Project

mobile networking; quality of experience; human factors; mobile Internet; mobility; quality of service; machine learning

Lead
Mobiles Internet (MI) erlaubt Internetdienste von unterwegs aus zu nutzen, zum Beispiel auf einem Smartphone oder Laptop, im Auto oder im Zug. Heute werden in einem zunehmenden Masse wichtige berufliche und private Aufgaben 'on the go' über MI erledigt, wie zum Beispiel die Recherche nach Informationen oder die Fertigstellung eines dringenden Dokuments. Allerdings hängt die wahrgenommene Qualität von MI-Diensten stark von Ort, Zeit, Bewegungsverhalten und anderen Faktoren ab. Wenn aufgrund ungenügender MI-Qualität eine Aufgabe nicht erledigt werden kann, kann das zu Frustration und Enttäuschung führen, und zudem einen negativen Einfluss auf das berufliche oder private Leben der Benutzer haben.Wir sollten die Erwartungen und Bedürfnisse der Benutzer bezüglich MI besser verstehen. Wir sollten versuchen, den Benutzern in jeder Situation die bestmögliche Qualität zu ermöglichen, oder sie alternativ über mangelhafte Qualität zu informieren, so dass sie ihre Erwartungen anpassen können.
Lay summary
Gegenstand und Ziel Unser Hauptziel ist es, die Erwartungen und Bedürfnisse der Benutzer zu verstehen und den MI-Diensten ein Mittel zu geben, sich so anzupassen, dass diese Erwartungen und Bedürfnisse bestmöglich erfüllt werden können. Zudem sollten die Benutzer im Voraus über zu erwartende Einschränkungen informiert werden. In diesem Projekt wollen wir ein prediktives Modell für MI-Qualität entwickeln, welches auf gesammelten realen Erfahrungsdaten der Benutzergemeinschaft beruht. Stellen Sie sich vor, Sie pendeln jeden Tag mit dem Zug zwischen Bern und Zürich. Heute wollen Sie an einer Videokonferenz teilnehmen während Sie im Zug sind. Basierend auf unserem prediktiven Modell kann das Videokonferenzprogramm nun zum Beispiel proaktiv die Übertragunsrate anpassen, für Unterbrechungen (z.B. Tunnel) kompensieren und so das Auftreten von Störungen minimieren. Alternativ könnte das Programm Sie auch informieren, dass eine Videokonferenz zurzeit auf der gewünschten Strecke nicht möglich ist, und Ihnen vorschlagen auf einen simplen Sprachanruf auszuweichen oder für die Videokonferenz in Bern zu bleiben. Soziotechnischer Kontext Unsere Forschung wird Benutzern helfen MI-Dienste effektiver und frustrationsfreier zu nutzen und daher private und berufliche Aufgaben besser erledigen zu können. Die Benutzer werden ein besseres Verständnis dafür gewinnen, was sie in welcher Situation von MI erwarten können. Unsere Ergebnisse können auch Anbietern von MI-Diensten helfen ihre Kundenzufriedenheit zu erhöhen, indem sie zum Beispiel eine durchgehende Minimalqualität sicherstellen.

Direct link to Lay Summary

Last update: 09.02.2015

Lead
L’Internet Mobile (IM) permet l’accès à différents services Internet grâce à un smartphone ou un ordinateur portable, en voiture ou dans le train. Ces applications et services contribuent à l’accomplissement par l’utilisateur de tâches professionnelles ou personnelles lors de ses déplacements, p. ex., passer un appel, accéder à une information ou terminé l’édition d’un document avant la date limite d’un projet. Cependant, la qualité de l’IM varie en fonction de l’endroit, de l’heure, du niveau de mobilité et d’autres facteurs contextuels. Parfois, l’utilisateur est frustré ou déçu ne pas avoir pu réaliser sa tâche à cause d’une mauvaise qualité de l’IM. Cela peut influencer négativement sa vie personnelle ou professionnelle.Il est nécessaire de comprendre les attentes et besoins des utilisateurs de l’IM pour leur fournir le meilleur service IM possible ou au moins de les informer quand la qualité est mauvaise afin qu’ils ajustent leurs attentes et gèrent leurs tâches différemment.
Lay summary
Sujet et Objectif Notre objectif principal est de comprendre les besoins et attentes des utilisateurs de l’IM et d’adapter les applications mobiles et services pour mieux répondre à ceux-ci et, dans le cas où ils ne pourraient pas être atteints, informer l’utilisateur, idéalement à l’avance (de façon prédictive). L’objectif de ce projet est de construire un modèle de prédiction de la qualité de l’IM grâce à la collecte de données sur la qualité de l’IM par un grand nombre d’utilisateur utilisant la même application par le passé, au même endroit, jour et heure. Par exemple, vous voyagez en train tous les jours entre Bern et Zurich. Aujourd’hui, vous souhaitez participer à une vidéo-conférence lors de votre trajet en train. Le service pourrait automatiquement ajuster la qualité audio et vidéo basé sur la prédiction de la qualité de l’IM. Il serait capable de vous fournir la meilleure qualité possible en compensant les pertes de connexion et les changements d’antennes. Il pourrait également vous informer que la participation à la vidéo-conférence ne sera pas possible et vous suggérer de participer seulement en audio, ou encore de rester dans un café à Bern et de prendre un train plus tard pour une meilleure qualité. Contexte socio-scientifique Notre recherche aidera les utilisateurs à mieux gérer leur utilisation des différents services et applications mobiles tout en réduisant les frustrations liées à la qualité de l’IM pour mieux répondre à leurs besoins personnels et professionnels. Les utilisateurs comprendront ainsi mieux la qualité de l’IM qui leur est fournie et ce qu’ils peuvent en attendre dans différents contextes. De plus, comprendre le comportement des utilisateurs et leurs besoins aidera à mieux définir les règles des fournisseurs de l’IM, protégeant les utilisateurs et assurant un niveau minimum de qualité.

