Our research has been driven by thought experiments on conceptual and practical bottlenecks for the realization of NextGen mmWave communication and sensing. This approach has opened up intellectual challenges in a variety of areas.  We provide below a sampling of completed research, but watch this space for updates on our ongoing research.  


All-digital mmWave massive MIMO

Analytical design frameworks, architectures and algorithms for all-digital mmWave massive MIMO, addressing bottlenecks such as nonlinearities due to low-precision analog-to-digital conversion and RF imperfections, and phase noise.


Information-theoretic guidelines on degrees of freedom as a function of form factor, collaboration on hardware prototyping, fundamental limits and algorithms under severe constraints on analog-to-digital conversion.

mmWave mesh networks

Interference analysis guiding design of medium access control (MAC) for mesh networks with highly directional links where deafness rather than interference is the bottleneck (influencing Facebook’s Terragraph initiative on mm wave mesh networking). Optimization framework for joint routing and resource allocation.

mmWave to the mobile

Identifying and addressing fundamental bottlenecks such as blockage and beam tracking in the design of picocellular systems with 1000X capacity increase, including interference analysis and management, and novel compressive architectures for spatial channel estimation and tracking.  Along the way, we developed state of the art super-resolution algorithms and fundamental insights into compressive parameter estimation.

Indoor mmWave networking

 Early analysis of blockage in mmWave networks using physics-based cross-layer models.

mmWave imaging and sensing

Revisiting the fundamentals of radar and imaging modeling and sensing for novel geometries, and introducing advanced signal processing architectures and algorithms. Characterized the degrees of freedom for short-range imaging as a function of scene and array geometry, developed new patch-based target models that alleviate grating lobes obtained with the classical point target model for sparse arrays, demonstrated super-resolution in range-Doppler processing and used it, together with geometric information, for multi-sensor spatial association. We have also developed optimization algorithms for design of large-aperture sparse array of subarrays. Most recently, we introduced a novel compressive architecture for scaling MIMO radar