| This thesis introduces network-aware prefetching for mobile users in a heterogeneous wireless network. Network awareness is achieved by analyzing the location and movement of the user relative to the physical location of the nearby networks. Users optimize their overall utility cost by adapting their prefetching strategy based upon their likelihood of leaving the current network coverage area. A multiuser threshold algorithm is proposed using network-awareness, network costs and bit-rates. Under a standard non-priority system, multiple prefetching users cause a positive feedback, resulting in more aggressive prefetching, increasing WLAN traffic that potentially renders the system unstable. To increase the scalability and performance gains from WLANs, a priority queuing mechanism is proposed that applies lower priority to prefetch documents. This strategy keeps web response time low, essentially eliminating feedback and extending the scalability of the system, while allowing users to achieve considerable cost reduction for any number of users. |
