Text

theses

The dramatic growth in mobile Internet services and their diverse requirements motivate the need to significantly improve the design of protocols associated with spectrum use and data transfer over wireless networks. In this thesis, we will examine several algorithms and protocols for improving the spectrum utilization of wireless networks. Specifically, we propose new protocol frameworks and associated algorithms for: (1) multipath routing of data to devices equipped with multiple wireless interfaces such as Wi-Fi and cellular; (2) efficient information-centric multicast over wireless networks; and (3) dynamic spectrum access by wireless networks in both licensed and unlicensed bands. Control and data plane protocol design is examined along with applicable distributed algorithms for each of the above cases, and corresponding performance evaluation results compared with existing solutions are provided. The unifying theme in terms of data transfer protocols is the use of globally unique identifiers derived from a new information centric Internet architecture called MobilityFirst, along with the use of fully distributed algorithms to achieve the desired functionality.
The thesis starts with an investigation of mobile access challenges including device mobility and multi-interface connectivity and identifies the shortcomings of existing solutions such as transport layer solutions (MPTCP) and dual-connectivity/mobility mechanisms adopted in cellular carriers (5G). To address these challenges, the proposed design uses the concept of named-objects, based on separation of naming and addressing and dynamic binding of connections to identifiers, in the context of a future internet architecture "MobilityFirst". Building on the MobilityFirst architecture, we propose protocols for in-network data splitting based on explicit edge feedback. Detailed performance evaluation of our proposed scheme compared with existing ones is presented using simulations and real-world experiments.
The second chapter presents techniques for extending the usage of named-objects within the cellular network architecture to enable dynamic and efficient wireless multicast while greatly reducing overhead and latency. This novel cross-layer design is important for requirements of emerging services, like large-venue live video streaming applications, autonomous driving assistance and massive IoT. The main idea is to introduce awareness of information centric network (ICN) identifiers at the wireless MAC layer in order to enable efficient multicast services on the wireless channel. To provide a holistic multicast solution, two important aspects are considered: radio resource scheduling and protocol design.
To enable radio resource scheduling, an algorithm for proportional fair resource allocation to a mix of multicast and unicast flows which is deployable in time-slotted systems is proposed. Numerical analysis of the impact of various parameters like total number of UEs/multicast services and grouping of multicast UEs are presented. Finally, comprehensive comparison of ICN multicast with the current wireless multicast standard (eMBMS) is carried out in order to quantify achievable performance gains for the use case of context-based services. The superior performance of ICN multicast in terms of packet delivery time and throughput is illustrated for this use case.
The third chapter considers dynamic spectrum management motivated by the need to share spectrum across the frequency band, including licensed and unlicensed spectrum, more efficiently. We present the case for decentralizing the architecture for dynamic spectrum management as a means to address artificial spectrum scarcity and overcoming spectrum under-utilization. To this end, we propose SMAP (distributed spectrum management architecture and protocol), which enables coordination of spectrum use among wireless networks and devices through an Internet-based common spectrum control plane. Drawing insight from the success of distributed protocols like BGP in Internet, SMAP facilitates peer exchange of radio usage and control parameters among autonomous wireless network domains. It also provides interfaces to higher level cloud services including spectrum aggregators which facilitate broader cooperation and business relationships between wireless domains in the same area, or to regional spectrum databases such as the SAS (Spectrum Access System). The chapter starts with a technology outline of the architecture, specifying protocols and network entities. Then it establishes three classes of distributed algorithms for optimization of transmission parameters and showcases the feasibility and performance, for a variety of scenarios including coexistence of heterogeneous wireless technologies in the unlicensed band and priority-based spectrum access by bringing all the designed protocols and elements into a single experimental platform on an open-access research testbed. These experimental results are presented to demonstrate significant improvements in system throughput and fairness. We then analyze sample distributed algorithms using large-scale simulations, considering efficiency, scalability, convergence and their adherence to global and local policies.

ETD_10335

Ph.D.

Includes bibliographical references

doi:10.7282/t3-6wr3-bq91

ETD doctoral

The author owns the copyright to this work.