Mid Sweden University

Next generation context-aware mobile applicationswill require a continuous update of relevant information about auser’s surroundings, in order to create low latency notificationsand high quality of experience. Existing mobile devices alreadycontain a large number of built in sensors which are capableof producing huge amounts of sensor data, exceeding boththe capacity of the local storage and the Internet connection.Therefore, we will in this paper study the limits when sharingcontextual information from mobile devices, as well as finding theimpact of this information overload for the Internet-of-Things.Furthermore, we present an evaluation model for assessing theeffort required to present applications with relevant contextinformation. In conclusion, the model shows that one feasiblesolution for the future Internet-of-Things is a peer-to-peer basedsolution which can control the flow of information without anycentralized authority, to circumvent earlier limitations.

The increase in sensor-assisted applicationssuch as sensing campaigns and interactions in large-scalepopulations put new demands on seamless sharing ofsensor information on a global scale. Existing support islimited due to simplified information models reflectingnarrow application scenarios and poorly scaling architectures.In response to this, we present an application architecturefor pervasive internet-of-things applications whichcircumvents these limitations. Our architecture enablesapplications to utilize information from sensors and wirelesssensor networks via a peer-to-peer overlay. Theoverlay shares sensor information in an generic informationmodel, which is extensible and thus enables intelligentapplication behavior. Our evaluation of the architectureindicates that it supports the sharing of sensor informationon a global scale in sensor-assisted applications, with lowresponse times. Finally, we evaluate the feasibility ofrunning such applications in end devices with limitedresources in a sensor-assisted application prototype.

Research in Internet-of-Things infrastructures hasso far mainly been focused on connecting sensors and actuatorsto the Internet, while associating these devices to applicationsvia web services. This has contributed to making the technologyaccessible in areas such as smart-grid, transport, health, etc.These early successes have hidden the lack of support forsensor-based applications to share information and limitationsin support for applications to access sensors and actuatorsglobally. We address these limitations in an novel open-sourceplatform, MediaSense. MediaSense offers scalable, seamless,real-time access to global sensors and actuators via hetero-geneous network infrastructure. This paper presents a setof requirements for Internet-of-Things applications support,an overview of our architecture, and application prototypescreated in order to verify the approach in a test bed withusers connected from heterogeneous networks.

The SensibleThings platform is an open source architecturefor enabling Internet-of-Things based applications.During its development, multiple problems have been faced andsolved, for example issues related to networking, information dissemination,sensors, and application access. This paper describesthese problems and the technical solutions that are implementedin the platform. We also present the current progress and aseries of demonstrator applications, which show the wide rangeof possibilities enabled by the platform. Finally, we present futurework and how it will be used in future research endeavors andcommercial interests.

Context-aware applications require local access to current and relevant views of context information derived from global sensors. Existing approaches provide only limited support, because they either rely on a network broker service precluding open-ended searches, or they adopt a presence model which has scalability issues. To this end, we propose a fully distributed architecture employing context user-agents co-located with data-mining agents. These agents create and maintain local schemas using ranking of global context information based on context proximity. Continually evolving context information thus provides applications with current and relevant context views derived from global sensors. Furthermore, we present an evaluation model for assessing the effort required to present local applications with current and relevant contextual views. We show in a comparison with earlier work that the approach achieves the provisioning of evolving context information to applications within predictable time bounds, circumventing earlier limitations.

This article analyzes the challenges of supportingcontinual changes of context information in Internetof-Things applications. These applications require aconstant flow of continuously changing information fromsensor based sources in order to ensure a high quality-ofexperience.However, an uncontrolled flow between sourcesand sinks on a global scale wastes resources, such ascomputational power, communication bandwidth, andbattery time. In response to these challenges we presenta general approach which focuses on four layers wherewe provide a proposed solution to each layer. We haverealized the general model into a proof-of-concept implementationrunning on devices with limited resources,where we can moderate the information exchange basedon relevance and sought after quality-of-experience bythe applications. In conclusion, we evaluate our solutionand present a summary of our experiences regardingthe impact of continuously changing information on theInternet-of-Things.

Current context-aware applications often use thelocation of a user as the only indication of the current situation.These existing applications are therefore limited in their situationawareness, because of the poor indoor resolution of the locationsensor and its high resource consumption. In response to theselimitations we present an approach to estimate the contextualsituation of a user without using resource inefficient locationsensors. Our proposed solution utilizes a wide range of lowpowered sensors, together with two modified machine learningtechniques to estimate the situation in a more resource efficientmanner. Simulations and a proof-of-concept application show thatthe situation of a user can be determined within 50 ms at anaccuracy above 90%, when only using the low energy sensorsavailable on a smartphone and its limited processing power.