My research interests are broadly in the area of high-performance integrated circuits and systems aimed towards creating the next-generation cutting-edge technology for a diverse set of applications from communication to personalized healthcare and medical diagnostics. I believe, in near future, a host of highly impactful technological innovations will come through a close alliance of multiple scientific disciplines such electrical engineering, applied physics and mathematics, material science, chemistry and biology. Such research topics, which potentially have high positive impact on society and stretch the boundaries between electrical engineering and applied sciences, motivate me. I find such spaces much less explored, and equipped with tools from core disciplines, and with understanding of underlying mathematics and physics, the solutions and methods are often only limited by imagination and creativity. My research approach is to leverage the strengths of concepts and techniques across disciplines and blend them to create novel and high-performance systems with a diverse set of applications from wireless communications, biomedical sensing, imaging and radar. Currently, my intellectual focus is on these broadly defined topics
Imagine the possibilities of a billion transistors with cut-off frequencies progressing towards THz frequencies. The unique ability to integrate very fast transistors, along with complex signal processing all in one chip, has opened up a portion of the electromagnetic spectrum in the mm-Wave and THz frequency ranges, previously unavailable to integrated technology. This can not only miniaturize bulky, expensive, custom optics-based instrumentation existing in this range, but the versatility of silicon technology can open doors to high-impactful new applications in ultra-high speed wireless communication, high-resolution imaging to spectroscopic analysis of biomolecules, non-invasive biomedical imaging besides the more conventional use in astronomy, condensed-matter physics and security. The path-breaking innovations will not happen piggybacking on technology advancements with scaling, but through new architectures, design methodologies that can fully leverage the true potential of such versatile integrated platforms.
We demonstrated such an approach in our concept of "Distributed Active Radiation," that enabled power generation, frequency multiplication, radiation and filtering at THz frequencies simultaneously in a small silicon footprint. It revisits the fundamental concepts of wave propagation and radiation from a new perspective by combining concepts from electromagnetics, circuits, antenna and nonlinear dynamics through a holistic approach. We demonstrated the first integrated, 4x4 phased-array with onchip radiators in CMOS, radiating nearly 10 mW of Effective-isotropic radiated power at 280 GHz (1000x previously recorded in CMOS), and 80 degrees of digital beam-scanning in 2D space. We also demonstrated the world's most sensitive THz camera without silicon lens or post-processing, enabling the world's first all-silicon THz imaging system with a CMOS source.
We proposed a Inverse Maxwel design philosophy in the evolution of the DAR, where we directly synthesize the surface currents responsible to create the desirable electromagnetic fields. This is in contrast to the traditional, partitioned way of designing where we are limited to the set of known classical circuit blocks. This opens up a new and broader design space, where concepts from different fields can synergistically merge together to create an integrated chip surface on which one can dynamically synthesize, control and manipulate the electromagnetic near-fields. This ultimate realization of a versatile, reconfigurable 'electromagnetic carpet' could be in radio-frequencies for manipulation of dielectric or magnetic nanomaterials, at THz frequencies for onchip spectroscopy or even at optical frequencies. A host of different fields from electromagnetism, optics, plasmonics, metamaterials, circuits can merge together to create radically new integrated, low-cost devices in biosensing and bioactuations.
Silicon transistors have shrunk into dimensions, where variation in the countable number of dopant atoms cause significant change in performance and behavior. Added to this, parasitic and process variations, modeling inaccuracies, aging, environmental changes cause a severe degradation in system performance.
'Self-healing' or reconfigurable CMOS integrated CMOS systems constantly monitors the `health' of the circuits and sub-systems within the chip and takes actuation measures to correct for failure. where the optimization of performance and control of all sensors and actuators, is done in the digital domain, where transistors come almost free of cost. Such concepts can be extended to programmable, flexible RF transceivers which are spectrally aware, can dynamically switch between available spectral bands and where communication standards can be programmed post-fabrication. Design of such transceivers, which can also meet all standards, can be extremely challenging and leaves plenty of room for innovations at the architectural level. We demonstrated self-healing in a fully integrated mm-Wave power amplifier capable of onchip improvements in gain, power dissipation and efficiency at backoff and in presence of process variations and load-mismatch events and even when some of the amplifier transistors are intentionally destroyed (RFIC 2012 best paper award).
I am interested in understanding and analysis of various fundamental problems such as limits of efficiency of radiation and surface-wave fields inside silicon substrates, inverse problems such as synthesis of currents for given electromagnetic fields, minimal loss in arbitrary impedance matching networks, as well as limits of noise achievable in various circuit blocks. I am also interested in techniques that can overcome such limitations, either with innovations in architecture, or using signal processing methods or leveraging the benefits of heterogenous integration, such as integration of different device technologies on the same substrate (GaN-Si, silicon photonics) etc.