Estimating aging-related degradation and failure of embedded nanoscale circuits (ICs) before they occur is crucial for the creation of aging prevention / mitigation policies and in turn to prevent unexpected circuit failures in the field. Effectively, IC operating conditions can be monitored in real time to predict aging degradation and timing mistakes owing to the device's aging. Existing techniques only take into account certain operating conditions (e.g., temperature or workload). In this article, I propose a fresh method of real-time IC aging prediction. This method extends prediction systems to a comprehensive model. It is easy to put any time-varying dynamic working conditions suitable for aging prediction into this model and get a prediction. By using a prediction model for machine learning and the idea of equivalent aging time, it shows that our approach performs better than present methods with regard to aging precise under different operating circumstances. Index Terms—Real-time IC aging prediction, equivalent aging time, negative/positive bias temperature instability, hot carrier injection, machine learning.
