together to solve the challeng- RSC# 16 @ www.embedded-computing.com/rsc outweigh the disadvantages. Sometimes es of interoperability, awareness, innova- announcing membership can signal ad- tion, and new market development. vance intentions a company may not be ready to share publicly. Shrewd use of With the landscape changing constantly, we membership can be played in multiple will continue to identify new organizations ways to help create confusion among com- and update existing ones. Visit our web- petitors, and if not played well can give site at www.embedded-computing.com to the impression a company is struggling view the most current listing and to learn with their strategic direction. Participation more on how these organizations interact.

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution Consortia List

Organization Technology URL

1394 Trade Association IEEE 1394 multimedia connection www.1394ta.org

Accellera Electronic design tools www.accellera.org

Ada Resource Association (ARA) Ada compilers www.adaic.org

American National Standards Institute (ANSI) Various electronic standards www.ansi.org

AS-Interface AS-Interface networking protocol www.as-interface.net

Association for Automatic Identification and Mobility RFID and barcoding www.aimglobal.org (AIM)

Blade Systems Alliance (BladeS) Blade systems www.bladesystems.org

Bluetooth Special Interest Group Bluetooth wireless technology www.bluetooth.com

CAN in Automation (CiA) CAN-based higher-layer protocols www.can-cia.org

Communications Platforms Trade Association (CP-TA) Specification-based communications platforms www.cp-ta.org CompactFlash Association CompactFlash and CF+ www.compactflash.orgly Consumer Electronics Linux Forum (CELF) Linux www.celinuxforum.orgn Continental Automated Buildings Association (CABA) Home automation Owww.caba.org Continua Health Alliance Personal telehealth systems t www.continuaalliance.org ControlNet International ControlNet networking rin www.controlnet.org Eclipse Foundation Eclipse platform P www.eclipse.org Embedded Microprocessor Benchmark Consortium Processor benchmarking www.eembc.org (EEMBC) le EtherCAT Technology Group (ETG) EtherCATg www.ethercat.org Ethernet Powerlink Standardization Group (EPSG) inEthernet Powerlink www.www.ethernet-powerlink.org Fibre Channel Industry Association (FCIA) S Fibre Channel www.fibrechannel.org Fieldbus Foundation r Fieldbus www.fieldbus.org HART CommunicationF Foundationo (HCF) HART protocol www.hartcomm.org HyperTransport Technology Consortium HyperTransport www.hypertransport.org

InfiniBand Trade Association (IBTA) InfiniBand www.infinibandta.org

Information Technology Systems Embedded research www.cordis.lu/ist/embedded

Institute of Electrical and Electronics Engineers, Inc. Various electronic standards www.ieee.org (IEEE)

Intelligent Transportation Systems of America (ITSA) Intelligent transportation systems www.itsa.org

INTERBUS Club INTERBUS protocol www.ibsclub.com

Interchangeable Virtual Instruments (IVI) Programming test instruments www.ivifoundation.org

International Engineering Consortium (IEC) Various electronic standards www.iec.org

ISA Automation www.isa.org

LXI Consortium, Inc. LAN extensions for instrumentation www.lxistandard.org

Modbus-IDA Modbus communication protocol www.modbus.org

Mountain View Alliance Platforms and building blocks www.mountainviewalliance.org

Network Centric Operations Industry Consortium Network centric www.ncoic.org (NCOIC) Unified Modeling Language (UML) and Model Driven Object Management Group (OMG) www.omg.org Architecture (MDA)

Open Core Protocol Specification Intellectual Property (IP) core interfaces www.ocpip.org

Continued on next page

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution Consortia List, continued

Organization Technology URL

Network technologies based on the Common Industrial Open DeviceNet Vendor Association (ODVA) www.odva.org Protocol (CIP)

Open Group Real-Time and Embedded Systems Forum Real-time and embedded systems www.opengroup.org/rtforum

Open source Linux in the context of machine and plant Open Source Automation Development Lab (OSADL) www.osadl.org control systems

Open Source Development Labs (OSDL) Open source Linux to enterprise computing www.osdl.org

Open SystemC Initiative (OSCI) SystemC www.systemc.org

OpenCores.org Intellectual Property digital modules or cores www.opencores.org

Optical Internetworking Forum (OIF) Optical internetworks www.oiforum.com

OSGi Alliance Java applications and services www.osgi.org

PC/104 Embedded Consortium PC/104 small form factor boards www.pc104.org PCI-SIG PCI bus and PCI Express www.pcisig.comly High-performance telecommunications and industrial n PICMG computing boards and systems, CompactPCI, www.picmg.org AdvancedTCA, MicroTCA t O PLCopen IEC 61131-3 control programming forn PLCs www.plcopen.org Power.org Power Architecture ri www.power.org PROFIBUS International (PI) PROFIBUS and PROFINET P www.profibus.com PXI Systems Alliance PCI extensionsle for instrumentation www.pxisa.org RapidIO Trade Association RapidIOg embedded fabric interconnect www.rapidio.org SATA-IO inSerial ATA www.sata-io.org Carrier grade base platforms for service provider SCOPE Alliance S www.scope-alliance.org r applications Small Computer System Interface (SCSI) and Serial SCSI Trade Association (STA)o www.scsita.org F Attached SCSI (SAS) Digital motion control bus, SErial Real-time SERCOS International (SI) www.sercos.org COmmunications System (SERCOS)

Highly available carrier grade systems with off-the- Service Availability Forum (SAF) shelf hardware platforms, middleware, and service www.saforum.org applications

Software Defined Radio Forum (SDRF) Software defined radio www.sdrforum.org

Standard Performance Evaluation Corporation (SPEC) Processor benchmarking www.spec.org

