Mobile awareness in a wide area wireless network of info-stations

JVireless networking is becoming an increasingly important communication means, yet high wid~area wireless data connectivity is difficult to achieve due to technological and physical imitations. To alleviate these problems we experiment with an alternative byplacing many high bandwidth local “islands”of injo-stations dispersed throughout the low bandwidth wid~area wireless network. The Iocation and distribution of the individual stations is crucial for the network’s overall effectiveness. as demonstrated by our investigations. The info+ tations are deployed in a transparent manner, often notgeographlcally visible to the user. Applications must be designed to be mobileaware and able to account for changing network characteristics by optimally utilizing the available network resources. \Ve simulate alternative network layouts and determine their effectiveness by experimenting with an incremental mapdownloading application for road travelers that uses intelligent prefetching totakead\antage of theinfo-statiom. The prefetching algorithm uses location, route, and speed information to predict future data access. Our experiments show substantial performance improvement of the mobll~aware application in the info+tation network over a motill~unaware application in a conventions widearea wireless network. The prefetching alg~ rithm proves to hide latency from the user better than a naive prefetching algorithm. We have achieved between 16% to S0% improvements, depending on the prefetched amount. The results also suggest that a network design with frequent short range infestations is better than one with fewer, longer range stations.