| dc.description.abstract | Wireless Mesh Networks (WMNs) have gained considerable popularity recently. Most research
on WMNs focuses on improving network capacity, because WMNs suffer from severe
throughput degradation due to interferences. Among them, routing has been an active area of research
for a long time, and it is also the focus of this dissertation.
Routing metrics are critical for selection of a path with a maximum throughput in multi-hop
wireless networks. Due to the nature of wireless losses and multi-rate support, wireless links in one
single network may have a broad range of characteristics such as loss rates and data transmission
rates. In addition, wireless nodes may be equipped with multiple radios tuned to non-overlapping
channels in order to improve network performance. All these technologies bring new challenges
to route selection in wireless multi-hop networks, and the traditional minimum hop count metric
does not suffice in such a complicated environment, since it does not take any aforementioned
characteristics into account.
To design an effective routing metric, it is necessary to accurately acquire the link properties
that might affect path performance, most importantly, the loss rate. To better assess individual link
loss rate, this study proposes several modifications to existing broadcast based probe method that leads to a more accurate link loss rate measurement. In recent years, although many link qualitybased
routing metrics have been explored, none of the current routing metrics is capable of capturing
the necessary link properties, which is critical for a comprehensive evaluation of routing path
performances and the selection of the best route. In this study, we explore the drawbacks of current
routing metrics and propose two improved strategies, improved Expected Time Transmission
(iETT) and improved Bottleneck Aware Transmission Delay (iBATD), respectively, to address the
challenges for single-radio and multi-radio wireless mesh networks. The proposed routing metrics
will be explored in various wireless network scenarios through ns-2 simulations and their performances
will be compared to current routing metrics, such as minimum hop count metric (HOP),
Expected Transmission Count (ETX), Expected Transmission Time (ETT) and Weighted Cumulative
ETT (WCETT) metrics. Additionally, an extended Dynamic Source Routing (DSR)-based
routing protocol (eDSR), which works effectively in multi-radio scenarios and performs seamlessly
with any quality aware routing metrics, has also been designed to aid the performance comparisons.
Moreover, we develop iETT routing metric and three other selected representative routing metrics
on a Linux based test bed and assess their performances with extensive experiments. | en |