Storage Networking Industry Association (SNIA) Storage networks www.snia.org

USB Implementers Forum, Inc. Universal Serial Bus (USB) www.usb.org

UXP Initiative 10 Gbps systems www.uxpi.org

Critical embedded computing boards and systems, VITA www.vita.com VMEbus, VPX, and PMC

VSI Alliance System-on-Chip and Intellectual Property (IP) www.vsia.org

WiMedia Alliance Ultra wideband wireless RF-based communications www.wimedia.org

Wireless Industrial Networking Alliance (WINA) Wireless RF-based communications www.wina.org

XTX Consortium XTX form factor embedded motherboards www.xtx-standard.org

ZigBee Alliance Wireless RF-based communications www.zigbee.com

Z-Wave Alliance Wireless RF-based communications www.z-wavealliance.org

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution y nl t O rin e P gl in r S Fo

RSC# 19 @ www.embedded-computing.com/rsc Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution y nl t O rin e P gl in r S Fo

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution y nl t O rin e P gl in r S Fo

RSC# 21 @ www.embedded-computing.com/rsc

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution Technology trends in embedded computing

Open source opens doors for improved developmently On t By Dr. Inder M. Singh rin P An industrye in flux The benefits of Today’s embeddedl devices are more com- standardization plex g and incorporate richer feature sets Many reasons explain why this shift is nthan ever before. With more powerful happening now. First, during the past he complexity of today’s ihardware, pervasive network connectivity, few years, embedded device develop- Sand enhanced user expectations, embed- ers have begun standardizing platforms embedded devices is ded devices no longer are limited to single based on open standards such as POSIX r function, dedicated systems in which the and open source software such as Linux. causing a shifto in how embedded device does just one dedicated Just as many factors influence the choice T F job. At present, embedded devices fre- between commercial or open source prod- quently have secondary or tertiary func- ucts, many benefits can result from stan- developers approach designing tions that enhance the primary function dardization, including performance gains, or sometimes allow for an entirely dif- faster time to market, and lower develop- the software that drives these ferent purpose altogether. Moreover, the ment costs. functionality of the device may often be systems. Dr. Inder M. Singh will dynamically altered, more like a personal computer than a traditional embedded device. address how this shift is trans- By standardizing a com- In addition to the increased richness and forming the embedded landscape complexity of embedded software, devel- opment cycles are getting shorter all the mercial or open source and explain how open standards, time. Very often, there simply isn’t enough time available to develop all the required platform, developers can open source elements, and spe- functionality from scratch and, at the same time, meet new product deadlines. streamline the code cific technologies such as Linux Because of this shift, the embedded in- dustry is undergoing a sea change in its writing process by re- and Eclipse development tools approach to development. Where previ- ous generations of embedded developers are facilitating a more efficient were required to write unique operating using core software that systems and application code from scratch development process. for each specific device, developers are doesn’t change from now increasingly opting to use – and reuse – COTS software to build their embedded device to device … devices.

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution By standardizing a commercial or open Dr. Inder M. Singh is chairman of To learn more, contact Dr. Singh at: source platform, developers can stream- LynuxWorks and served as CEO until line the code writing process by reusing 2006. He has founded core software that doesn’t change from and led numerous LynuxWorks device to device, freeing development companies including 855 Embedded Way teams to focus on the elements that dif- Exelan and Kalpana. ferentiate their product. In addition to al- He was formerly San Jose, CA 95138 lowing developers to concentrate on the board chairman and key elements of their device, using COTS president for the 408-979-3300 Embedded Linux software based on open standards speeds [email protected] the development process and gets new de- Consortium. He vices to market faster. holds PhD and www.lynuxworks.com M. Phil. degrees in Computer Science Linux as a foundation for from Yale University and an MSEE from embedded systems Polytechnic Institute of New York. A key enabler of this trend toward software reuse lies in the growing ecosystem of both embedded software and applications for Linux. As embedded software gets too complex to build everything in-house, de- y velopers are relying on third-party or open l source software to keep the development n process efficient. In the case of Linux, us- ing open source software also alleviates O some of the problems traditionally en- t countered when running applications over n different embedded hardware platforms, ri much as it has done in the enterprise soft- ware market. The growing use of Eclipse, P an open source development framework that provides enterprise-strength tools for le embedded developers, supports the adop- g tion of Linux in embedded devices. in A final factor influencing the trend toward reusing software is the increasing “openS- ness” of today’s embedded devices.r More and more, suppliers ando sometimes even end users demandF the ability to custom- ize device functionality by download- ing updates or other software pieces. For obvious reasons, it is more efficient to do this for a device built on an open standard platform. In the case of mobile phones, for example, network providers like to customize the handsets to support value-added services they provide and to differentiate the look and feel from other providers. Furthermore, end users are in- creasingly downloading additional appli- cations available on the Internet. Similar trends are expected in the case of automo- tive infotainment systems and TV set-top boxes.

More innovation ahead The growing availability of reusable soft- ware code for embedded systems and de- velopers’ increasing willingness to use it is prompting a rapid shift in the embedded industry. By removing the chore of writ- ing time-consuming infrastructure code, these COTS software packages are setting the stage for a new wave of innovation in the embedded industry. RSC# 23 @ www.embedded-computing.com/rsc

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution y nl t O rin e P gl in r S Fo

RSC# 24 @ www.embedded-computing.com/rsc

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution Technology trends in embedded computing

The advent of COTS middleware use y in embedded systemsnl t O in By Andrew Foster Pr le s embedded systems become increasinglyng more complex and heterogeneous, OEMs and system integrators face escalating technical, organizational,i and cost challenges. Embedded system engineers must tackle the S Adaunting tasko of designingr and building complex hardware that is multifunctional and highly compatible, yet as cost sensitiveF as possible. The good news is that COTS middleware can simplify embedded systems develop- ment by making system integration far easier, improving application portability, and increasing software reuse.

