The use of a PDA with ad-hoc built-in information retrieval functionalities can help people in visiting an historical site in a natural manner instead of traditional audio/visual pre-recorded guides. The aim of this work is to build a versatile virtual-guide system adaptable to the user needs of mobility and therefore usable on different devices (e.g. PDAs, Smartphones). An information retrieval service is included and is easily accessible through a spoken language interaction. The system takes the advantages of chat-bot and speech recognition technologies, allowing a natural interaction with the user. The system has been implemented on a Qtek PDA with Windows Mobile 2003 with the aim of assisting an user during a visit to a cultural heritage institution.
Simultaneous operation of multiple network interfaces can benefit mobile devices with diverse network interfaces. This paper outlines an IP-level network activation strategy that aims at optimum use of the complementary features of the underlying networks. A key design issue in such strategies is the energy cost of different network interfaces. As a step towards validating and fine-tuning our strategy, we devised an experimental setting to measure the relative energy consumption of WiFi and GPRS interfaces in two modern mobile devices. The paper reports the experimental method and results. Our study indicates that the proposed strategy –in this case, reaching devices from the GPRS interface and, whenever appropriate, handing over some sessions to the WiFi interface– is plausible from energy consumption viewpoint and in some cases it outperforms the existing methods. The energy consumption model obtained will be used to make the network activation strategy energy aware. Key words—Mobile devices, multiple network interfaces, connectivity management, energy cost, WiFi-GPRS.
In this paper, we present a policy-based architecture to be used in identifying the location of users within a pervasive system environment. Mobile users can define their own policies governing the granularity of their location identification. Such granularity is governed by a combination of policy predicates and premise topology information. Privacy can vary in a spectrum of levels from total obstruction to total transparency of location information. Sensors and agents coupled with such sensors continuously forward user locations and policies to a policy and state repository. The policy and state repository can be queried for user location information. Location information is filtered to comply with user defined policies at the agents closest to the source of the query. Historical information is eventually archived at the policy repository, and can also be used to query for possible predictions of futuristic location of users.
To overcome shortcomings of Exact Matching Method (EMM) and Substitute Description Method (SDM), an Inexact Matching Method Based on Ontology and Semantic Distance (OSDIMM) is introduced. This method, firstly, describes resources information with Ontology languages to get Resource Ontology Description (RODs); then, infers Concept Hierarchy Trees and Property Hierarchy Trees from RODs; lastly, computes semantic distance between a required resource and each of existing resources based on Hierarchy Trees, and selects resources by semantic distances and thresholds. The results of experiments indicate: firstly, OSDIMM is averagely 3.8889 times as large as EMM at the aspect of matching count when the threshold of holistic semantic distance is zero and inputs are same, i.e., OSDIMM can utilize fully resources discovered than EMM; then, OSDIMM can adapt more smartly than SDM to the condition of lots of resources and properties being in pervasive computing environment; at last, it can differentiate the weightiness between entities and properties, and can describe the degree of resources satisfying query with property tuple suitability degree.
