Reza Abdolee

Reza Abdolee

University of California, Los Angeles

California

PhD student Reza Abdolee worked closely with Prof. Ali H. Sayed at the Adaptive System Laboratory (ASL) of the University of California, Los Angeles (UCLA). During this period, he was engaged in research on distributed strategies for online adaptation, learning, and optimization over networks, which is an emerging field of study originally introduced by Prof. Sayed and his research team.

The focus of Mr. Abdolee’s internship was on developing diffusion adaptive strategies in sensor networks meant for monitoring a general class of physical phenomena that are well described by partial differential equations (PDEs). Research to date on diffusion adaptive strategies has shown that these strategies can serve as efficient and powerful learning mechanisms for solving estimation problems in a distributed manner and in real-time over networks. Diffusion adaptive algorithms have been used successfully to model several instances of organized behavior encountered in nature such as bird flight formations, fish schooling, bee swarming, and bacteria mobility. These algorithms, thanks to their adaptive feature, are potentially instrumental in sensor networks deployed for monitoring or control of dynamic systems over the space whose underlying parameter of interest changes over time. To apply diffusion adaptive strategies to such dynamic networks, Mr. Abdolee’s research at UCLA aimed to create a link between the inverse problems constrained by PDEs and the analytically tractable models in adaptive networks. Creating this link and resolving the associated issues will eventually enable the implementation of distributed adaptive strategies in sensor networks for parameter identification problems encountered in monitoring a wide variety of physical phenomena in nature.

Joint manuscripts describing this research have been written by Mr. Abdolee with Professors Benoit Champagne and Ali H. Sayed and have been published.