An attractive solution Building increasingly complex and diverse systems causes a num- Thanks to recent middleware developments including technol- ber of obvious side effects. Initial build and integration costs are ogy advancements, standardization initiatives, and new innova- higher. In the automotive sector for example, it is common for the tive approaches to building embedded systems, OEMs and system software development process to be highly distributed between integrators no longer need to think in terms of trade-offs when different OEMs and for each OEM to have considerable freedom developing their systems. With COTS middleware, assuming as to how they realize the required functionality. functionality will come at a higher price or that a highly com- patible system will severely limit system resources is no longer An OEM typically has only a black box specification of how the necessary. different parts of the system should be connected, making inte- gration and error detection difficult. In contrast, business appli- The embedded community has been following a path to a COTS cations are developed with usually far greater constraints on the middleware solution for several years. In sectors such as defense software technologies used to facilitate a much smoother integra- and aerospace, system requirements are increasingly complex, tion process. Compared to desktop-based applications, the goal dynamic, and demanding as far as scalability (vast numbers of of software reuse in embedded systems has been more difficult embedded sensors in a network) and quality of service is con- to achieve. In many cases applications are completely rewritten cerned. Today’s solutions are often highly proprietary, brittle whenever changes to the underlying platform are made, creating a and nonadaptive, expensive to build and maintain, and extremely very expensive system upgrade process. As systems become more vulnerable to changes in the underlying platform and hardware complex to build in the first place, this approach becomes less and technologies. Other sectors such as consumer electronics are en- less tenable, and a COTS middleware solution becomes more and countering the same types of challenges fueled by the growing more attractive. complexity of electronic devices including expanded features, networking, and multimedia capabilities, which are driving the re- Middleware technologies such as Java 2 Platform Enterprise quirement to add full commercial operating systems in many cases. Edition (J2EE) and Common Object Request Broker Architecture

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution (CORBA) have achieved wide acceptance across market sectors Standardization initiatives for building enterprise scale systems. Middleware provides a Another key facet of this next-generation embedded middleware much higher level of abstraction for software engineers to write is that through a combination of support for standard interfaces applications against. Benefits of using middleware to build soft- defined by OMG and product developments it is now very easy to ware applications include: support highly specialized transports (other than TCP/IP), such as those available in the form of high-speed bus-based interconnects. n Simplifying component integration using different This allows the middleware to be used efficiently while maintain- platforms, thus facilitating seamless interworking between ing excellent latency and throughput characteristics throughout heterogeneous systems the system. n Simplifying functionality distribution, independent of physi- cal location or business domain The result of these advances is that it is now possible for em- n Allowing a system to scale as usage requires and facilitating bedded developers to utilize a highly efficient, standards-based, load balancing across components unified middleware layering across all processors in the system, n Enabling distributed component management simplifying component integration within the system, signifi- n Simplifying fault-tolerant and secure operation assurance cantly improving application portability, and facilitating greater software reuse, as illustrated in Figure 1. Addressing middleware hang-ups In spite of these major benefits, middleware technologies such as CORBA have been slow to gain wide acceptance within the em- bedded community. This sluggish take-up has mainly been due to concerns about the overhead using middleware imposes on ly an embedded application, particularly in terms of size (memory footprint), performance (latency, throughput), and predictability n (timeliness). Now, however, new middleware developments in- O cluding technology advancements, standardization initiatives for t embedded systems such as Object Management Group’s (OMG’s) CORBA/e standard, and innovative approaches to building em- in bedded systems are making advanced middleware use viable in r even the most resource-constrained environments. P Technology advancements Middleware vendors such as PrismTech are producing extremelyle lightweight, highly optimized CORBA implementationsg that can be used efficiently in just about any embedded environment. For Figure 1 example, in the software-defined radio domainin it is quite common to find small form factor radios consisting of a multiprocessor environment, which includes General S Purpose Processor (GPP), DSP, and FPGA processor setsr as part of the signal processing Innovative development approaches chain. Until recently, however,o middleware would only be used to COTS middleware vendors are also combining Model Driven De- host parts of the applicationF running on the GPP. velopment (MDD) techniques with advances in middleware for use in embedded systems to produce tool chains that support the Due to concerns about footprint and performance as well as a full embedded development life cycle. By using graphical tools lack of COTS middleware support, communication with any to model the software components in the system including the parts of the application hosted on the DSP and FPGA processors interfaces between components, it is now possible to generate a was previously based on proprietary communication drivers and complete executable application and deploy it into a distributed custom message formats. Obviously, if the underlying hardware embedded environment from a tool. MDD tools, with the empha- changed, large parts of the application would have to be rewritten. sis on hiding many of the underlying complexities from the devel- PrismTech’s approach, for example, has been to develop highly oper and automatic code generation, remove the final perceived optimized yet standards-conformant middleware implementa- hurdle to using middleware in embedded systems – complexity. tions that can support GPP, DSP, and FPGA environments. Other Using these tools, any middleware complexities are completely COTS vendors are also producing implementations that can sup- hidden from the embedded developer, allowing more focus on de- port these environments. veloping application logic and not on infrastructure concerns.

Lightweight CORBA PrismTech’s embedded COTS middleware product, OpenFusion e*ORB, supports distributed CORBA components whose footprint can take up less than 100 KB in a DSP environment, easily fitting into the onboard memory available in most modern DSPs. The company also has implemented COTS CORBA middleware in hardware, the Integrated Circuit ORB, for use in an FPGA environment.

