Vishnu N Moothedath

Research and Projects

Research Experience

Stochastic characterisation of delay violation probability of 5G URLLC

Research project in collaboration with Siemens AG, where the aim is to characterise the DVP (Delay Violation Probability) in an industrial 5G-URLLC setting. The DVP is characterised as a function of the resource utilisation and the retransmission mechanism with a specific focus on HARQ.

Hierarchical inference learning at the edge

Tiny machine learning (TinyML) models deployed on resource-constrained edge devices require efficient offloading strategies that delegate only complex tasks. This project develops algorithms that learn in real time which tasks should be offloaded, based on an adaptively learned complexity threshold. The resulting Hierarchical Inference (HI) framework provides provably sub-linear regret bounds and forms the basis of my subsequent work on cost-sensitive event detection with asymmetric misclassification penalties, extending HI to real-time decision-making for edge intelligence.

Optimal sampling of human-in-the-loop edge computing systems

The latency sensitive human-in-the-loop systems are developed, in many cases, with extremely high sampling rate which is unnecessary for relatively slow humans. A too low sampling rate on the other hand impact the responsiveness experienced by the humans. The project looks at WCA (Wearable Cognitive Assistant) applications and find an optimum periodic as well as aperiodic sampling scheme that minimises energy consumption.

Optimal downlink MISO beamforming and pareto boundary calculation

Master's thesis research where analytical solutions and iterative algorithms are developed to find the capacity boundary and maximum sum-rate providing beamforming vector in a MISO-IC with multiple frequency bands.

Compressive Sampling of Analog signals

Bachelor's project where the sub-nyquist sampling is studied.

Adaptive Delta Modulation - Modulation, Transmission and Reception : A Hands-On

Bachelor's project where a working model is made for data transmission over a microwave link using ADM (Adaptive Delta Modulation)

Seminar on Controller Area Network with case study

A case study and Seminar on CAN (Controller Area Network) and passenger car ABS

Thesis/Seminars

Towards Efficient Distributed Intelligence: Cost-Aware Sensing and Offloading for Inference at the Edge

Doctoral thesis

Optimal downlink MISO beamforming and pareto boundary calculation

Master's thesis

Compressive Sampling of Analog signals

Bachelor's project

Adaptive Delta Modulation - Modulation, Transmission and Reception : A Hands-On

Bachelor's project

Seminar on Controller Area Network with case study

A case study and Seminar

Industrial Collaboration

Siemens

On the characterization of delay violation probability of 5G HARQ

Ericsson

Various master thesis supervisions at Ericsson

Collaboration via TECoSA research center and ExPECA testbed

Scania

CaMeL project on multi-node hierarchical inference offloading problem

Implementation/Prototyping

ExPECA testbed

The ExPECA testbed at KTH Royal Institute of Technology is a reconfigurable edge-and-cloud research platform combining software-defined radios, COTS wireless links and containerised compute nodes to enable repeatable end-to-end experiments in communication and computing.

Google Sites

Report abuse