Archive

Panoptic Dynamic Voltage Scaling (PDVS)

Panoptic Dynamic Voltage Scaling (PDVS) is an exciting approach to ultra low power (ULP) design to reduce energy without sacrificing performance. The objective of PDVS is to dynamically scale energy of a digital circuit to meet real-time energy constraints and thus extend battery life. Consumers demand longer battery life. Some batteries in remote sensors such a forest or desert cannot be changed. Changing batteries within biomedical devices could have adverse effects on the patient.

New UltraThin SRAM Layout

Technology scaling has made it nearly impossible to maintain SRAM bitcell stability for dense cells. In technologies below 32nm, the SRAM bitcell size has begun to rise above the area predicted by the scaling trend followed over recent technology generations. We propose a new layout for the 6T SRAM that substantially increases the lithographic friendliness of the cell.

Advanced Power distribution methods and their effects on the on-chip power network

Aggressive scaling according to Moore’s Law, combined with the push to incorporate more functionality and higher performance on a single chip, has created the widely known power problem in modern ICs. Power consumption has become a major limiting factor in many designs. A number of solutions have been preliminarily explored to help improve the energy efficiency of chips, including clock gating, power gating, local voltage regulation, and dynamic voltage scaling. While theoretical benefits of these techniques are known, their full impact on the overall power distribution network remains uncertain. This project aims to address the impact of Panoptic Dynamic Voltage Scaling (PDVS) on the power distribution network and compare them against alternative low power techniques.

RFID Temperature Monitor for Lab Mice

The goal of the RFID Temperature Monitor for Lab Mice is to research an efficient way to monitor the well-being of mice in a vivarium. Medical research laboratories may house very large numbers of research animals, and each animal must be visually inspected every day. In addition to being a tedious task for research personnel, visual inspection is not the most effective method for accurately assessing the well-being of the animals. This project focuses on developing an implantable RFID-like tag that allows for remote, accurate, and continuous monitoring of lab animals.

Implementable Privacy and Security for RFIDs

This project explores algorithm, architectural and circuit ideas for implementation of Security Functions on RFID tags. An RFID tag harvests its power from the RF field that a reader sets-up while reading the tag. Typical power consumption, gate count and cost are below 10uW, 50K and 10 cents respectively. We explore energy efficient sub-threshold logic and memory design that can operate at sub-10uW power levels, while also delivering performance required for computationally intensive encryption algorithms.

Artificial Seal Whisker

Animal research activities conducted under NMFS permit 14535

The artificial seal whisker project is a joint effort with the University of Virginia’s Mechanical Engineering Department and the University of California Santa Cruz to detect and track underwater wakes using an array of bio-inspired sensors. Previous biological work found harbor seals are able to track wakes using only their whiskers. In this project, the seal whisker team is focused on understanding how seals sense wakes using their whiskers, designing a capacitance based whisker-like sensor, and designing the electrical backend printed circuit board for sensing, storing, and transmitting data. The biologically inspired sensor’s design is based on seal whiskers and previous effort in the field involving spider hairs and fish lateral lines. All components will be integrated in the Wake Information Detection and Tracking System (WIDTS) to be carried by a trained harbor seal for testing.

Circuit Techniques for Lowering SRAM VMIN

We investigate the application of these peripheral assist methods to reduce the operating voltage of SRAM, called VMIN. Lowering VMIN helps to decrease power consumption and also keeps pace with the dropping logic VDD, allowing for easier integration. Large scale 6T SRAM beyond 65nm will increasingly rely on assist methods to overcome the functional limitations associated with scaling and the inherent read stability/write margin trade off.

Dynamic stability

Static metrics for SRAM read and write stability do not take into account the dynamic nature of these operations. They are either pessimistic or optimistic in predicting cell failure due to the assumption of an infinite word-line (WL) pulse. In this project, we investigate TCRIT, the minimum WL pulse required for the cell to flip, as a dynamic measure of cell writability. We also explore models, both analytical and fast-simulation, that capture the statistics of dynamic margin in SRAM.

High Reliability SRAM for Extreme Environmental Conditions

This project investigates circuit and architectural techniques to improve the reliability of SRAM at extreme temperatures, particularly for automotive applications. The goal is to design and implement SRAMs that are capable of monitoring and adapting their circuits and micro-architectures to very wide range changes in environmental conditions.

Syndicate content