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution embedded application portability can be improved significantly while maximizing code reuse at the same time. In the short term, Middleware vendors are producing complex embedded applications that use middleware are easier to build and integrate, and over the life of a system can protect innovative new technologies … that the embedded OEM from expensive system rewrites whenever the hardware is changed or upgraded.

are enabling middleware to be used in Andrew Foster is product manager for PrismTech’s CORBA middleware technolo- environments previously deemed gies. He is responsible for providing both the strategic vision and managing the evolu- impossible or inappropriate. tion of the highly successful range of CORBA middleware products. He has more than 12 years of experience developing Distributed Easier integration, lower costs Real-time and Embedded (DRE) software In conclusion, standards bodies such as OMG have been working applications and products. He frequently hard to more adequately address the needs of the embedded devel- presents at conferences on subjects relating opment community. With broad industry support, new specifica- to distributed middleware and embedded technologies. Andrew tions such as OMG’s CORBA/e standard are enabling vendors to holds a B. Eng (honors) in Digital Systems Engineering as well produce highly optimized, standards-compliant middleware im- as an MS in Computer-Based Plant and Process Control from plementations that can be used successfully in the most resource- the University of Sunderland in the Unitedly Kingdom. constrained embedded environments. Middleware vendors are producing innovative new technologies such as a hardware ORB To learn more, contact Andrew at:n for FPGAs that are enabling middleware to be used in environ- O ments previously deemed impossible or inappropriate. tPrismTech Corporation Combined with modern Model Driven Development approaches 6 Lincolnin Knoll Lane, Suite 100 • Burlington, MA 01803 to tooling, building applications using advanced embedded COTS r 781-270-1177 middleware is becoming easier through automating much of the middleware code generation and allowing the embedded devel- P [email protected] oper to concentrate on writing successful application logic.e By www.prismtech.com leveraging the higher-level abstractions middlewareg provides,l in r S Fo

RSC# 27 @ www.embedded-computing.com/rsc

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution y nl t O rin e P gl in r S Fo

RSC# 28 @ www.embedded-computing.com/rsc Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution Technology trends in embedded computing

Key industry movements y driving designs in n2007l t O in By Wade Clowes Pr le Adoptingg standards-based Outsourcing model on the rise he embedded industry nproducts As OEMs move to open standards-based iHistorically, many OEMs designed and products they are increasingly partnering faces several challengesS manufactured proprietary systems for with their vendors. With the constant push successive generations of products from for OEMs to expand their global footprint, r the ground up. While such hardware pro- more are migrating toward outsourcing Tin 2007. OEMso confront vided the necessary performance, these the design and manufacturing of building F solutions were costly and inflexible. The blocks and systems. OEMs are looking to mounting pressure to introduce increasing adoption of open standards- partner with companies that have an in- based products counteracts this approach depth understanding of their end user’s innovative products more quickly as it enables OEMs to reduce total cost applications. They are increasingly choos- of ownership and increase their design ing partners that can provide long-life, while reducing costs and staying flexibility. standards-based products with custom- ized features. Embracing the outsourcing Standards-based development allows de- model allows OEMs to reduce fixed costs, competitive in a global market. signers to take advantage of a wider selec- decrease time to market, and access indus- tion of commercially available software try experts that can act as an extension of Getting increased performance and hardware components than what is the company. These outsourcing benefits available with a proprietary design, en- allow the OEMs to focus on their core and reliability from processors abling OEMs to pick the best building businesses and develop increasingly dif- blocks and focus on the areas of design ferentiated end products. and embedded systems remains unique to their core competencies. The ecosystem of hardware and software Multicore processors bring providers that exists for these standards- great benefits a constant desire as well. As based components drives both higher The increased availability of multicore performance and lower cost from compe- processors such as Intel’s Core Duo pro- the new year gets underway, a tition. In the embedded market, it is also cessor families is making its mark on the important to economically create low-cost embedded industry. Multicore processors number of trends are gaining variants of standard products that en- include two or more cores in a single pro- able features important to OEM product cessor that enable lower power and higher differentiation. performance than is possible with single momentum within the embedded core processors. Dual core processors can (Editor’s note: Many new options are available for operate at low voltages and minimize clock computing marketplace. standards-based, processor, board, and products. Our March issue will provide a and signal switching, resulting in lower complete rundown on the numerous choices.) power dissipation in the active state.

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution These advances help bring innovative performance for specific algorithmic- performance to processor-intensive data- based applications. Unfortunately, they crunching applications such as imaging, usually do so at the expense of program- gaming, test and measurement, and in- ming flexibility. By using a combination The shift to open dustrial automation applications. Taking of processor types, the system designer advantage of this breakthrough, OEMs attempts to balance processing perfor- standards-based can design products and systems with mance, cost, and programming flexibility. higher-performance capabilities in smaller Such heterogeneous multiprocessing sys- physical and thermal footprints. Multicore tems contain multiple types of processors. products, outsourcing, processors will expand from PCs to em- bedded computers in 2007. One possible architectural approach for multicore adoption, and heterogeneous systems is to attach a Heterogeneous multiprocess- number of application-specific hardware ing delivers flexibility accelerators to an embedded server host heterogeneous multi- Using heterogeneous multiprocessors is computer. In this case the accelerators also gaining traction. Hardware accelera- can be either attached directly to the host processor use is clearly tors often provide very high processing CPU’s PCI Express root complex or via a driving the next genera- tion of designs.y nl high-speed switch fabric. Heterogeneous multiprocessorsO are particularly well suit- t ed for imaging applications and will help further advance developments within this rin industry segment. As OEMs search for ways to reduce cost P and speed time to market we will see these e trends fully develop. The shift to open l standards-based products, outsourcing, g multicore adoption, and heterogeneous multiprocessor use is clearly driving the in next generation of designs. S Wade Clowes, VP r of RadiSys Corp.’s o commercial seg- F ment, brought nearly 30 years of industry expertise when he joined the company in 2005. Wade has held management posi- tions with a number of companies, including a 22-year career with HP as well as the principal of Clowes and Associates, a consulting firm specializing in business and organization development. Wade received BS and MS degrees in Mechanical Engineering from Montana State University and completed the Stanford Executive Program.

