CAREER: Robust Behavioral Fault Simulation Algorithms for Multilevel Simulation

职业：用于多级仿真的鲁棒行为故障仿真算法

• 批准号: 9733584
• 负责人: Karen Panetta
• 金额: $ 27.28万
• 依托单位: Tufts University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 1998
• 资助国家: 美国
• 起止时间: 1998-03-01 至 2003-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9733584&HistoricalAwards=false
• 关键词: CAREER; Robust; Behavioral; Fault; Simulation

This research focuses on improving simulation and test methodologies through the use of Concurrent Simulation (CS) and Multiple Domain Concurrent Simulation (MDCS) algorithms. This project develops an efficient behavioral fault simulation architecture that integrates well with structural and RTL models. The inherent ability of CS to represent and store only unique behaviors from the fault free experiment allows for excellent internal observation and comparative experimentation. Furthermore, the use of MDCS algorithms provides compression techniques for large simulations, overcoming memory bottlenecks while supporting the creation of multiple experiment scenarios. This allows a designer to vary parameters within any model and simulate it against other model configurations in a single simulation. Activities being explored under this grant include: (a) Developing algorithms for behavioral fault simulation, observability and signature traces using CS and MDCS; (b) Reduction of memory requirements through enhanced erratic experiment (infinite loop) detection, experiment compression using MDCS algorithms and fault dropping. (c) Evaluation of algorithms that allow the seamless integration of multilevel models.

本研究的重点是通过使用并发仿真（CS）和多域并发仿真（MDCS）算法来改进仿真和测试方法。该项目开发了一个有效的行为故障仿真架构，与结构和RTL模型集成良好。CS仅表示和存储无故障实验中的独特行为的固有能力允许进行出色的内部观察和比较实验。此外，MDCS算法的使用为大型模拟提供了压缩技术，克服了内存瓶颈，同时支持创建多个实验场景。这允许设计人员改变任何模型中的参数，并在单个仿真中针对其他模型配置对其进行仿真。在该赠款下探索的活动包括：（a）使用CS和MDCS开发行为故障模拟、可观测性和特征跟踪算法;（B）通过增强不稳定实验（无限循环）检测、使用MDCS算法压缩实验和故障丢弃来减少内存需求。(c)评估允许无缝集成多级模型的算法。