Design and Implementation of Embedded Configuration - Power Integrated
As the application area of embedded system widely broadened in recent years, its working environments are more complex and constraint. Thus it needs much more dedicate and systematic architecture design for meeting the requirements of the application such as real time, concurrence, efficiency, etc. Processes fulfilling different tasks are not simply added in together for an integrated system The relationship amongst processes needs to be considered more carefully. For embedded software system, having many peripherals is quite normal. Device drivers are responsible for controlling the behaviors of the peripherals. Since devices are rather independent of center module, and have much more uncertainty time cost because of different tasks such as data collection, status control or data preprocess, the drivers then are converters that connect slower speed devices with higher performance center modules. In conventional architecture, drivers usually directly interact with the center modules .When the number of driver tasks is enormous and much more synchronization constraints existing for those parallel processes, such a directly coupling structure seems not very suitable, which may lead to a disaster for the performance of system. Since directly coupling architecture is not fit for solution of the previous problem, software agent design seems proper for effectively controlling multi-access and parallel processing [6][7]. Software agent can be defined as referring to a component of software and/or hardware which is capable of acting exactingly in order to accomplish tasks on behalf of its user [8]. In this issue, agent is deputy for the peripheral drivers to concurrency and synchronization. It rearranges the multi-access sequence for supporting as many parallel processing requests, and with specific scheduling algorithm [9] [10] such as Rate Monotonic Scheduling (RMS) algorithm [11] [12] [13] for satisfying time constraints of tasks. The indirect coupling architecture of agent can well adapt to meeting the requirements of parallel processing for better performance, and it has good extensibility for future multi-agent applications. This issue is organized as follows. Section 2 introduces software agent architecture design for parallel processing embedded applications. View More
many peripherals is quite normal. Device drivers are responsible for controlling the behaviors of the peripherals. Since devices are rather independent of center module, and have much more uncertainty time cost because of different tasks such as data collection, status control or data preprocess, the drivers then are converters that connect slower speed devices with higher performance center modules. In conventional architecture, drivers usually directly interact with the center modules [4] [5].When the number of driver tasks is enormous and much more synchronization constraints existing for those parallel processes, such a directly coupling structure seems not very suitable, which may lead to a disaster for the performance of system. Since directly coupling architecture is not fit for solution of the previous problem, software agent design seems Read More proper for effectively controlling multi-access and parallel processing . Software agent can be defined as referring to a component of software and/or hardware which is capable of acting exactingly in order to accomplish tasks on behalf of its user [8]. In this issue, agent is deputy for the peripheral drivers to concurrency and synchronization. It rearranges the multi-access sequence for supporting as many parallel processing requests, and with specific scheduling algorithm [9] [10] such as Rate Monotonic Scheduling (RMS) algorithm [11] [12] [13] for satisfying time constraints of tasks. The indirect coupling architecture of agent can well adapt to meeting the requirements of parallel processing for better performance, and it has good extensibility for future multi-agent applications. This issue is organized as follows. Section 2 introduces software agent architecture design for parallel processing embedded applications. Section 3 applies RMS algorithm for scheduling concurrency tasks by agent.
No comments:
Post a Comment