Electrical and Computer Engineering

College of Engineering and Applied Science

Dr. Suresh Muknahallipatna's Research

> Visit Dr. Muknahallipatna's webpage

Computer Networks and High Performance Computing (HPC)

 

High Performance Computing

GPGPU based HPC is a new approach to achieve Super Computer computing capabilities at minimal cost. This involves developing parallel software satisfying the graphical processing unit (GPU) architecture to achieve high computation speedups. Recently, Dr. Muknahallipatna and a team of collaborators have developed parallel algorithms in the area of computer vision, radio propagation maps and snow modeling to execute on a GPGPU cluster. These parallel algorithms are useful in applications like real-time remote rehabilitation, real-time terrain based radio propagation map generation and flood forecasting

Online Princeton GPU Hackathon - June 2020

The ECE HPC research team at the University of Wyoming is one of the groups invited to mentor faculty and research scientist teams participating in the Princeton GPU Hackathon hosted by Princeton University, Oak Ridge National Laboratory, and NVIDIA. The Hackathon is conducted over two weeks starting on the June 1^st and ending on June 9^th, 2020. Three graduate students Sumathi Lakshmiranganatha, George D. Dickerson, and Pranay R. Kommera under the supervision of Dr. Muknahallipatna will mentor two research teams:

Fusion Recurrent Neural Network – Sumathi Lakshmirangantha and George D. Dickerson

Fusion energy is a safe, clean, potentially unlimited energy source, however, it has not been commercially realized due to technical and scientific challenges. Fusion fuel needs to be confined at extremely high temperatures (in a plasma state) and reactors are subject to events called disruption - an extremely fast loss of confinement leading to severe forces and heat loads on the reactor. Our team has recently developed deep learning algorithms that can predict disruptions early enough to safely shut down the reactor. We are now looking to develop the code/hardware to deploy the predictor in real-time on experimental reactors throughout the world. Programming Languages and Models: Python, Matlab, CUDA-C, cuBLAS, cuDNN, Keras, tensorrt, TensorFlow

STRIDE – Pranay R. Kommera

Rapid calculation of linear ideal MHD stability. Given the state of a tokamak plasma, it calculates whether the equilibrium is stable against perturbations. Currently it is used for both offline analysis of fusion experiments, and an online version is also in development for calculating stability in real time to avoid harmful disruptions in fusion devices like DIII-D and (in the future) ITER. Programming Languages and Models: FORTRAN, OpenMP/OpenACC, cuBLAS/cuFFT

The ECE HPC team consists of graduate and undergraduate students pursuing Ph.D., Masters and UG computer engineering degrees in the areas of HPC, Machine Learning and Augmented Reality. Dr. Muknahallipatna leads the HPC research team.

 

Mobile Adhoc Network

The Mobile Adhoc network (MANET) provides the capability to rapidly deploy a communication network for applications like battlefield communications, collection of environmental data for natural system analysis to mention a few.  A MANET consists of mobile nodes forming a network to relay information across the network. However, developing a communication network infrastructure on an ad-hoc basis, poses significant technology challenges including design and fielding of antennas, radios, terminals, and higher-layer communications protocols. Some of the fundamental challenges of an ad-hoc network include harsh radio frequency (RF) propagation environments, spectrum limitations, varying degrees of terminal (node) mobility, scalability, energy consumption, network lifetime, self-healing, localization, and routing.
Dr. Muknahallipatna and a team of collaborators are addressing this technological challenges and the outcome of the research can be found at http://www.uwyo.edu/electrical/research/signal-processing.html