Deep learning for optimal dynamic control of the internet of things
Abstract
In recent years, industrial Internet of Things (IIoT) has gained considerable
attention in both industry and academia. In manufacturing sector, factories use the
real-time big data collected from their IoT-enabled machines to derive new insights, which
can be used to optimize their supply chain dynamics. To this end, researches focused on
fully utilizing the opportunities provided by the IIoT, to meet the stringent requirement of
industry in terms of resource efficiency, and delay. However, to efficiently manage the
IoT-enabled resources while guaranteeing a return of investment for the IoT service
providers, is still challenging. In this proposal, I propose a control mechanisms using
auctions to optimally manage the IoT-enabled physical resources, that not only meets the
resource demands from users, it guarantees that the service providers can earn a profit.
The proposed auction-based mechanisms are truthful, individually rational, prevents
buyers from unfairly affecting the results, and its assignments satisfy physical resource
capacity constraints. I conducted extensive simulations, and see that the proposed auction
mechanism can find control decisions that both improve the utilization of IoT-enabled
resources, and profit of service providers.
Description
Thesis (Ph.D.)-- Wichita State University, College of Engineering, Dept. of Electrical Engineering & Computer Science