| dc.description.abstract | Networks are constantly flooded by heterogeneous services, and while some services such as Virtual Reality (VR) and Voice over IP (VoIP) have strict requirements regarding real-time packet delivery, others such as downloading large files have non-real time overflow traffic.
In the past years, research community had sufficient study to provide optimal scheduling policies under different requirements on the network communication. However, in the studies, a distinct lack of network fluidity exists: there has only been one network base station supporting multiple services, and the optimal scheduling policy for one service doesn't necessarily serve the other services ideally.
In this thesis, I present a new software-defined MAC system that employs network slicing to support multiple heterogeneous scheduling policies in a uniform framework. After successfully constructing the testbed, I further integrated Software Defined Networking (SDN) into the testbed as a tool to manage slicing and scheduling configuration. With the help of SDN, I designed and implemented an algorithm to effectively approach and calculate the optimal resource allocation between slices. I'm able to demonstrate that the testbed outperforms the system with baseline single scheduling policy. In addition, the proposed algorithm can further increase the system performance. | en |
