THESIS
2008
x, 57 leaves : ill. ; 30 cm
Abstract
Wireless mesh networks (WMNs) provide network coverage to areas where infrastructure is difficult to install. A general WMN consists of multiple access points (APs) providing gateway service to the Internet for associated mesh routers. Many peer-to-peer applications such as file sharing and streaming require an effective object location scheme. In this thesis, we study how objects can be located efficiently in WMNs, where peers (mesh routers) share and query for objects on behalf of users. We propose MOLE (Multi-hop Object Location for wireless mesh networks), a scalable, network-adaptive, efficient and distributed object location protocol for WMNs. MOLE employs a novel request filtering technique based on bloom filter aggregation to reduce search overhead significantly. It also uses a...[
Read more ]
Wireless mesh networks (WMNs) provide network coverage to areas where infrastructure is difficult to install. A general WMN consists of multiple access points (APs) providing gateway service to the Internet for associated mesh routers. Many peer-to-peer applications such as file sharing and streaming require an effective object location scheme. In this thesis, we study how objects can be located efficiently in WMNs, where peers (mesh routers) share and query for objects on behalf of users. We propose MOLE (Multi-hop Object Location for wireless mesh networks), a scalable, network-adaptive, efficient and distributed object location protocol for WMNs. MOLE employs a novel request filtering technique based on bloom filter aggregation to reduce search overhead significantly. It also uses a cross-border search technique to shorten search paths between peers associated with different APs. Using ns-2 simulations, we show that MOLE, as compared to a recent bloom-filter-based scheme (IDLP), saves much overhead (up to 70%), and is light-weight in terms of processing and memory requirements. The tradeoff is that MOLE's search paths to objects are slightly longer than shortest paths. In addition, MOLE achieves high success rate for search even in the presence of node churn. Our cross-border searches shorten search paths significantly (by up to 20%), while incurring low overhead.
Post a Comment