Date of Award
12-2015
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Legacy Department
Computer Science
Committee Chair/Advisor
Martin, James
Committee Member
Dean, Brian
Committee Member
Luo, Feng
Committee Member
Westall, James
Abstract
In this dissertation, we consider the downstream bandwidth management in the context of emerging DOCSIS-based cable networks. The latest DOCSIS 3.1 standard for cable access networks represents a significant change to cable networks. For downstream, the current 6 MHz channel size is replaced by a much larger 192 MHz channel which potentially can provide data rates up to 10 Gbps. Further, the current standard requires equipment to support a relatively new form of active queue management (AQM) referred to as delay-based AQM. Given that more than 50 million households (and climbing) use cable for Internet access, a clear understanding of the impacts of bandwidth management strategies used in these emerging networks is crucial. Further, given the scope of the change provided by emerging cable systems, now is the time to develop and introduce innovative new methods for managing bandwidth. With this motivation, we address research questions pertaining to next generation of cable access networks. The cable industry has had to deal with the problem of a small number of subscribers who utilize the majority of network resources. This problem will grow as access rates increase to gigabits per second. Fundamentally this is a problem on how to manage data flows in a fair manner and provide protection. A well known performance issue in the Internet, referred to as bufferbloat, has received significant attention recently. High throughput network flows need sufficiently large buffer to keep the pipe full and absorb occasional burstiness. Standard practice however has led to equipment offering very large unmanaged buffers that can result in sustained queue levels increasing packet latency. One reason why these problems continue to plague cable access networks is the desire for low complexity and easily explainable (to access network subscribers and to the Federal Communications Commission) bandwidth management. This research begins by evaluating modern delay-based AQM algorithms in downstream DOCSIS 3.0 environments with a focus on fairness and application performance capabilities of single queue AQMs. We are especially interested in delay-based AQM schemes that have been proposed to combat the bufferbloat problem. Our evaluation involves a variety of scenarios that include tiered services and application workloads. Based on our results, we show that in scenarios involving realistic workloads, modern delay-based AQMs can effectively mitigate bufferbloat. However they do not address the other problem related to managing the fairness. To address the combined problem of fairness and bufferbloat, we propose a novel approach to bandwidth management that provides a compromise among the conflicting requirements. We introduce a flow quantization method referred to as adaptive bandwidth binning where flows that are observed to consume similar levels of bandwidth are grouped together with the system managed through a hierarchical scheduler designed to approximate weighted fairness while addressing bufferbloat. Based on a simulation study that considers many system experimental parameters including workloads and network configurations, we provide evidence of the efficacy of the idea. Our results suggest that the scheme is able to provide long term fairness and low delay with a performance close to that of a reference approach based on fair queueing. A further contribution is our idea for replacing `tiered' levels of service based on service rates with tiering based on weights. The application of our bandwidth binning scheme offers a timely and innovative alternative to broadband service that leverages the potential offered by emerging DOCSIS-based cable systems.
Recommended Citation
Hong, Gongbing, "Downstream Bandwidth Management for Emerging DOCSIS-based Networks" (2015). All Dissertations. 1579.
https://open.clemson.edu/all_dissertations/1579