Date of Award
Doctor of Philosophy
Dr. Ala Al-Fuqaha
Dr. Ajay Gupta
Dr. Fahad Saeed
Dr. J. Michael Tarn
Machine learning, reinforcement learning, Hadoop MapReduce, Software Defined Networking, OpenFlow Management, Big Data
For the ongoing advancement of the fields of Information Technology (IT) and Computer Science, machine learning-based approaches are utilized in different ways in order to solve the problems that belong to the Nondeterministic Polynomial time (NP)-hard complexity class or to approximate the problems if there is no known efficient way to find a solution. Problems that determine the proper set of reconfigurable parameters of parametric systems to obtain the near optimal performance are typically classified as NP-hard problems with no efficient mathematical models to obtain the best solutions. This body of work aims to advance the knowledge of machine learning techniques for the adaptive applications that depend on the set of configurable parameters, particularly the reinforcement learning approach, to address such problems. This work focuses on applying reinforcement learning to two chosen applications: (1) the MapReduce framework, and (2) the flow management in Software-Defined Networking (SDN). To demonstrate the effectiveness of this work, three studies were conducted: 1- The literature review of SDN technologies, architectures, applicable applications, underlying protocols, and deployments. 2- The use of reinforcement learning algorithm to search for the set of reconfigurable parameters that yields the near optimal performance automatically and adaptively, including the development of the simulator for evaluating the results and applying the obtained results to the real world application, and 3- The feasibility for the development of the emulation-based environment for the utilization of reinforcement learning in flow management in SDN, in order to address the limited capacity issue of the Ternary Content-Addressable Memory (TCAM) installed in an OpenFlow enabled switch. The MapReduce simulator was implemented using Network Simulator 3 (NS-3) simulation engine and the SDN emulation was built on top of Mininet network emulator. The results of these studies provide information on the effectiveness and efficiency of reinforcement learning algorithms for improving the performance of the parametric systems adaptively and automatically.
Mu, Ting-Yu, "Toward Self-Reconfigurable Parametric Systems: Reinforcement Learning Approach" (2019). Dissertations. 3546.