This item is under embargo and not available online per the author's request. For access information, please visit http://libanswers.wustl.edu/faq/5640.
Resource Allocation and Management of Computer Networks
Date of Award
Doctor of Philosophy (PhD)
Resource Allocation and Management (RAM) in Computer Networks has been an active research topic for decades. The complexity of the problem has increased for many reasons, such as the available mix of many different networks and scarce network resources along with an exponential increase in user demands and types of applications. An additional problem in wireless networks is that the frequency spectrum is fixed while the user demands are growing exponentially. Therefore, the problem is how to use the limited resources over the diverse networks efficiently to meet application specific requirements, such as throughput, delay, and delay jitter. In this dissertation, our focus is on three main research areas over three different networks - Part I: Congestion control in Datacenter Networks (DCNs), Part II: Resource allocation and management in IEEE 802.16 WiMAX Networks, and Part III: Resource management in Next Generation Wireless Networks (NGWNs). These three parts reflect the growing complexity of the problem space and the sequence of latest developments in the networking industry.
In the past, Infiniband and Fibre Channel were common in DCNs because they offer better resource management due to their sophisticated congestion and flow controls. There is a need to move to use Ethernet in data centers due to its low cost, but it lacks the congestion control mechanisms required for reliable transfer of large volumes of data. Part I deals with congestion control in DCNs.
The second important recent development is that of the broadband wireless metropolitan area networks, i.e., IEEE 802.16 WiMAX. Since these networks use licensed spectrum, which is very costly and scarce, resource management in these networks is very important. One of the key approaches for resource management is to design an ef- ficient scheduler that takes advantages of the characteristics of the targeted wireless technology. Being a paid service, one of the key features of WiMAX is its strong quality of service (QoS). The standard supports multiple QoS classes for voice, video, and data applications. Therefore, an efficient scheduler should be able to provide such kinds of services with quality guarantees. The IEEE 802.16 standard does not specify a standard scheduling mechanism and leaves it for vendor differentiation. This is Part II of this dissertation.
In Part III, we investigate key features/factors/issues required for NGWNs. We also propose a framework to overcome the issues, i.e., mobility, multihoming, and location privacy, and to aid resource management in a multi-interface mobile scenario, commonly used in NGWNs.
Min, Shyam Parkeh, Robert Pless