基于FPGA的微控制器芯核软错误敏感性评估方法研究
发布时间:2018-08-14 16:00
【摘要】:随着半导体工艺技术的持续改善,使得芯片的集成度、工作频率、功耗等指标参数不断提高。但是特征尺寸的缩小也使得电路翻转所需的临界电荷降低,导致微控制器对内部和外部环境变得更加敏感。另一方面,随着微控制器广泛应用在航空航天等特殊应用领域的苛刻任务中,要求微控制器需具有更高的可靠性。在太空环境中,微控制器很容易受到单粒子效应的影响,使微控制器中的逻辑单元发生翻转而产生故障,严重影响微控制器的可靠性。故障注入作为一种灵活方便、便宜有效的方法广泛应用于可靠性评估领域。在故障注入的实现中,常用的有基于物理的、仿真的或者FPGA的故障注入方法。为了自动快速地分析微控制器对软错误的敏感性,该论文对基于FPGA的微控制器芯核软错误敏感性评估方法进行了研究。首先,对微控制器故障注入方法的研究现状、FPGA内部结构、以及开发语言进行简单介绍。其次,利用Verilog HDL对PIC16F54微控制器芯核进行了设计,包括取指、译码、ALU、寄存器组、堆栈以及ROM,并在FPGA开发板上进行验证。然后,利用该微控制器芯核作为实验对象,提出一种基于FPGA故障注入的微控制器敏感性评估方法。在FPGA芯片上同时运行有故障和无故障的两个微控制器,并充分利用FPGA的并行性,把故障注入控制、故障分类、故障列表等模块均在硬件上实现。自动地完成全部存储位的故障注入,并且能够快速的对故障注入结果进行分析。最后,注入大约30万个软错误,把敏感性结果与其他的方法相比较,验证本文方法的正确性。另有实验数据表明,使用该方法进行故障注入及敏感性分析所需的时间比软件仿真的方法提高了 4个数量级。
[Abstract]:With the continuous improvement of semiconductor technology, the chip integration, working frequency, power consumption and other parameters are constantly improved. However, the reduction of the characteristic size also reduces the critical charge required for the circuit to flip, which makes the microcontroller more sensitive to the internal and external environment. On the other hand, with the wide application of microcontrollers in the demanding tasks of special applications such as aerospace, microcontrollers are required to have higher reliability. In the space environment, the microcontroller is easily affected by the single particle effect, which causes the logic unit in the microcontroller to flip and causes the fault, which seriously affects the reliability of the microcontroller. As a flexible, cheap and effective method, fault injection is widely used in the field of reliability evaluation. In the implementation of fault injection, there are physical, simulation or FPGA based fault injection methods. In order to analyze the sensitivity of microcontroller to soft error automatically and quickly, this paper studies the sensitivity evaluation method of core soft error based on FPGA. Firstly, the internal structure of FPGA and the development language are briefly introduced. Secondly, the core of PIC16F54 microcontroller is designed by using Verilog HDL, including the reference, decoding Alu, register group, stack and rom, and verified on the FPGA development board. Then, using the core of the microcontroller as the experimental object, a sensitivity evaluation method of the microcontroller based on FPGA fault injection is proposed. Two microcontrollers with and without faults are run on FPGA chip at the same time, and taking full advantage of the parallelism of FPGA, the modules of fault injection control, fault classification and fault list are implemented in hardware. The fault injection of all storage bits is completed automatically, and the fault injection results can be analyzed quickly. Finally, about 300,000 soft errors are injected, and the sensitivity results are compared with other methods to verify the correctness of the proposed method. Other experimental data show that the time required for fault injection and sensitivity analysis using this method is 4 orders of magnitude higher than that of software simulation.
【学位授予单位】:合肥工业大学
【学位级别】:硕士
【学位授予年份】:2017
【分类号】:TN791
本文编号:2183383
[Abstract]:With the continuous improvement of semiconductor technology, the chip integration, working frequency, power consumption and other parameters are constantly improved. However, the reduction of the characteristic size also reduces the critical charge required for the circuit to flip, which makes the microcontroller more sensitive to the internal and external environment. On the other hand, with the wide application of microcontrollers in the demanding tasks of special applications such as aerospace, microcontrollers are required to have higher reliability. In the space environment, the microcontroller is easily affected by the single particle effect, which causes the logic unit in the microcontroller to flip and causes the fault, which seriously affects the reliability of the microcontroller. As a flexible, cheap and effective method, fault injection is widely used in the field of reliability evaluation. In the implementation of fault injection, there are physical, simulation or FPGA based fault injection methods. In order to analyze the sensitivity of microcontroller to soft error automatically and quickly, this paper studies the sensitivity evaluation method of core soft error based on FPGA. Firstly, the internal structure of FPGA and the development language are briefly introduced. Secondly, the core of PIC16F54 microcontroller is designed by using Verilog HDL, including the reference, decoding Alu, register group, stack and rom, and verified on the FPGA development board. Then, using the core of the microcontroller as the experimental object, a sensitivity evaluation method of the microcontroller based on FPGA fault injection is proposed. Two microcontrollers with and without faults are run on FPGA chip at the same time, and taking full advantage of the parallelism of FPGA, the modules of fault injection control, fault classification and fault list are implemented in hardware. The fault injection of all storage bits is completed automatically, and the fault injection results can be analyzed quickly. Finally, about 300,000 soft errors are injected, and the sensitivity results are compared with other methods to verify the correctness of the proposed method. Other experimental data show that the time required for fault injection and sensitivity analysis using this method is 4 orders of magnitude higher than that of software simulation.
【学位授予单位】:合肥工业大学
【学位级别】:硕士
【学位授予年份】:2017
【分类号】:TN791
【参考文献】
相关期刊论文 前6条
1 黄正峰;钱栋良;梁华国;易茂祥;欧阳一鸣;闫爱斌;;65nm工艺下单粒子加固锁存器设计[J];计算机辅助设计与图形学学报;2016年08期
2 黄正峰;陈凡;蒋翠云;梁华国;;基于时序优先的电路容错混合加固方案[J];电子与信息学报;2014年01期
3 武振平;付方发;肖立伊;;基于VHDL故障注入的处理器敏感性分析[J];微电子学与计算机;2012年10期
4 黄海林;唐志敏;许彤;;龙芯1号处理器的故障注入方法与软错误敏感性分析[J];计算机研究与发展;2006年10期
5 董剑,曲峰,刘宏伟,崔刚;嵌入式故障注入器HFI-4的研究与设计[J];小型微型计算机系统;2003年12期
6 孙峻朝,王建莹,杨孝宗;故障注入方法与工具的研究现状[J];宇航学报;2001年01期
相关博士学位论文 前1条
1 周婉婷;辐照环境中通信数字集成电路软错误预测建模研究[D];电子科技大学;2014年
相关硕士学位论文 前4条
1 洪映森;基于FPGA的故障注入系统研究[D];哈尔滨工业大学;2014年
2 江丽君;数字集成电路故障模型研究及故障注入平台设计[D];哈尔滨工业大学;2013年
3 吴杰;基于VHDL的故障注入工具的研究与实现[D];哈尔滨工业大学;2013年
4 路遥;基于HDL的故障注入工具的研究与实现[D];国防科学技术大学;2008年
,本文编号:2183383
本文链接:https://www.wllwen.com/kejilunwen/dianzigongchenglunwen/2183383.html