To learn more, contact Wade at:

RadiSys 5445 N.E. Dawson Creek Drive Hillsboro, OR 97124 503-615-1100 www.radisys.com

RSC# 30 @ www.embedded-computing.com/rsc

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution y nl t O rin e P gl in r S Fo

RSC# 31 @ www.embedded-computing.com/rsc

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution www.embedded-computing.com/products

Ethernet-in-a-kit with Eclipse tools With network connections now everywhere, designers need easy ways to add interfaces to existing hardware. For just $99, a new kit provides a simple method of adding Ethernet to a design. The NetBurner Eclipse Ethernet Development Kit offers a complete set of powerful developer hardware, software, and tools that enable rapid Active drive for longer development of embedded network-enabled products. InfiniBand cables The included NetBurner Mod5270 Ethernet Core Module features a 32-bit 147 MHz Freescale ColdFire processor with a 10/100 Speaking of networking, often the performance of the network is Ethernet port, 47 digital I/O lines, three UARTs, and I2C and SPI only as good as the cable used. A quality cable can provide better ports. The kit also includes NetBurner’s Eclipse signal integrity and span longer distances. A new solution for Double Integrated Development Environment Data Rate (DDR) InfiniBand cables combines conductor technology delivering time- with tiny active analog signal drivers to allow cables three times saving features such longer than previously possible. as single-click compile and load capability, intel- GORE Extended Reach DDR Cable Assembliesly utilize the patented ligent code completion, self-equalizing GORE EYE-OPENER+ Conductor Technology com- bined with Q:Active ultrasmall low-powern analog semiconductors and integrated graphical debugger. This enables from Quellan to deliver exceptional signal fidelity over longer dis- tances in the smallest cross-sectionalO area. Fitting application needs developers to interact with external devices by writing in between passivet copper and optical cables, these cables offer code in ANSI C/C++ within lower crosstalk, less latency, and increased airflow in sizes ranging from 24 to 30n gauge and lengths ranging from 5 to 50 meters. the Eclipse environment i and compile and load it into r a NetBurner Ethernet Core W. L. Gore and Associates / www.gore.com PRSC# 32283 Module in seconds. e NetBurner / www.netburner.com l RSC# 25136 g in Modeling for S AUTOSAR USB 2.0 Hub IPr cores applications Adding a USB hub and multipleo ports to a system can be a bit of With more and more compute a challenge for designers. IPextreme and Cypress Semiconductor power and software being designed into automotive environments, have teamed up toF offer a low-power USB 2.0 hub core in an IP developers need better tools to get their complex ideas implemented format for adding to third-party products using Cypress’ proven and to market faster than ever. The AUTOSAR (AUTomotive Open technology from the EZ-USB HX2LP and other products. System ARchitecture) specifications provide an open standard architecture for developing vehicular software, user interfaces, and IPextreme’s USB20Hub is configurable to accommodate two to management. seven ports. Customers can choose one transaction translator for all ports to minimize production cost or one transaction translator The Rhapsody AUTOSAR pack delivers an AUTOSAR modeling per port to maximize throughput. The USB20Hub design has very environment based on UML and SysML that allows users to capture low power consumption AUTOSAR system models using AUTOSAR-specific diagrams, nota- and runs easily on USB tions, and terminology. This is achieved with five new AUTOSAR bus power. Once imple- diagrams: mented, the USB20Hub has externally configu- n Systems Diagram captures the overall AUTOSAR system rable options such as the n Software Component Diagram defines the software architecture number of active ports, n Internal Behavior Diagram specifies the interface between the power management, port AUTOSAR System and the standard AUTOSAR Run-Time indicators, and remote Environment with which it will integrate wake-up. The USB20Hub n ECU Diagram defines the Electronic Control Unit (ECU) types and IP core is fully synthesiz- their communication ports able and comes packaged in an IPextreme XPack to maximize ease n Topology diagram defines the physical topology or physical ar- of integration through a simple IP configuration user interface with chitecture of the system including all the ECUs in the automobile support for common EDA tools. The USB20Hub IP subsystem has and how they are connected received USB-IF compliance certification. Telelogic / www.telelogic.com IPextreme / www.ip-extreme.com RSC# 32284 RSC# 32285

Editor’s Choice Products are drawn from OSP’s product database and press releases. Vendors may add their new products to our website at www.opensystems-publishing.com/vendors/submissions/np/ and submit press releases at www.opensystems-publishing.com/news/submit. OSP reserves the right to publish products based on editors’ discretion alone, and does not guarantee publication of any product entries.