Direct link to Lay Summary

Last update: 09.02.2015

Lead
Internet Mobile (IM) permette ai suoi utilizzatori di accedere a servizi internet con i loro smartphones e/o laptops quando in movimento in auto o in treno. Le suddette applicazioni e servizi contribuiscono a completare compiti professionali o personali importanti quando gli utilizzatori sono in viaggio, accedere a delle informazioni o terminare dei documenti prima di una scadenza di consegna. Purtroppo, la qualità dell’IM varia a dipendenza della sua posizione, l’ora, la sua mobilita e altri fattori della sua situazione. Può accadere che la connettività offerta dall’IM sia scarsa e che l’utilizzatore non riesca a portare a termine uno dei suoi compiti. Questo può influenzare negativamente la sua vita.È necessario comprendere i bisogni dell’utilizzatore e le sue attese nell’IM. Dobbiamo essere in grado di fornire il migliore servizio IM, o almeno informare gli utilizzatori della sua prevista bassa qualità così che essi possano gestire i loro compiti in modo differente.
Lay summary
Il nostro scopo principale è di comprendere i bisogni e le attese degli utilizzatori dell’IM e di adattare le applicazioni mobile per adempiere al meglio a questi ultimi. Nel caso in cui non si possa adempiere a questi requisiti l’utilizzatore deve essere informato. Idealmente ciò andrebbe fatto in anticipo, cercando di prevedere la qualità della connettività dell’IM. L’obiettivo di questo progetto è di creare un modello per prevedere la qualità della connessione offerta dall’IM. Il modello sarà basato sui dati collezionati da molti utilizzatori dell’IM che hanno utilizzato la stessa applicazione nel passato nel medesimo posto, giorno e ora. Per esempio: l’utilizzatore si reca tutti i giorni in treno da Locarno a Lugano. Oggi, esso desidera partecipare a una videoconferenza mentre si trova in treno. L’applicazione utilizzata potrebbe attivamente adattare la qualità video a dipendenza della capacita offerta secondo il modello della qualità dell’IM. Il modello può creare le condizioni per fornire all’utilizzatore la miglior qualità possibile mentre gestisce le perdite di connessione o gli spostamenti dell’utilizzatore da un’antenna di ricezione all’altra. Alternativamente, l’utilizzatore può essere informato del fatto che la sua partecipazione alla prevista videoconferenza sarà impossibile per quel treno. Il sistema può suggerire all’utilizzatore di partecipare (solo chiamata vocale), o di restare in un bar a Locarno per la videoconferenza e prendere il treno seguente per una migliore qualità di trasmissione. Il nostro lavoro aiuterà l’utilizzatore ad adempiere ai suoi compiti gestendo meglio le applicazioni mobili senza essere frustrato. Gli utilizzatori comprenderanno meglio quale sia la reale qualità offerta dall’IM e in quali contesti. Comprendere il comportamento dell’utilizzatore e le sue necessità aiuterà a meglio definire le regole per i fornitori della connessione dell’IM proteggendo l’utilizzatore assicurandogli un margine definito di qualità.

Direct link to Lay Summary

Last update: 09.02.2015

Lead
Mobile Internet (MI) enables its users to access diverse Internet applications ‘on the move’, for example on a smartphone or a laptop in a car or a train. On a growing scale, these applications contribute to achieving important professional or personal tasks when out and mobile. For example, having an important call, accessing information or finalize documents before a deadline. However, the quality of the MI the user experiences vary depending on the user’s location, time, mobility level and other context factors. If it happens that a user does not manage to achieve the task due to a low quality offered by the MI, he/she may become frustrated. Additionally user’s professional or personal life may get negatively influenced.It is necessary to understand individual’s needs and expectations for MI. We need to provide them the best possible MI service, or at least inform them about the low quality, such that they can adjust their expectations and manage their tasks differently.
Lay summary
Subject and Objective Our principal aim is to understand the needs and expectations of the users for Mobile Internet (MI) quality and to adapt the mobile applications and services to best meet these expectations and needs. In case these cannot be assured we need to inform the user, ideally ahead of time (i.e., in a predictive manner). The objective of this project is to build a prediction model for MI quality based on quality data collected by many MI users being in the same location, time and using the same application in the past. For example, you are commuting every day from Bern to Zurich by train. Today, you wish to participate in a videoconference happening while you are in that train. The service may proactively adapt the video and audio quality based on the MI quality prediction model. It can manage to provide you the best possible quality while compensating for disconnections and network handovers. Alternatively, it may inform you that the videoconference participation is impossible to be achieved and suggest you either to participate via an audio call only, or stay in café in Bern and take a later train for the best quality. Socio-scientific context Our research will help users to manage better their use of diverse mobile applications and services, while not getting frustrated with the MI quality to better meet their professional or personal goals. The users will better understand what MI quality is provided to them and what they can expect in which context. Understanding users behaviour and needs will also help to better define policies for providers of MI, protecting the users and assuring their quality, for example at some minimum requested level.

Direct link to Lay Summary

Last update: 09.02.2015

Active participation

Self-organised

Self-organised

Communication	Title	Media	Place	Year

Title	Year

Number	Title	Start	Funding scheme

On a growing scale, we use mobile applications and services ‘on the move’ (e.g., in a car or a train) to fulfil critical goals that rely on prompt and to-the-point mobile interactions ranging from last minute information about the important meeting we are about to join, to a health procedure that can save our fellow train passenger’s life. The challenge arises from the fact that the service-ability of these applications and our experience rely on the connection quality provided by the underlying mobile networking infrastructure, supporting the data exchange between the mobile device we use (e.g., a smartphone) and some remote application server(s) (situated somewhere “in the cloud”). This underlying mobile networking infrastructure we denote as the Mobile Internet (MI), and its connection quality as MIQ. As research studies show, the MIQ is a best-effort level quality. Nevertheless, the users develop high MIQ expectations; the more they value the utility of the mobile application at hand, and the more critical the use of this application is to their personal goal, the more critical they become of accepting a given experienced MIQ. It is especially so, if the experienced MIQ is surprisingly different from their expected MIQ. The MIQ contributes to the overall user experience (QoE); the QoE also embraces the service design (user interface) and service information quality, which are not treated in here. The MIQ is defined by its speed, accuracy, dependability of the mobile connection establishment and then the subsequent MIQ stability. We denote these as Quality of Service (QoS) criteria for MI, which further depend on the user’s complex context (location, time, mobility speed and heading, access network, etc.). We present research project on developing timely and accurate contextual MIQ model that predict MIQ for mobile users in different context. We research the predictive accuracy, speed and computational complexity of the MIQ model, as well as we focus on precision and recall of this model to predict MIQ ‘surprises’, via modelling and predicting unexpected (by the user) MIQ conditions (e.g., bad access network). The model is based on a proactive assessment of QoS and MIQ context, and predictive modelling of MIQ for any future point in time. The MIQmodel is user-centric and enables to proactively control interactive applications, in order to meet the user’s expected MIQ levels and enable them to meet their goals. It contrasts with the current operator-based approaches - focusing on providing access to their wireless networks, yet not assuring the user’s experience for mobile applications. We evaluate our research with real network quality measurements data collected ‘in situ’ with a set of real mobile users ‘on the move’. The proposed project lies at the frontier of Human-Computer Interaction (HCI), pervasive and ubiquitous computing, network performance management, machine learning and predictive analytics. The project is timely - there is a significant need for research and commercial interest for mobile networking solutions, as smartphones became personal communication and computing devices affordable for the masses. Additionally, there is a growing number of applications providing time-constrained services to their mobile users, relying the success of their delivery, hence their users’ experience, on the ‘best-effort’ quality of the underlying wireless networks. To conduct the proposed research we have the necessary equipment including latest smartphones as well as infrastructure for collecting smartphone-based data from mobile users called mQoL Living Lab (mQoL-LLab). So far our research focused on the ‘basic’ system (QoSIS.com) for collecting network performance data from mobile users and using simple machine learning on that data. The next step taken in the project is to do in depth research and model the user’s experience and expectations in context, and its relation with mobile networking. Mr Fanourakis - the PhD student proposed for this project has already started some initial research towards it. There are various potential exploitation areas of the MIQmodel research, including novel application development leveraging MIQmodel in their application quality management, and mobile networking solutions by network operators, leveraging the model towards enhancing their users’ experience ‘on the move’. The MIQmodel will be initially exploited via our on-going research projects and collaborations in the area of mobile networking, especially applied in healthcare, where the lack of quality assurance for mobile services can even result in endangering the patient’s life. It is our moral obligation to ensure quality in this emerging area, propelled by the exponential needs of the ageing, and chronically ill population.