2018 SE Doctoral Dissertation Showcase – Exemplary Recognition
Computational Models for Architecture of Autonomous Networked Systems
Complex systems, comprised of heterogeneous constituents that are capable of localized autonomous decision-making, are becoming increasingly ubiquitous in a wide range of technical and socio-technical contexts. Study of such systems requires methods that are beyond hierarchical design and control approaches. The traditional engineering approaches do not capture individual components’ autonomy, the way components interact and (dynamically) share resources, and these system’s responsiveness to highly uncertain and rapidly changing environments. Hence, a key question is that of how to promote desired behavior in such systems through mechanisms that leverage individual agents’ autonomy – as opposed to directly controlling them. In this work, we investigate the way connectivity structure of interactions between individual components can be altered as a mechanism to govern the behavior of socio-technical systems in uncertain environment. We develop a computational framework that helps in decision making about architecture of complex systems under uncertainty and covers a wide range of systems from fully integrated and monolithic to distributed systems with evolving network architectures, which are capable of dynamic resource sharing. For distributed systems with autonomous units, we develop strategic network formation models that capture components heterogeneity (e.g., processing capacity, state, and bandwidth) and the system environment uncertainty profile, typical characteristics of many technical systems. Under various conditions of heterogeneity, we derive analytical solutions for the efficient and stable connectivity structures and show that, for a range of costs and benefits of connections, these networks exhibit certain structural characteristics, such as the Core-Periphery. Additionally, to capture the effect of connectivity structure on the collective behavior of systems comprised of adaptive technology agents/human agents whose strategies evolve over the course of interactions, we use a multi-layered agent-based model based on evolutionary game theory on graphs. We focus on the case of collective fair behavior in social systems and show that two network parameters, namely, community structure, as measured by the modularity index, and network hubiness, represented by the skewness of degree distribution, have the most significant impact on the emergence of collective fair behavior. These two parameters can explain much of the variations in fairness norms across societies and can also be linked to hypotheses suggested by earlier empirical studies. Finally, we complement the theoretical and simulation-based work on system level behavior by developing a data-driven model that captures collective behavior from textual data using natural language processing models.
Mohsen Mosleh is a postdoctoral scholar at Sloan School of Management, Massachusetts Institute of Technology (MIT). Prior to joining MIT, he was a postdoctoral scholar at Psychology Department at Yale University. He received his PhD in Systems Engineering with a graduate certificate in Business Intelligence and Analytics from Stevens Institute of Technology. He holds a master’s degree in Management and a BS degree in electrical engineering both from Sharif University. He also worked as a Systems and Software Integration Lead for five years prior to starting his PhD. He has been the recipient of several academic awards including Fabrycky-Blanchard Award for excellence of academic performance in Systems Engineering Research in 2017, the best paper award of Systems Engineering journal in 2016, and INCOSE Foundation Doctoral Award for promising research in systems engineering and integration in 2015. His research interests include architecture of distributed and networked systems, economic and technological networks, modeling the emergence of social norms and collective behavior in socio-technical systems, and computational social sciences.
Mohsen Mosleh, Postdoctoral Associate, Sloan School of Management, Massachusetts Institute of Technology, Cambridge, Massachusetts 02142, USA
Phone: +1 617 324 6710