Analog to Digital Converters
Noise-Shaping SAR ADCs.
Introduced by the group in 2012, the noise-shaping SAR ADC technique has become an essential ADC design technique. This technique is widely adopted and enables some of the world’s most efficient ADCs.
SAR-Assisted Pipeline ADCs and Ring Amplifiers
Our work in 2010 introduced the SAR assisted pipeline ADC architecture revolutionizing the efficiency and performance of the pipeline ADCs. More recently we have harnessed ring-amplifiers to deliver record efficiency from the SAR-assisted-pipeline ADC architecture.
Continuous-Time and Bandpass Continuous-Time Delta-Sigma ADCs
Continuous-time (CT) delta-sigma ADCs are very appealing because they simplify input filtering and provide an easy-to-drive signal input. Our work improves both the performance and efficiency of CT delta-sigma ADC. We also introduce new compelling new architectures for bandpass conversion and multi-band digitization.
Introduced by the group in 2012, the noise-shaping SAR ADC technique has become an essential ADC design technique. This technique is widely adopted and enables some of the world’s most efficient ADCs.
SAR-Assisted Pipeline ADCs and Ring Amplifiers
Our work in 2010 introduced the SAR assisted pipeline ADC architecture revolutionizing the efficiency and performance of the pipeline ADCs. More recently we have harnessed ring-amplifiers to deliver record efficiency from the SAR-assisted-pipeline ADC architecture.
Continuous-Time and Bandpass Continuous-Time Delta-Sigma ADCs
Continuous-time (CT) delta-sigma ADCs are very appealing because they simplify input filtering and provide an easy-to-drive signal input. Our work improves both the performance and efficiency of CT delta-sigma ADC. We also introduce new compelling new architectures for bandpass conversion and multi-band digitization.
Machine learning and neural networks
RRAM Crossbar Computing
Memristors and memristor crossbar arrays have been widely studied for neuromorphic and other in-memory computing applications. For optimal system performance, however, it is essential to integrate memristor crossbars with peripheral and control circuitry. We build complete compute systems based on RRAM devices and assisted by mixed-signal circuitry.
Mixed-signal and Digital Spiking Neural Networks
We take inspiration from biology to build spiking neural networks with analog and digital circuits. The current revolution in machine learning is based on digital feed-forward CNN's comprising many layers. In contrast, the Locally Competitive Algorithm accomplishes similar tasks with fewer layers. LCA relies on a biologically-inspired analog Leaky Integrate-and-Fire (LIF) neuron model, suggesting an efficient analog implementation.
Memristors and memristor crossbar arrays have been widely studied for neuromorphic and other in-memory computing applications. For optimal system performance, however, it is essential to integrate memristor crossbars with peripheral and control circuitry. We build complete compute systems based on RRAM devices and assisted by mixed-signal circuitry.
Mixed-signal and Digital Spiking Neural Networks
We take inspiration from biology to build spiking neural networks with analog and digital circuits. The current revolution in machine learning is based on digital feed-forward CNN's comprising many layers. In contrast, the Locally Competitive Algorithm accomplishes similar tasks with fewer layers. LCA relies on a biologically-inspired analog Leaky Integrate-and-Fire (LIF) neuron model, suggesting an efficient analog implementation.
Wireless Systems and Digital Beamforming
Digital Wireless Beamforming
Beamforming is an essential technique for mm-wave communication (e.g., 5G). Large-scale beam-beamforming mitigates the substantial path loss in an mm-wave link, rejects spatial interferers, and significantly relaxes both the transmit power and receiver sensitivity requirements. The ability to generate multiple simultaneous beams is essential for multi-input multi- utput (MIMO) systems.
Clock Generation, Radar Chirps Generation and PLLs
Accurate, low-noise clocks are essential for wireless communication, analog-digital conversion, and radar. Our work explores new mostly-digital PLLs architectures aided by analog-to-digital converters. We use these PLLs for wireless links and for mm-wave radar chirp generation.
Beamforming is an essential technique for mm-wave communication (e.g., 5G). Large-scale beam-beamforming mitigates the substantial path loss in an mm-wave link, rejects spatial interferers, and significantly relaxes both the transmit power and receiver sensitivity requirements. The ability to generate multiple simultaneous beams is essential for multi-input multi- utput (MIMO) systems.
Clock Generation, Radar Chirps Generation and PLLs
Accurate, low-noise clocks are essential for wireless communication, analog-digital conversion, and radar. Our work explores new mostly-digital PLLs architectures aided by analog-to-digital converters. We use these PLLs for wireless links and for mm-wave radar chirp generation.
Systems and applications
Brain Prostheses
Deep brain stimulation (DBS) of the subthalamic nucleus is an effective therapy for numerous neurological disorders including Parkinson's disease, which affects more than 1.5m people in the US. Our work demonstrated the first closed-loop system for deep-brain stimulation.
Integrated Circuits for Nuclear Treaty Verification and for Detection of Nuclear Material
Verification of arms control treaties may require information barriers to protect sensitive data acquired during the verification process. These information barriers are commonly implemented in software, but must store and operate on data that are vulnerable to attack and tampering. This research resulted in a new secure ADC that is resilient to eavesdropping.
Real-Time Synthetic Aperture Imaging
Imaging radiometry over millimeter-wave frequencies (30–300 GHz) has the potential to transform security, health care, and remote sensing. Applications include sensing of concealed weapons, detection of skin cancer, and mapping of water in the atmosphere.
Deep brain stimulation (DBS) of the subthalamic nucleus is an effective therapy for numerous neurological disorders including Parkinson's disease, which affects more than 1.5m people in the US. Our work demonstrated the first closed-loop system for deep-brain stimulation.
Integrated Circuits for Nuclear Treaty Verification and for Detection of Nuclear Material
Verification of arms control treaties may require information barriers to protect sensitive data acquired during the verification process. These information barriers are commonly implemented in software, but must store and operate on data that are vulnerable to attack and tampering. This research resulted in a new secure ADC that is resilient to eavesdropping.
Real-Time Synthetic Aperture Imaging
Imaging radiometry over millimeter-wave frequencies (30–300 GHz) has the potential to transform security, health care, and remote sensing. Applications include sensing of concealed weapons, detection of skin cancer, and mapping of water in the atmosphere.