Research

Over the past years, I worked on some interesting problems in the broad area of signal processing for communications and radar. I was fortunate to have worked with great mentors and amazing colleagues at IIT Madras, UT Austin, Univ. of Oulu., Tampere Univ., Qualcomm, Samsung, and most recently TU Delft. This page summarizes my research directions.

Note: The sections in grey contain an image carousel with multiple images. Please feel free to flip through them.

Current group members:

Postdoc- Dr. Chen Quan (joint with Dr. Geethu Joseph)

PhD - Ir. Edoardo Focante (BSc- UNIVPM, MSc- TU Delft)

Ir. Mohammed Aasim Shaikh (BSc- IIT GN, MSc- KAUST)

Ir. Iosef Vardakis (BSc- TU Crete, MSc- TU Crete)

Ir. Remco Huijsen [joint with Dr. Holger Caesar] (MSc- TU Delft)

MSc - Xian Cheng Hsiao, Luis Alcantara Flores, Kinge Keulen

Graduates:

PhD- Ir. Hamed Masoumi (Next job- Postdoc at EEMCS, TU Delft)

Research Engineer - Thymon Rhemrev, Federico Scari (joint with Dr. Arkady Zgonikkov)

MSc - Surya Natarajan Subramani, Weijia Yi, Aron Bevelander, Liam Rees

BSc, Appl. Physics - Tom Entes

BSc, ME - Floris van Leeuwen, Luke Netjes, Matthijs Klasens, Looka Schoneveld, Matthijs Jansen, Martijn Smak, Job Keijts, Demi Spee, Kinge Keulen, Koen van Dulst, Noortje van Vulpen, Tijn Jeletich, Roger Kalkman, Gideon van Triest, Emma de Jong, Jordy Pronk, Thymon Rhemrev, Chris Hogendoorn, Ruben Wosten, Marnix Zimmerman, Paul Barend Groen, Bein Frederik Jacob Kamminga, Lars Jisse Hoogland, Boele van Schaik, Anniek Christine van der Veen

TU Delft students who are interested in my research may send me an email for BEP/ MSc thesis projects. Externally funded researchers (MSc/ PhD/ Postdoc) can check out this link.

Robust compressed sensing under carrier frequency offset (CFO)

Compressed sensing (CS) is a promising solution to accelerate beam alignment at millimeter wave (mmWave). Most CS techniques at mmWave use random phase shift-based antenna weights to sense the sparse mmWave channel. The CS matrix with random weights, however, results in poor beam alignment under a CFO error. In our research, we developed a new concept of G-space trajectories to model a special class of CS matrices. We then constructed a set of partial randomized G-space trajectories that achieve robustness to CFO. [Video]

Waveform design for mmWave communication and radar

Commercial phased arrays have finite resolution phase shifters, a single radio frequency chain, and antenna weights that are approximately flat over the bandwidth. All of these create new challenges such as limited spatial sampling, beam squint, imperfect beam patterns and space-time aliasing artifacts. On the communications front, my research has addressed some of these challenges using ideas from perfect binary arrays, periodic complementary pairs and quaternions. On the radar front, I worked on a different set of problems related to space-time waveform design and beam squint compensation.

Data-driven signal processing for situation-aware link configuration

I have also looked at optimizing some components of a link configuration algorithm using a data-driven approach, in my research. One specific example is on site-specific beam alignment using tools from optimization and deep learning. Another example is on learning-aided channel estimation. In both these examples, our focus was on how to augment signal processing algorithms with data-aided priors, instead of solving the problem with a black-box deep learning approach.

Short range link configuration

Short range links arise in chip-to-chip, data center and wearable communications. As the transceiver distance in these systems is smaller than the Fraunhoffer limit, conventional link configuration algorithms perform poor when applied to short range settings. In our research, we proposed a new geometry-aided message passing algorithm that exploits the structure in short range channels for link configuration. Recently, we also worked on the misfocus problem in short range beamforming and proposed a spatial chirp-based phase profile for misfocus mitigation.

Message passing-based joint channel estimation and frequency synchronization in one-bit receivers

Receivers with one-bit analog-to-digital converters (ADCs) are potential candidates for next generation radios due to their low power consumption. The coarse quantization in one-bit ADCs, however, complicates signal processing tasks. Imagine estimating a vector from just signed measurements! Using a collection of several measurements and prior information on the signal structure, the vector can still be estimated upto a "reasonable" accuracy. In our research, we developed message passing-based algorithms for joint frequency synchronization and channel estimation in one-bit receivers.

Physical layer security in low resolution radios

The use of low resolution transceivers results in a reduced power consumption. Unfortunately, the low power consumption comes at the expense of compromising communication security. To address this issue, it is important to revisit the physical layer security problem from the scratch. In our recent research, we developed attack and defense strategies for secure mmWave communication with low resolution phased arrays. Our defense strategy outperforms other well known techniques based on antenna subset modulation and directional modulation. [Video][Paper]