Infrastructure for Context-Aware Wireless Network Applications

点击翻译按钮获取中文摘要

Information systems are now mobile, wearable, multimodal, real-time, scalable from workstations todesktops to notebooks to palmtops to cellular, collaborative and ubiquitous. Network access is becomingincreasingly important as a part of the computing infrastructure. Unfortunately, for the near term, un-limited bandwidth, anytime access to a network is not feasible, particularly in large urban environmentswhere interference, occlusion and collision are ongoing problems. To alleviate this, we are developing aset of context-aware Autonomous Information Retrieval (AIR) pods. These are small, hardened low-costcomputers that require only electric power. AIR pods have sufficient local storage to hold relatively staticinformation and are equipped with two wireless interfaces: one short-range, unlicensed high-speed interfacesuch as IEEE 802.11 and one long-range, low-speed interface such as CDPD or other Wide Area Networkconnection which may be intermittent. AIR pods can be either stationary or mobile. Stationary AIR podscan be attached to lamp posts and traffic lights, hidden in lighted store signs or in subway stations. MobileAIR pods can be attached to delivery trucks, postal service vehicles, buses, police cruisers, taxis or othervehicles that roam city streets. AIR pods are also small enough to be carried in backpacks. We propose to develop and implement the prototype hardware and software AIR idea, and explore whatkinds of infrastructure support are necessary to make these devices a key component of network applications.Research issues addressed will be cooperative data sharing, resource scheduling and anticipatory caching,message propagation and wide-area resource discovery. The proposed mobile networking infrastructure will be tested in a demanding set of context-aware mo-bile research projects. The first project is a wearable augmented reality system that allows outdoor users to tour a campus interactively. A second related project allows indoor users to collaborate with those outside over wireless networks. The third project is a mobile robot sensing system that can autonomously explore the campus and create rich 3D, texture mapped, site models. All of these applications need to interact with host computers through limited network access, and we propose to optimize this interaction over bandwidth, devices, and locality using context-aware wireless networks.