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution Enterprise software in embedded applications

COTS databases for embedded systems

By Steve Graves any embedded computer systems are excellent ycandidates for Membedded databases. Steve defines whatn anl embedded database is and what features developers can texpect O to be or not to be in them, and discusses which of those ifeatures,n whether present or not, make embedded databases appropriater for embedded systems. e P Databases, large and small n Japanesel consumer products giant JVC embeds a database in MP3 player technology Certain types of software are small foot- forg playlists and other data organization features print almost by definition. For example, n In the United States, an industry-leading provider of emergency communications boot loaders and device drivers aren’t ex- ingear integrates a DBMS into its product family of wireless infrastructure devices pected to hog memory and CPU resources, so embedded systems developers generS- Today, a COTS database system can be crucial to a product’s success. To support expand- ally don’t think twice about incorporatingr ing features, intelligent devices must manage growing volumes of more complex data. such code into their projects.o On the other Writing data storage and manipulation code from scratch – the “homegrown” solution hand, certain softwareF virtually screams, – simply takes too long, especially when factoring in quality assurance, portability, and “BIG!” and is rarely used in resource- scalability issues. Conversely, an off-the-shelf database system saves development time constrained systems. and, by separating application logic from data management logic, simplifies code reuse and maintenance. Given the competitive need to roll out and update products quickly, a Database Management Systems (DBMSs) COTS DBMS provides distinct advantages over homegrown code. once fit squarely in the second category. Certainly enterprise databases demand When investigating DBMSs, developers quickly discover that a dozen or so vendors of- far more memory and processing cycles fer embedded databases. This creates confusion because only a few embedded database than typical embedded systems provide. systems are practical for use in embedded systems such as cell phones, industrial automa- Even most embedded databases, a prod- tion, and the like. Not every embedded database provides the performance, efficiency, and uct category developed about 20 years ago small footprint needed for embedded systems. for embedding in server and desktop soft- ware, are simply too big, slow, and com- Which database management technology fits? plex to build into, say, a set-top box, cell For database management systems, devices represent a new development domain with phone, or onboard automotive system. unique needs. Requirements include small footprint because manufacturers seek to re- duce devices’ memory, CPU, and storage requirements for economic reasons. Another But as embedded systems have evolved to important need is real-time performance. Communications equipment, consumer devices, include devices with faster (32-bit) CPUs, aerospace and defense systems, and other categories of embedded software are expected to more available memory, and sophisticated respond instantly with consequences ranging from annoying to dire if this fails to happen. Real-Time Operating Systems (RTOSs), Therefore, data management for embedded systems must be inherently fast and efficient. COTS database systems are finding their way into devices where they would never A database for embedded systems must be reliable. While a software crash might be have been expected. For example: tolerated once in a while in desktop or server software, it’s not accepted at all in a cell phone or set-top box. The user will simply demand to return the product. For mission- or n Major European mobile phone life-critical embedded applications (for example, medical, aerospace, and some commu- handset manufacturers rely on a nications systems) a database will ideally include built-in high availability or a method COTS DBMS for address book, user for establishing and updating multiple copies of a database with automatic failover in notification, and message retrieval case of disruption on the primary node.

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution Embedded systems rarely manage data that fits into neat, tabular structures. Instead, they must address complex data such as trees and arbitrarily long arrays of simple or complex fields. A DBMS for embedded systems must provide the tools to work efficiently with such data types. Another reason an embedded systems database needs such tools is be- cause the developer usually predefines on-device data access. Tight integration between data management code and the application creates greater runtime efficiency, reducing the need for CPU cycles and other computing resources. Therefore, sophisticated develop- ment tools also contribute to a smaller footprint.

Sorting through database software features Given that embedded database software is diverse, what aspects of the technology best meet the previously described requirements? By considering a few of the key features embedded databases offer, some answers – or at least a clearer picture of the trade-offs – begin to emerge.

Storage modality On-disk storage The slowest function of most DBMSs is disk access, the mechanical process by which working copies of information are transferred to persistent media. Traditional on-disk databases are designed to cache frequently requested data in memory for faster access, but also to write all database updates, insertions, and deletes throughly the cache to be stored to disk. Most DBMSs are on-disk. The advantage of this storage approach is permanence. An in-memory data store, after all, can disappear if its hardwaren or software environment fails as in a power outage, for example. O In-memory storage t A newer approach is the In-Memoryin Database System (IMDS), which eliminates disk access and stores data in mainr memory, sending data to the hard disk only when speci- fied by the application. This all-in-memory storage means IMDSs are very fast, and that speed advantage has madeP them popular for real-time embedded systems. While elimi- nation of mechanical disk I/O is the main reason for in-memory databases’ speed, their streamlinedl designe usually also removes the internal handing off of multiple data copies as wellg as logical processes, especially those related to caching, that are no longer needed. This simpler design also gives IMDSs a smaller code footprint, which can be crucial for inembedded systems such as consumer electronics devices in which system resources are Sstrictly limited. r It is true that system failure can disrupt data stored in memory. But IMDSs compensate o for this risk with features such as maintaining the database in nonvolatile RAM, backing F up to secondary data storage (disk or flash), high availability schemes that manage redun- dant database copies, and transaction logging, in which changes (data transactions) are recorded in a transaction log file to enable recovery if needed. IMDSs include Polyhedra from Enea and McObject’s eXtremeDB.

API SQL API The SQL API is the universal enterprise database interface, and is also used in some embedded databases. This familiarity may shorten embedded system developers’ learn- ing curve if they’ve had prior experience with enterprise database systems. In addition, with its ability to express complex queries relatively succinctly, SQL can make a claim to greater coding efficiency.

But is SQL applicable to embedded systems? The answer may hinge on how complex the system’s queries are and how deterministic or predictable a response is required. SQL’s simplicity also makes it a “black box,” inside of which developers exercise little control over performance or resource consumption. Simply put, with SQL statements it is dif- ficult to see at the level of program execution what the DBMS is doing. This increases the chance that some change in software will cause the SQL optimizer, the software re- sponsible for deciding an SQL statement’s execution plan, to choose an inferior execution RSC# 34 @ www.embedded-computing.com/rsc plan. Unless queries are so complicated that SQL’s expressiveness lends an advantage, the SQL API’s overhead and inherent unpredictability may not be justifiable.

