非线性系统
灵敏度(控制系统)
光子学
算法
电子工程
计算机科学
正在测试的设备
集成电路
噪音(视频)
极小极大
自动测试设备
工程类
数学
内置自检
标量(数学)
干扰(通信)
稳健性(进化)
自动测试模式生成
有限集
模式(计算机接口)
散射参数
数字图形发生器
逻辑门
作者
Lawrence Schlitt,Priyank Kalla,Steve Blair
出处
期刊:Proceedings
[Institute of Electrical and Electronics Engineers]
日期:2026-10-01
卷期号:: 576-581
标识
DOI:10.1109/itc64303.2026.00082
摘要
Interference and resonance can hide a photonic defect under one stimulus and expose it under another. Our optimization-based automatic test pattern generation (ATPG) method ranks tests by their weakest predicted defect response. This maximin score uses a linear sensitivity model and second-order cone programs (SOCPs), with an analytical solution for scalar targets. Nonlinear simulation checks the selected tests and adds tests for missed samples. Test-set compaction reduces the test count while preserving coverage. Across three benchmarks, 31, 89, and 22 patterns detect all 4,614 samples that pass an initial detectability check. Of 19 modeled defect mechanisms, 14 pass this check somewhere in the benchmarks. With access, actions, and observation of the intended guided mode fixed, the assumed noise threshold lowers attainable coverage at the physical severity limits from 94.3-100.0% to 41.5-58.4%. These results concern finite samples in illustrative models; continuous-range guarantees and physical validation remain open.
科研通智能强力驱动
Strongly Powered by AbleSci AI