Hybrid seminar at GERAD or Zoom.

Modern networked infrastructures, such as power grids, communication networks, and cyber-physical systems, rely on decentralized and centralized agents that must allocate limited resources while facing uncertainty, limited communication, and sometimes competition. Designing decision-making mechanisms that remain efficient and reliable in such environments requires combining tools from distributed optimization and game theory.

In the first part of the talk, I consider cooperative resource allocation over time-varying networks motivated by adaptive under-frequency load shedding in power systems. The goal is to select an optimal subset of loads to disconnect in order to stabilize the grid after sudden disturbances, while accounting for socio-technical priorities among network components. Because the decision variables are discrete, classical gradient-based distributed optimization methods are not applicable. I present a distributed root-finding algorithm based on a cumulative criticality function that enables agents to collectively determine near-optimal actions using only local information and intermittent communication.

In the second part of the talk, I consider resource allocation in adversarial environments using a networked General Lotto game model. A defender and attacker compete by distributing resources across nodes, and network structure determines whether connections remain secure. I characterize equilibrium behavior for bipartite networks and derive performance guarantees for general networks. The results reveal that connectivity can fundamentally change optimal strategies and that randomization is an intrinsic feature of optimal defense policies

Biography: Adel Aghajan is a researcher in electrical and computer engineering specializing in distributed optimization, stochastic systems, game theory, and control of complex networks. He completed his Ph.D. in Electrical Engineering at the University of California, San Diego, where his dissertation focused on distributed computation and optimization over random networks. Following his doctoral studies, he held postdoctoral research positions at the University of California, Santa Barbara and later at the University of California, San Diego.