Various natural language processing (NLP) tasks necessitate deep models that are fast, efficient, and small based on their ultimate application at the edge or elsewhere. While significant investigation has furthered the efficiency and reduced the size of these models, reducing their downstream latency without significant trade-offs remains a difficult task....
Simultaneous speech-to-text translation remains a difficult yet important problem for modern machine learning models whereby a text translation is generated concurrently with receiving partial speech inputs. One state-of-the-art simultaneous speech-to-text model is the augmented memory transformer whose encoder breaks a speech input into fixed-size overlapping segments composed of left, right,...
High Performance Computing can find ubiquitous applications in the industry. HPC-applications are specifically designed to take advantage of the parallel nature of the computing systems which is often enabled by Multi-core/Many-core architectures. With this advent of Multi-processors in the mainstream systems, inter-core communication has been one of the major challenges...
Portable, high power efficiency communication devices is a growing market in the semiconductor industry. Analog-to-digital converters (ADC) are key interface that are used to digitize the sensed information. Recently, digital techniques have been proposed to improve analog building block power efficiency in sub-micron technologies. This research focuses on mixed signal...
The increasing demand for higher data-rates in spectrum-deficient indoor WiFi networks calls for the adaptation of hybrid systems, consisting of RF and optical wireless channels. The novel hybrid WiFi Free-Space Optical (WIFO) system was introduced to enhance the wireless capacity of indoor WiFi networks. In this thesis, an optical wireless...
Traditional localization techniques rely on triangulation or trilateration, where in a set of three or more stationary known locations is used to estimate a “client” position. For inertial navigation, these techniques can estimate client positions merely using the measured data from tri-axial accelerometers and gyroscopes. However, the use of double...
Wearable electronic devices are becoming increasingly widespread and are being used to fulfill a variety of needs. These devices range from purely infotainment type systems such as smartwatches and fitness oriented devices to specialized medical devices such as insulin pumps and heart-rate monitors. A challenge faced with all of these...
The CMOS two-stage Operational Transconductance Amplifier (OTA) has been a key enabler for mixed-signal IC design for nearly four decades . This research focuses on a modified two-stage CMOS OTA that features load-pole cancellation (LPC); i.e., the resulting architecture is essentially a two-stage CMOS OTA with no load capacitance. The...
Ring amplification has emerged as an efficient technique to drive large capacitive loads in switched capacitor circuits. We propose circuit techniques to demonstrate the first application of a ring amplifier in a non-capacitive feedback system of a LDO. These techniques enable a simple cap-less LDO structure in 180nm CMOS that...
Low-power receivers (RX) with 100$\mu W$-scale power consumption can enable several power/energy-constrained IoT applications. However, achieving sensitivity, interferer tolerance and wide operating range with low power presents a challenge for existing architectures, particularly those constrained to highly integrated solutions without high-Q off-chip components. Existing solutions rely heavily on high quality...