TCS’ Embedded Devices and Intelligent Systems research focuses on intelligent sensing systems that are the building blocks of perceptive machines. This research area is centered on computational sensing, signal processing, embedded computing, and their new research paradigms. The outcomes include the development of novel, frugal, and deployable solutions for intelligent IoT systems in industrial manufacturing, oil and gas, aerospace, medical devices, remote healthcare, smart energy, supply chain, smart transportation, and earth observation.
While catering to the constrained nature of embedded systems, the TCS team addressed the following challenges:
Investigation of a wide range of spectrum from infrasound to terahertz for information sources - 1D, 2D, and 3D
Uncertainty of information collection under dynamic and noisy environments
Increasing intelligence efficiency by improving deployment of sensing systems and optimization of sensing cost
Development of advanced embedded intelligence platforms and algorithms
The primary focus areas of this research are:
Embedded sensing systems (ESS): With designing and developing novel embedded systems being the focus, TCS uses the latest technologies and techniques such as sensing modalities (acoustic, RF, microwave, and optical), radar systems, phased array systems, embedded programming & sensor data acquisition systems, signal conditioning and pre-processing, computational sensing systems, nano-sensing, and bio-sensing. Through novel sensing, they explore better ways to understand the environment and ultimately impart intelligence to devices by efficiently processing the sensed information.
Signal analysis and machine vision (SAMV): This group focuses on the creation of novel algorithmic suites for signal, image, and video processing and 3D vision. Core research topics include multi-modal fusion, array signal processing, and beamforming, model compression, spiking neural networks, adaptive systems, active sensing, and affective computing. In addition, the focus is on the fusion of classical signal processing methods with emerging deep neural networks leading to the development of hybrid perceptive machines. The fusion of science-based knowledge models with data-driven ML/DL models and Quantum ML are some of the other new topics being explored.
Embedded computing and communication platforms (ECCP): This group focuses on distributed embedded systems, real-time systems, time-sensitive networks, model partitioning and optimization, sensor reliability and planning, and edgified machine learning on specialized processors and devices like neuromorphic, RISC-V, FPGA. TCS also looks at distributed computing across the edge, fog and cloud with a focus on optimization of communication cost, execution latency, energy footprint, and so on.
People, Publications and Patents
Lead: Dr. Arpan Pal, Chief Scientist
Team: Dr. Tapas Chakravarty, Dr. M Girish Chandra, Ramesh Ramakrishnan, Dr. Arijit Mukherjee, Chirabrata Bhaumik, Soma Bandyopadhyay, Dr. Hrishikesh Sharma, Dr. Jayavardhana Gubbi, Arijit Ukil, Naveen Thokala, Dr. A. Anil Kumar, Dr. Himadri Sekhar Paul, Debatri Chatterjee, Sounak Dey, Swarnava Dey, Kriti Kumar, Dr. Anwesha Khasnobish, Arijit Chowdhury, Arijit Sinharay, Gitesh Kulkarni, and Dr. Kartik Muralidharan
Academic partners: MIT Media Lab, USA; IIT Kharagpur, India; ISI Kolkata, India; IISc Bangalore, India; IIIT Delhi, India; and Jadavpur University, India
Patents and Publications (last three years): Over 70 patent grants and more than 110 publications in top-tier conferences and journals