Navigational API Many embedded databases provide a navigational database API closely integrated with programming languages like C and C++. Embedded systems programmers tend to have advanced knowledge of these languages, and navigational APIs leverage these skills. In contrast to the high-level SQL, navigational APIs work on one record at a time, literally

34 / January 2007 Embedded Computing Design Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution navigating through the database, though much of the traversal occurs program- matically via looping and other structured programming techniques. Compared to SQL, navigational APIs are deterministic – it is known exactly how the data will be traversed when the application is com- piled – and are faster, executing at the speed of compiled C/C++ code while avoiding the parsing, optimization, and execution dynamic SQL requires.

Embedded database systems featuring SQL include Pervasive Software’s PSQL, McObject’s eXtremeSQL, and Solid Technology’s EmbeddedEngine. Figure 1 DBMSs with navigational APIs include McObject’s eXtremeDB, BerkeleyDB from Oracle, and Birdstep Technology’s RDM Embedded. Database systems with ly both SQL and proprietary navigational pro- gramming interfaces include McObject’s n eXtremeDB, Birdstep Technology’s RDM O Server, and Faircom’s c-treeSQL Server. t System architecture in Client/server r Client/server has been the predominant distributed computing model since the mid P ’90s. In this architecture, client applica- tions send requests to a database or other le shared application that resides on a server. g The World Wide Web is global client/ server computing. If you’re reading this in article online, your Web browser is likely the client with the article itself stored S on a Figure 2 Web server, perhaps on anotherr continent. Even when client and servero processes re- side on the same computer,F they are sepa- Advantages of client/server include the ability to appropriately size a network by install- rated by an interprocess communication ing the server software on a powerful computer with thin clients. In addition, client/server layer. architecture lends itself naturally to supporting larger numbers of users. A server will typically start more threads as the workload increases, so if it’s running on a multi-CPU Many embedded databases are based on box under symmetric multiprocessing, it can scale. In contrast, a database library offers this client/server model. But a handful no independent processing and relies on its host application to start additional threads. of others, which might be called more embedded or truly embedded, consist not High availability of client and server modules, but of code High availability is so mandatory for some telecommunications equipment that it has libraries (functions in the Java, C, and its own slang. Systems with the requisite fault tolerance are said to have 5-nines or C++ languages) actually embedded in 99.999 percent uptime. This presents challenges for databases. An in-memory database and compiled with a software application. can lose its stored content if memory fails, the server in client/server architecture can Companies including Birdstep, McObject, fail, and hard disk crashes can devastate an on-disk data store. and Empress provide such truly embedded databases. Database replication Maintaining replicated copies of data distributed over multiple nodes can achieve high This architecture is simpler, which means availability. Strategies for managing replicated databases include eager (synchronous) it has a smaller code size and is less prone and lazy (asynchronous) replication, based on their approach to pushing out changes from to defects because less complexity means the primary database to replica nodes. In eager replication (Figure 1), all updates are less risk of things going wrong. Removing applied to the primary and replica nodes as part of the original transaction. While ensur- interprocess communication eliminates a ing transaction integrity across nodes, the eager approach might exhibit longer resource performance barrier, and data storage and holding time and, consequently, execute more slowly. retrieval is further accelerated by elimi- nating server tasks such as the necessity In contrast, lazy replication (Figure 2) commits a transaction without waiting for the to manage sessions and connections and updates to complete on the replicas. This means shorter resource holding time and faster to allocate and de-allocate resources on behalf of clients.

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution performance. However, lazy replication is also more likely to lose a committed transac- tion if the primary node fails and transactions have not yet been propagated to the replicas. Employing a Generally speaking, lazy replication provides less transaction durability.

Another technology strategy for achieving high availability with predictable database time-cognizant eager response times is to apply a strict time requirement or processing deadline to eager replication. Employing a time-cognizant eager replication protocol ensures predictable replication protocol resource-holding time. If the primary database instance does not receive acknowledge- ment of its communication by a preset deadline, it assumes the uncommunicative replica ensures predictable database has failed, decommissions it, and continues with normal processing. Advantages bolster COTS database utilization Growth of intelligent, connected devices is soaring. Whether found in the shirt pocket, resource-holding time. automobile, airplane cockpit, or factory floor, these smart systems all incorporate embed- ded software and offer features that depend on searching, sorting, retrieving, and storing growing volumes of increasingly complex data. These are database management tasks, and to achieve them efficiently, embedded systems software developers are overcoming their propensity to build from scratch. Includingly a COTS database in an embedded systems project is not yet as commonn as using a commercial RTOS. ButO commercial databases’ time to mar- t ket, reliability, and resource consumption advantages are likely to drive the product rin category near that level of ubiquity. Steve Graves co- P founded McObject in 2001. As the com- le pany’s president and g CEO, he has spear- headed McObject’s in growth and helped the company attain S its goal of providing r database technology o that makes embedded systems smarter, F more reliable, and more cost effective to develop and maintain. Prior to McObject, Steve was president and chairman of Centura Solutions Corporation and VP of worldwide consulting for Centura Software Corporation. He also served as president and COO of Raima Corporation. Steve is a member of the advisory board for the University of Washington’s certificate program in Embedded and Real-Time Systems Programming.

