基于概率统计的单粒子多单元翻转信息提取方法

航空航天电子系统对电子器件选型评估时需考虑器件的多单元翻转（MCU）情况，而MCU信息提取面临的最主要困难是缺少器件的版图信息。本文提出一种基于概率统计的单粒子MCU信息提取方法，其可在无版图信息条件下以较高精度提取单粒子翻转（SEU）实验数据中的MCU信息。该方法通过统计分析SEU实验数据中不同翻转地址间的按位异或和汉明距离以提取MCU模板，然后利用该模板提取MCU信息。采用一款位交错SRAM器件的重离子实验数据对上述方法进行了验证，结果表明，该方法能以较高的精度提取实验数据中的MCU信息。该方法可省去对器件进行逆向工程的时间和成本，提高科学研究和航空航天器件选型效率。

Statistical Method for Single Event Multiple Cell Upset Extraction

In the evaluation of electronic device for aerospace and avionics systems, the multiple cell upset (MCU) of the device should be considered. The main difficulty in MCU extraction is the lack of layout information. In this paper, a statistical method was proposed to extract MCU from single event upset (SEU) without layout information. Firstly, a MCU template was extracted by means of statistical analysis of the binary exclusive OR and the binary Hamming distance between the logical addresses of different upsets. Then the MCU could be extracted by using the MCU template. Heavy ion test data of an interleaved SRAM were used to verify the method. The results show that MCU can be extracted with a high accuracy. Using the proposed method, the time and cost of reverse engineering can be saved, and the efficiency of related scientific research can be significantly improved.