Intelligent information dissemination in collaborative, context-aware environments

Abstract

Today's mobile computing environments aggregate many entities, all of them with a number of internal sensors, processing applications and other, user given information that can be shared with others over the available networks. Tomorrow's ubiquitous computing environments, where the number of sensors is assumed to be even significantly higher, face the challenge that information exchange between entities has to be controlled, not only to protect privacy and to unburden the wireless networks -but also to reduce load on the receiving entities that have to process all incoming information.The approach we propose in this work measures the importance of some information and the interest of the receiver in it, before it is transferred. We apply it in two scenarios with limited resources. In vehicle-to-vehicle communications the transmission time while cars meet and bandwidth availability of the wireless channel is the critical resource, forcing to reduce information exchange. In context inference on mobile devices processing power and battery life are limited and responsiveness to user actions is most important. Hence only the most important information should be processed.

Segmenting Bayesian networks for intelligent information dissemination in collaborative, context-aware environments with Bayeslets

“…The application of the Bayeslet concept requires the introduction of an additional step before the actual context inference process: single Bayeslets have to be assembled to the BN eventually evaluated. In order to allow for efficient and precise Bayesian context inference taking into account the available information at request time, the concept of Dynamic Composition of Bayeslets has been introduced by the authors in [12,13]. To save transmission costs and inference time, the relevant Bayeslets have to be selected.…”

“…If, for instance, the car is located on the centre lane of a three-laned road, the preceding vehicle can be located on the same, the right or the left lane. Context inference can infer this from the available information and provide this information to CACC [13].…”

“…More importantly however, it offers the possibility that more specific utility functions and decisions are defined. The following demonstrates the latter option, according to [13] has been specified. The basic utility functions measure the Safety and the Efficiency utility of an action.…”

“…Variation of the value of information for different connection states of the Bayeslets for radar and V2V communications. Taken from[13].…”

Information Management for Adaptive Automotive Human Machine Interfaces

“…Further, the increasing number of available data will only be beneficial if information is fused adroitly [6]. This is particularly true if information is exchanged between several entities as in vehicular ad hoc networks [9]. Information fusion can be applied on several levels and helps to reduce the amount of messages to present [18].…”