Research

Creating Efficient Parallel Systems


Futurists often speak of society's inevitable technological "singularity", a point in the near future where computers will cease to exist as the distinct objects they are today, and instead manifest themselves as millions of smaller and ubiquitous units embedded in everyday objects and being interconnected by distributed networks which allow their seamless integration. The trend toward such a future is already being foreshadowed by the recent multicore processing via the network-on-chip approach, a novel paradigm which implements on-chip networks that enable platforms of extreme parallel capabilities. Our group seeks to develop not only the hardware technologies which enable such as fundamental shift, but also the optimization and resource management techniques which can facilitate the coordinated transactions between the units of a massively parallel computational paradigm. This communication-based design perspective can generate new mathematical approaches for energy-efficient, cost-effective, sustainable, large-scale distributed computational platforms based on multicores for both embedded and high-performance applications.

Read more about embedded systems here

Developing Models for Microscopic Robots


Development of microrobots that can swim and operate inside the human body for minimally invasive medicine seems to become possible due to the recent advances in nano-technology and molecular biology. Indeed, bacteria-based microrobots (i.e. bio-robots) can be designed to communicate and perform complex tasks such as diagnostic and targeted drug delivery at micro- and nano-scales. Such microrobots can sense chemical cues from the neighboring cells, swim towards the targeted sites, and finally identify and release the therapeutic agents. To model the dynamics of such biological systems, we develop analytical models involving advanced computational and statistical physics approaches that are meant to capture the intracellular dynamics and intercellular communication in such bacteria-based networks. We also integrate all these models into an open-source, parallel, stochastic, and multiscale simulator that can be used to calibrate the relevant parameters, simulate various biochemical reactions at molecular level, and then capture the dynamics of the entire system at population level.

Read more about biological cyberphysical systems here

Modeling Social Networks


Social networks disseminate real-time information through users’ dynamic interactions and collective behaviors. Consequently, social networks manifest remarkable bursts of keywords and topics that correspond to real-world events that draw intensive attention from the general public. To harvest the full potential of this new medium of information, our group focuses on developing methods that discover complex interactions and collective behaviors that determine how various types of bursty events in social networks (e.g., intensive attention of the general public) are generated and propagated. The results of this research can be used to promote or to impede the propagation of crucial information within social networks. Further, our group focuses on designing early detection and forecasting engines for bursty events that take place over large social networks. The results of this research represent an important step toward real-time and predictive management of social networks by making possible a response in space and time that is valuable for various applications, such as micro-blogging, viral marketing, emergency response, and disaster management.

Read more about social cyberphysical systems here


Copyright 2020 SLD Group @ UT Austin