Abstract: The problem of resource sharing is ubiquitous: transceivers need to share the scarce wireless spectrum; traffic flows need to share routers’ available bandwidth; applications and tasks need to share the central processing units (CPUs) and memory of computer systems; consumer appliances need to share the available energy supply etc. A key challenge in developing a methodology for optimally sharing resources in the long run is that the sequential decisions of the agents (users, transceivers, applications) are coupled in a complicated manner across time. Existing state-of-the-art methodologies and solutions to this important and ubiquitous problem are inefficient, require extensive and perfect feedback and monitoring, and cannot cope with dynamic entry and exit or self-interested users. In this talk, I will propose a novel, systematic and practical design framework for distributed resource sharing that is optimal (efficient), decentralized, requires minimal and imperfect feedback and monitoring, allows for dynamic entry and exit and for self-interested agents. (That is, agents – who may be heterogeneous, who are autonomous and who maximize their own payoffs subject to constraints such as minimum throughput requirements, maximum energy consumption constraints, delay constraints, etc.). We model the problem as a stochastic game among agents with conflicting objectives and coupled constraints. A crucial and necessary feature of our solutions –optimal policies – is that they are nonstationary: each agent’s action depends not only on its current state but also on the history of previous states and actions. Our solution provides a novel and rigorous methodology to determine optimal spectrum sharing policies that flexibly optimize the desired system performance in terms of throughput, energy consumption, delay or utility-based fairness. The application of our proposed methodology to wireless spectrum sharing problems enables tripling the spectrum efficiency or achieving up to 90% energy saving, compared to state-of-the-art spectrum sharing policies. The proposed optimal resource sharing methodology can be used for a variety of other systems such as video over Internet-of-Things enabled networks, smart grids, cloud computing, crowdsourcing etc.
Bio: Mihaela van der Schaar is Chancellor's Professor of Electrical Engineering at University of California, Los Angeles. She is an IEEE Fellow, was a Distinguished Lecturer of the Communications Society (2011-2012), the Editor in Chief of IEEE Transactions on Multimedia (2011-2013) and a member of the Editorial Board of the IEEE Journal on Selected Topics in Signal Processing (2011). She received an NSF CAREER Award (2004), the Best Paper Award from IEEE Transactions on Circuits and Systems for Video Technology (2005), the Okawa Foundation Award (2006), the IBM Faculty Award (2005, 2007, 2008), the Most Cited Paper Award from EURASIP: Image Communications Journal (2006), the Gamenets Conference Best Paper Award (2011) and the 2011 IEEE Circuits and Systems Society Darlington Award Best Paper Award. She received three ISO awards for her contributions to the MPEG video compression and streaming international standardization activities, and holds 33 granted US patents. She is also the founding director of the UCLA Center for Engineering Economics, Learning, and Networks (see netecon.ee.ucla.edu). Her research interests include engineering economics and game theory, multi-agent learning, online learning, decision theory, network science, multi-user networking, Big data and real-time stream mining, and multimedia.