SHF: Small: Online Detection and Recovery from Electrostatic Discharge Induced Transient Errors

如果来自静电放电（ESD）的能量到达移动终端内的集成电路组件，则可能导致故障。通常需要多个设计周期，产品才能通过ESD资格测试，只有零星的，相对良性的，软故障发生。软故障的表现包括挂起应用程序、操作系统崩溃、信号保真度差、数据传输错误和瞬时显示闪烁。这些故障被称为软故障，因为它们可以从以下故障恢复，尽管可能需要操作员干预，例如，人们可能不得不关闭电源并再次打开。本项目将开发技术，用于演示基于操作系统的方法，以减轻ESD引起的软故障。预计这将减少重新设计硬件以满足ESD规范的需要，从而缩短移动的设备和其他电子产品的上市时间。这些技术也可以潜在地随后应用于减轻广泛的可靠性和安全性危害。该项目将涉及研究生和本科生，包括代表性不足的群体的成员，从而将有助于扩大信息和通信技术的劳动力。ESD引起的噪声的特征可以在集成电路（IC）中检测到，当这种噪声被检测到时，整个IC可能会发生逻辑错误。将演示ESD检测器可以放置在IC内部;当激活时，将向操作系统发送中断信号。为了功率和面积效率，将不会尝试识别IC内受影响的寄存器，也不会以不同方式处理不同类型的错误。相反，操作系统将被设计为对即将发生的错误的通知做出响应，并随后将系统回滚到先前的“已知良好”检查点。所提出的解决方案将在硬件中进行验证，这将允许ESD引起的错误在应用程序级别进行表征。为了进一步研究ESD引起的故障分析、诊断和恢复，还将进行ESD引起的错误的故障建模。

If energy from an electrostatic discharge (ESD) reaches the integrated circuit components inside a mobile device, failure may result. Multiple design cycles are generally needed before a product can go through ESD qualification testing with only sporadic, relatively benign, soft failures occurring. Manifestations of soft failures include a hanging application, an operating system crash, poor signal fidelity, data transmission errors, and momentary display flickering. These are referred to as soft failures because they are recoverable from, although operator intervention may be needed, e.g., one may have to turn the power off and on again. This project will develop techniques for demonstrating an operating system based approach for mitigating ESD-induced soft failures. It is expected to reduce the need for hardware to be redesigned to meet ESD specifications, thereby resulting in a shorter time-to-market for mobile devices and other electronic products. Such techniques can potentially subsequently be applied to mitigate a broad range of reliability and security hazards as well. The project will involve graduate and undergraduate students, include members of underrepresented groups, and will thus help enlarge the workforce in information and communication technologies.The signature of ESD-induced noise can be detected in an integrated circuit (IC) and when such noise is detected, logic errors are likely to occur throughout the IC. It will be demonstrated that ESD detectors may be placed inside an IC; when activated, an interrupt signal will be sent to the operating system. In the interest of power and area efficiency, there will be no attempt to identify the affected registers inside the IC, nor to handle the different types of errors differently. Instead, the operating system will be designed to respond to the notice of impending errors and will subsequently rollback the system to a previous "known good" checkpoint. The proposed solution will be validated in hardware and this will allow ESD-induced errors to be characterized at the application level. To enable further research in ESD-induced failure analysis, diagnosis and recovery, fault modeling of ESD-induced errors will also be undertaken.