Entertainment & Media

A Survival Kit: Adaptive Hardware/Software Codesign Life-Cycle Model

A Survival Kit: Adaptive Hardware/Software Codesign Life-Cycle Model
of 3
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  A Survival Kit: Adaptive Hardware/Software Codesign Life-Cycle Model T he cost of a digital tele-vision such as an LG 42-inch LCD—about $1,800 in June 2007—had plummeted to about $1,000 by December 2007. Within a half year of the initial marketing, 44.4 percent was wiped from the price tag. How can a consumer electronics com-pany survive this cutthroat market competition?Many CE makers that produce not only digital TVs but also mobile phones, digital cameras, digital video recorders, MP3 players, or other CE embedded systems are struggling to resolve this problem. They reduce expenses by lowering manufacturing costs, improving productivity, and using other management methods. However, few CE producers have considered using hardware/soft-ware codesign technology to lower production costs. CE embedded system makers can use hardware/software codesign technology to replace expensive hardware com-ponents with inexpensive software components. Manufacturers can dra-matically reduce cell phone handset production costs, for example, by replacing radio frequency hardware chips with software programs such as software-defined radio. Few studies have examined the cost impact of a codesign decision. Existing hardware/software codesign research focuses only on optimizing the hardware and software combi-nation in the design phase, but the optimized combination remains the same thereafter, even though market conditions may change. We introduce an adaptive life-cycle model of hardware/software codesign to optimize the production cost of CE embedded systems by modifying various combinations of hardware and software components over the life cycle, according to market condi-tions such as cost, price, revenue, and time-to-market requirements. TYPICAL HARDWARE/SOFTWARE CODESIGN DEVELOPMENT PROCESS The embedded system develop-ment process is unique, since two different processes—hardware development and software develop-ment—are considered in combination. Arnold S. Berger (  Embedded System  Design , CMP Books, 2002) presented a typical embedded system develop-ment process, which consists of seven phases, as Figure 1a shows. Hardware/software partitioning (phase 2) and integration (phase 5) are distinctive features that distinguish this process from general software development processes. Overall, the industry employs a development process that is similar to a typical embedded system develop-ment process, especially in terms of the phased structure. Phases 3 and 4, which involve design and implemen-tation, are slightly different, as Figure 1b depicts. Most vendors develop their Dong-hyun Lee and Hoh Peter In, Korea University   Keun Lee, Samsung Electronics   Sooyong Park, Sogang University   Mike Hinchey, Lero—The Irish Software Engineering Research Centre An adaptive hardware/software codesign model enables consumer electronics companies to reduce production cost and maximize profits by progressively modifying codesign options over the life cycle. COMPUTER  100 SOFTWARE TECHNOLOGIES Published by the IEEE Computer Society 0018-9162/09/$25.00 © 2009 IEEE    own software components but rely on hardware suppliers for hardware designs. In other words, hardware component design is based on the production plan, while the software design is based on the development plan. Moreover, hardware imple-mentation (with buyout options) is more rapid than in-house software implementation. Consequently, hard-ware and software development lead times and costs differ. A strategy is necessary for adaptive optimization of hardware/software partitioning options over the life cycle. ADAPTIVE LIFE󰀭CYCLE PROCESS The adaptive life-cycle process of hardware/software codesign involves reducing the production cost and determining an optimized transition sequence via the codesign life-cycle transition sequence. This method is highly applicable to the embedded system field, since it changes hard-ware/software codesign options for a product to reduce production costs. Figure 2 is a conceptual diagram of the adaptive hardware/software codesign life-cycle sequence of a CE product. There are three types of components in the product: software-only, hardware-only, and components that can be implemented as both hardware and software. Replacing hardware components with software components reduces production costs. The hardware cost is constant in the production stage, after release; the software cost is almost zero during the same pro-cess. For this reason, we assume that a transition can occur in only one direction, from hardware to software. Figure 2 shows the transition from hardware-intensive Design A to soft-ware-intensive Design D over the life cycle. For simplicity, we also assume that there are no other parameters affecting a component transition from hardware to software. After release, the optimized hard-ware/software codesign life-cycle transition sequence determines the design transition sequence in a manner that maximizes revenue from a product in certain production environments—for example, market conditions and the product devel-opment schedule. It consists of two steps:  partitioning  and transition .The hardware/software partition-ing step generates design options for a product and determines the initial version of the product with certain input parameters—for example, hardware/software components, time to market, and hardware/ software development lead time—      P     h    a    s    e     1  –    p    r    o     d   u    c    t     i    o    n    s    p    e    c     