Conventional radar systems only detect and localize objects within the radar's field-of-view that reflect incident signals directly back to the radar. However, in practice, radar signals bounce multiple times to other intermediate objects in the environment. We exploit these multiple signal bounces to detect & localize hidden objects outside the radar's field-of-view, e.g., objects located around-corners and behind-radars. Our implementation on a commercial millimeter-wave radar demonstrates 2x-10x improvement in localizing beyond-field-of-view objects over baselines that do not exploit multiple signal bounces.

N. Mehrotra, D. Pandey, A. Prabhakara, Y. Liu, S. Kumar and A. Sabharwal, "Hydra: Exploiting Multi-Bounce Scattering for Beyond-Field-of-View mmWave Radar," ACM MobiCom, 2024 (acceptance rate = 19%)

Autonomous navigation requires sensors that can not only estimate how fast objects move but also their direction of motion. However, automotive radars are only capable of detecting object motion towards or away from the radar, and require object tracking across multiple radar frames to estimate their true direction of motion. We develop a method that enables commercial radars to estimate motion vectors of objects within a single radar frame by processing multiple signal bounces from known, static landmarks, e.g., building pillars or walls. Our implementation on an open-source automotive radar dataset from Mercedez-Benz demonstrates 4.5x improvement in motion vector estimation.

N. Mehrotra, D. Pandey, U. Madhow, Y. Mostofi and A. Sabharwal, "Single-Frame MIMO Radar Velocity Vector Estimation via Multi-Bounce Scattering," IEEE Transactions on Computational Imaging, Special Section on Computational Imaging using Synthetic Apertures, 2025

Wireless devices are everywhere, and unlike visible light sources, are "always-on". This opens up the opportunity to leverage the ubiquity of wireless devices to sense the surrounding environment at unprecedented scale without having to deploy dedicated radar systems. However, the majority of wireless traffic comprises unknown data signals that cannot directly be utilized for sensing. Hence, existing systems transmit known signals (“pilots”) alongside data in every frame; however, at the cost of reduced communication throughput and large sensing latency. In recent work, we make the observation that data signals, once decoded, can be reused for sensing; thus enabling uninterrupted sensing with no loss in communication throughput.

N. Mehrotra and A. Sabharwal, "On the Degrees of Freedom Region for Simultaneous Imaging & Uplink Communication," IEEE Journal on Selected Areas in Communications, Special Issue on Integrated Sensing and Communication, 2022

S. R. Mattu*, N. Mehrotra* and R. Calderbank, "Differential Communication in Channels with Mobility and Delay Spread using Zak-OTFS," IEEE Wireless Communications Letters, 2025 (*co-primary authors)

Network users expect seamless connectivity – even in environments with extreme mobility, such as on bullet trains. Mobility causes legacy signaling schemes to fade, i.e., randomly fluctuate in signal amplitudes, negatively impacting network throughput and latency. Existing approaches solve this issue by modulating information in the delay-Doppler (radar) domain on pulse train-like waveforms that sacrifice energy efficiency for fading resilience. In recent work, we propose a new basis of "spread" communication waveforms that retain the fading resilience properties of delay-Doppler pulse trains while enabling energy-efficient implementation on practical 5G hardware.

N. Mehrotra*, S. R. Mattu* and R. Calderbank, "Zak-OTFS with Spread Carrier Waveforms," IEEE Wireless Communications Letters, 2025 (*co-primary authors)

N. Mehrotra*, S. R. Mattu* and R. Calderbank, "A Design Framework that Unifies 6G Modulation Schemes for Double Selectivity," IEEE Wireless Communications Letters, 2026 (*co-primary authors)