THESIS
2010
xi, 68 p. : ill. (some col.) ; 30 cm
Abstract
In the past decade, IEEE 802.11 protocol, which is originally designed for wireless local area networks (WLAN), has been invoked repeatedly in the context of wireless mesh networks (WMNs) to provide last-mile broadband service and large-scale metropolitan area networks thanks its high data rate, wide coverage and reduced infrastructure expenditure. Nevertheless, such deployment makes the well-known WLAN problems, such as hidden or exposed terminals, channel impairments such as multi-path fading, and so on, more severe, resulting in several types of service provisioning problem in WMNs. Among such problems that are yet to be addressed, we list, the lack of fairness among MAC flows, and flow starvation, which is the ultimate manifestation of the lack of fairness. In this thesis, we aim to...[
Read more ]
In the past decade, IEEE 802.11 protocol, which is originally designed for wireless local area networks (WLAN), has been invoked repeatedly in the context of wireless mesh networks (WMNs) to provide last-mile broadband service and large-scale metropolitan area networks thanks its high data rate, wide coverage and reduced infrastructure expenditure. Nevertheless, such deployment makes the well-known WLAN problems, such as hidden or exposed terminals, channel impairments such as multi-path fading, and so on, more severe, resulting in several types of service provisioning problem in WMNs. Among such problems that are yet to be addressed, we list, the lack of fairness among MAC flows, and flow starvation, which is the ultimate manifestation of the lack of fairness. In this thesis, we aim to design an easily deployable framework, to address these problems.
To begin with, we first study the trade-off between aggregate MAC flow throughput and the fairness in a WMN that utilizes scheduling on top of the CSMA/CA access scheme. For this purpose, we use a toy WMN topology to allow us to understand easily this tradeoff and propose an analytical model to study the interaction between contending links in this topology. Based on this model, we formulate the bandwidth-scheduling problem on top of CSMA/CA in this topology as an aggregate MAC flow throughput maximization problem subject to the fairness requirement dictated by the scheduler. The rational of such study is to provide a theoretical benchmark to understand the performance limits of our proposed non work-conserving scheduler designed to run on top of the CSMA/CA protocol in a distributed environment.
After understanding how much sacrifice in the aggregate MAC flow throughput or the fairness are made - due to the mismatch between the objective of the fair scheduler on one hand and the operation of the CSMA/CA algorithm on the other, we discuss the implementation of a distributed fair MAC flow allocation and scheduling framework for provisioning MAC flow fairness. The framework consists of (1) a modified wireless ad hoc routing protocol to take into account bandwidth requirement, (2) a link information dissemination protocol to propagate local topology information, (3) a cooperative gradient-based iterative algorithm to allocate fairly MAC flow bandwidth and (4) a distributed fair MAC scheduler to coordinate access to the channel according to the allocated bandwidth.
Post a Comment