Dynamic Modeling and Analysis for Design of Memristive and Static Random Access Memories
MetadataShow full item record
Nowadays, the trend of modern memory technology is going towards the following directions: (1) look for new nonvolatile devices; (2) keep scaling down the existing volatile devices. Although nonvolatile devices enable to switch off its power supply to further suppress standby power, the down sides are the low switching speed and the complicated dynamic cell characteristics. On the other hand, researchers are looking to scale down SRAM since it is the most reliable and fast. However, the SRAM suffers read and write failure due to lack of good stability optimizing metric. To tackle the above mentioned problems, this work first introduces a promising nonvolatile device called Memristor, which is said to be possible to replace our memory devices now. By starting from basic memristor device equations, this work aims to develop a comprehensive set of properties and design equations for memristor based memory. The introduced schemes are specifically targeting key device properties relevant to memory operations. Using the discovered properties, a simple design of read/write circuits is investigated. In the second part of this work, SRAM stability analysis is focused. SRAM verification and stability analysis has become an essential task to investigate soft-errors. This work aims to extend the SNM to a new era. Based on the introduced Region-Analysis in this work, SRAM stability can be explained using bifurcation theory, and closed form expression can be derived. The derived expression provides physical characterization of SRAM noise tolerance property; thus has potential to provide needed design insights. Overall, dynamics of memristor and SRAM are strongly emphasized. The derived memristor properties reveals that the memristor state change requires some time; it indicates that the memristor-based memory needs some “critical time” to flip the logic. Similarly to the SRAM, the SRAM write operation not only needs the injected current over a “critical current” but also need to maintain for some “critical time”. In short, both memristor-based memory and SRAM show the timely manner for read/write operation. Furthermore, the developed analytical formulae are able to reveal the dynamic aspect on memory read/write operations which address the key concern for modern memory technology.
Ho, Yenpo (2014). Dynamic Modeling and Analysis for Design of Memristive and Static Random Access Memories. Doctoral dissertation, Texas A & M University. Available electronically from