i         c    a    t     i    o    n     P     h    a    s    e     2  –     H     W     /     S     W     p    a    r    t     i    t     i    o    n     P     h    a    s    e     3  –     i    t    e    r    a    t     i    o    n    a    n     d     i    m    p     l    e    m    e    n    t    a    t     i    o    n     P     h    a    s    e     5  –     H     W     /     S     W     i    n    t    e    g    r    a    t     i    o    n     P     h    a    s    e     6  –    a    c    c    e    p    t    a    n    c    e    t    e    s    t     i    n    g     P     h    a    s    e     7  –    m    a     i    n    t    e    n    a    n    c    e    a    n     d   u    p    g    r    a     d    e Phase 4–detailed HW/SW designProduct releaseSW design path activitiesHW design path activitiesProduct releaseSW design path activitiesHW design path activitiesPhase 4Supplier managementDesign andsimulationDesign and implementation (b)(a)      P     h    a    s    e     3     P     h    a    s    e     2     P     h    a    s    e     1     P     h    a    s    e     5     P     h    a    s    e     6     P     h    a    s    e     7 Figure 1.  Differences in hardware and software development lead times in the embedded system industry: (a) typical embedded system development process and (b) real-world development process in the embedded system industry. … Design spaceAdaptive selection in the life-cycle model Partitioning phaseDesign ADesign BDesign CDesign DFinal transitionSecond transitionFirst transition SWSWSWHWSWSWSWSWSWSWSWSWSWSWSWSWSWSWHWHWHWHWHWHWHWHWHWHWHWHWHWHWSWHWHWHWHWHWHW Software-only componentHardware-only componentHardware or software enabled component SWHWSWHW Figure 2. Conceptual diagram of the adaptive hardware/software codesign life-cycle transition sequence. 101 FEBRUARY 2009  build your career IN COMPUTING www.computer.org/buildyourcareer  as well as the costs of developing hardware/software, based on hard-ware/software cost models. Among these parameters, time to market is the main market constraint on the hardware/software partitioning process, while developing and pro-ducing hardware and software are the main cost factors. We have adopted Cocomo II (B. Boehm,  Software Cost  Estimation   with Cocomo II  , Prentice Hall, 2000) as the model for estimat-ing the software costs, based on the assumption that the hardware costs represent the aggregate price of the hardware components.The hardware/software transition step determines a design transition sequence with the options generated in the hardware/software partitioning step. This step provides a road map for the optimization of the design tran-sition sequence for maximization of revenues. Figure 3 shows the total cost of design versions over the life cycle, as proposed in Figure 2. Estimating the total cost of each version requires information about market conditions such as the time to market and the estimated number of product sales. The first transition occurs when the development of design B is com-plete. The development lead time of software is generally longer than that of hardware, hence it takes time to replace hardware components with software components. The second and final transitions occur when designs C and D are complete. The yellow dotted line in Figure 3 indicates the transition sequence over the life cycle for cost minimization. Thus, the life-cycle transition plan is to adopt design A for months 1-3, design B for months 3-5, design C for months 5-6, and design D for months 6-9. With the road map of the design transition sequence, embedded system vendors can manage the development and production plan over the life cycle with respect to price cuts. P revious studies on hardware/software codesign mainly focused on development independent of the market require-ments, concentrating on various aspects of resource utilization such as the processor, memory, size, power consumption, and timing. The adaptive hardware/software codesign life-cycle transition model considers not only the development cost and lead time but also the production cost and product life cycle. It can maxi-mize profits for an embedded system product, even in cutthroat, red-ocean market conditions. Currently, we are investigating several industry projects funded by the Information Technology Research Center research program of the Institute of Information Tech-nology Advancement to estimate the return on investment of the proposed model. Contact Hoh P. In (hoh_in@korea.ac.kr) for further information on research results.  Dong-hyun Lee  is a PhD candidate at Korea University. Contact him at tellmehenry@korea.ac.kr.  Hoh Peter    In  is an associate professor at Korea University. Contact him at hoh_in@korea.ac.kr.  Keun Lee  is a principal engineer at  Samsung Electronics. Contact him at  gskeun@gmail.com.  Sooyong Park  is a professor at Sogang University. Contact him at sypark@sogang.ac.kr. Mike Hinchey , Software Technologies column editor, is the codirector of  Lero—The Irish Software Engineer-ing Research Centre. Contact him at mike.hinchey@lero.ie. Time (months)123456789      T    o    t    a     l    c    o    s    t     (     d    e   v    e     l    o    p    m    e    n    t    +     p    r    o     d   u    c    t     i    o    n     ) First transitionSecond transitionFinal transitionInitial versionreleased The road map of transitionDesign ADesign CDesign BDesign D(design A �  B �  C �  D) Figure 3. Road map of the transition sequence in the life cycle. The life-cycle transition plan is to adopt design A for months 1-3, design B for months 3-5, design C for months 5-6, and design D for months 6-9. COMPUTER  102 SOFTWARE TECHNOLOGIES
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks