CMStatistics 2020: Start Registration
View Submission - CMStatistics
Title: Phase-type distributions in a vector Markov process to analyse random telegraph noise in resistive memories Authors:  Christian Acal - University of Granada (Spain) [presenting]
Juan Eloy Ruiz-Castro - University of Granada (Spain)
Ana Maria Aguilera - University of Granada (Spain)
Juan Bautista Roldan - University of Granada (Spain)
Abstract: Advanced statistical techniques are key tools to model complex physical and engineering problems in many different areas of expertise, such as the field of Resistive Random Access Memories (RRAMs). One of the most important aspects to consider, prior to the massive industrialization, is the Random Telegraph Noise (RTN). This issue is a great concern because it can affect the correct operation of a device. A device can emit signals originated thanks to disturbances produced by several traps that provoke current fluctuations. This process can be represented as a process that evolves over time, going through multiple states. In this line, a vector Markov process, by considering macro-states in order to analyse, model and study the evolution, is introduced. Multiple measures of interest are worked out. So far, the usual statistical analysis performed on the sojourn times makes use of the exponential distribution; however, sometimes its fit is not accurate and therefore, another via must be considered. In this point, we propose a novel approach based on Phase-Type Distributions (PHD) where each stage of the device is a macro-state, in a way that the sojourn time distribution for each macro-state is PH distributed.