To learn more, contact Steve at:

McObject LLC 22525 S.E. 64th Place, Suite 302 Issaquah, WA 98027 425-831-5964 [email protected] www.mcobject.com

RSC# 36 @ www.embedded-computing.com/rsc

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution y nl t O rin e P gl in r S Fo

RSC# 37 @ www.embedded-computing.com/rsc Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution Enterprise software in embedded applications

Enterprise applications go embedded

By Collin Bruce Collin discusses recent technology changes that urge more applications to use a relationaly database and presents some examples of the types of applications that couldn benefit.l mbedded systems are moving into the most everyday of roles. Today, embedded issues.O As embedded devices become in- systems reach far beyond the traditional “hidden” embedded systems such as int creasingly complex, hosting a number of airplanes, machinery, and combat devices, to a new arena of enterprise softwaren applications that share and process large Eand systems – in unorthodox situations such as medical exam rooms,i delivery amounts of data, the need to efficiently trucks, and even the repair technician’s toolkit. This fundamental change has raffected how manage this data and provide users a way developers approach building their systems. to easily search for the content they want P has become paramount. For example, a Thanks in large part to increasingly powerful and affordablee hardware, today’s devices simple file system is just not up to the task can run enterprise class software and store an ever-expandingl amount of data. No longer of managing the immense volume of data driving operations behind the scenes, many embeddedg devices are bringing enterprise today’s devices can hold, let alone the software into situations where it previously couldn not reach and performing tasks un- complex ways in which it needs to be ac- imaginable just a few years ago. i cessed. What is needed is a sophisticated S Relational Database Management System Think of the traditional appliance repair technician. Rather than being connected via (RDBMS) that meets the footprint con- phone to the parts warehouse rand other support venues, today’s repair technician often straints of embedded devices while retain- arrives equipped with a ohandheld device that includes everything from infor- ing as much functionality as possible. mation to diagnosticF tests to parts availability to billing software. While some of these devices may be directly linked with the home office’s network, increasingly, these devices Though typical RDBMSs simply search can serve as complete stand-alone systems that can provide virtually any information the alphanumeric data, embedded software repair technician might need for any given service call. vendors have developed innovative new search technologies suited to the charac- Evolution of embedded software design teristics of embedded handheld devices. What has changed in the embedded industry that has allowed such a shift? For years, The proliferation of Global Positioning embedded device developers had been concerned with saving the last byte and reduc- System (GPS) and other location-based ing CPU cycles in order to wring the most performance out of the hardware’s limited technologies in embedded devices has capabilities. But a convergence of more powerful hardware systems and a proliferation of prompted developers to accommodate COTS embedded operating systems, middleware, and data management systems are shift- just this type of spatial data using an SQL ing priorities. Until just a few years ago, embedded developers wrote their own operating query (see Figure 1). Applications that systems. It might have been an inefficient use of developers’ time, but it was rationalized can take advantage of the device’s loca- by the idea that a purpose-built, customized embedded operating system may run leaner tion can execute multidimensional search, and faster than a commercial one. integrating spatial data with other data held on the device. More recently, a similar trend has emerged in embedded middleware and embedded data management systems. Catering to embedded developers’ space and memory constraints, Think back to the appliance repair tech- these new middleware and data management systems conform to as small a footprint nician. Suppose the technician finishes as possible, some needing as little as a few hundred kilobytes. With the proliferation of a service call early, leaving enough time lean, COTS embedded infrastructure software, embedded developers realized an unprec- to fit in a standby appointment before the edented ability to focus on adding core value to their devices, rather than on “invisible” next scheduled one. With multidimen- infrastructure code. As the growing number of COTS systems has allowed developers to sional search, the technician could locate focus on their core expertise, developers have maximized the opportunity to differentiate the next appointment by searching for an their embedded devices. appointment between starting and ending locations, with restrictions on the distance Location-based applications of the prescribed route and the length of Advancements in hardware and COTS systems, however, have brought to bear new time spent at an added service call. Rather

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution Collin Bruce is director of marketing with Hitachi America’s Embedded Business Group. Collin has more than 25 years of experience in strategic sales and marketing with companies including Solid Information Technology, Communication Intelligence Corporation, Chordiant Software Amdahl, and Santa Cruz Operations. Collin holds a Bachelor’s of Technology degree in Applied Physics, Math from Brunel University in London.

To learn more, contact Collin at:

Hitachi America Embedded Business Group 750 Central Expressway • Santa Clara, CA 95050 408-970-4400 • [email protected] www.hitachi.us/entier

Figure 1 than calling into dispatch, the technician has saved time, reduced expenses, and ly possibly improved customer service. n Similar enterprise applications of this O type of search include the obvious, such as t delivery personnel, to more unique appli- cations, like combining location aware- in ness and social networking. Imagine an r office-based social networking system in which participants can opt to reveal their P location to the people in their working e network. Then it becomes possible to inte- l grate contact data and location data to ask g questions like, “Who from my extended n team is in the building right now?” Si Alphanumeric search getsr a help- ing hand Even in devices that o are not location aware, the need toF efficiently find alpha- numeric data poses difficult data retrieval issues. The scores of addresses, part num- bers, customers, and other data that can be stored in today’s handheld device often means users are forced to methodically enter each and every letter and number.

Today’s new embedded RDBMS tech- nologies now can incorporate incremental text searching, which assists the user by progressively executing the search, nar- rowing down the results with each addi- tional character entered, and guiding the user with a list of possible next letters in the search term. This has obvious uses for any application that involves searching for entries such as names, part numbers, ad- dresses, and similar types of static alpha- numeric data.

Watch this space New technologies that make embedded de- vices accessible for enterprise use emerge almost daily. As hardware becomes more powerful, developers will continue to bring innovative applications and devices for use in new settings. RSC# 39 @ www.embedded-computing.com/rsc

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution y nl t O rin e P gl in r S Fo

RSC# 40 @ www.embedded-computing.com/rsc

Embedded Computing Design / January 2007 ©2007 OpenSystems Publishing Not for distribution