Computer Sciences

Pdas and handhelds: ict at your side and not in your face

J.Wishart , A.Ramsden , A.Mcfarlane


alt
37 views
alt alt alt alt alt
This paper stems from a trial at the Graduate School of Education in the University of Bristol supported by the Training and Development Agency for Schools and set up to evaluate the potential of personal digital assistants with internet access to support initial teacher training students in science. Fourteen students were given PDAs selected to represent the range of then currently available PDAs with mobile phone connectivity in the UK. As a result the following areas were identified by ITT students as definitely being of support during their course; the calendar or diary scheduler for organising themselves, the spreadsheet of attendance or mark book for organising their pupils and the use of a word processor to make notes on information and events immediately they are encountered. This electronic record was valuable as students could use it when they formally wrote up their experiences at a later date. However, internet access via the portable device was the application that the students considered to be most helpful to both teaching and learning. It was concluded that having the internet literally 'in the hand' affords access to tools and information in a way that does indeed function as distributed intelligence. The student plus PDA meets the person-plus description originally put forward by Perkins -they are able to do more than they can alone or with other resources. This is due to having ubiquitous access to internet as well as other electronic records and the less obtrusive nature of device. It appeared to be a particular benefit that the PDAs could be kept hidden in a pocket or handbag, at the student teacher's side instantly when needed and returned to the pocket once the need was gone, unlike other information and communications technology hardware that sits looming large in the classroom and demanding attention.

Le caching proactif dans les reseaux cellulaires 5g

M.Debbah , K.Hamidouche , E.Bastug , M.Bennis


alt
56 views
alt alt alt alt alt
Resume : Cet article etudie l'un des outils cles pour aller au dela des reseaux sans fils 4G, en exploitant le deploiement des reseaux a petites cellules , et plus precisement a travers le caching proactif. Ce nouveau paradigme permet d'exploiter les donnees du big data dans le domaines des reseaux sans fils. En munissant les stations de bases de capacites de prediction et en exploitant les developpements recents dans le stockage, la sensibilite au contexte et les reseaux sociaux, la gestion du trafic peut etre sensiblement amelioree. Ceci en servant les demandes previsibles des utilisateurs de maniere proactive, par la mise en cache des donnees dans les stations de base et les appareils des utilisateurs. Afin de montrer l'importance du caching proactif, nous etudions deux cas qui considerent la structure spatiale et sociale du reseau, ou le caching proactif joue un role crucial. Tout d'abord, nous proposons un mecanisme par lequel les fichiers sont mis dans le cache proactivement lorsque le trafic est faible en se basant sur la popularite des fichiers et les correlations entre les utilisateurs et les fichiers. Ensuite, en s'appuyant sur les reseaux sociaux et les communications appareil-a-appareil , nous proposons une procedure qui exploite la structure sociale du reseau en predisant l'ensemble des utilisateurs influents afin de mettre en cache les contenus strategiques et de les diffuser sur les connexions sociales via les communications AaA. L'exploitation du paradigme de mise en cache proactif peut apporter des gains importants en termes de reduction de trafic sur les liens backhaul et de satisfaction des utilisateurs.

Load balancing techniques for lifetime maximizing in wireless sensor networks

R.Kacimi , R.Dhaou , A.Beylot


alt
122 views
alt alt alt alt alt
Energy consumption has been the focus of many studies on Wireless Sensor Networks. It is well recognized that energy is a strictly limited resource in WSNs. This limitation constrains the operation of the sensor nodes and somehow compromises the long term network performance as well as network activities. Indeed, the purpose of all application scenarios is to have sensor nodes deployed, unattended, for several months or years. This paper presents the lifetime maximization problem in many-to-one and mostly-off wireless sensor networks. In such network pattern, all sensor nodes generate and send packets to a single sink via multi-hop transmissions. We noticed, in our previous experimental studies, that since the entire sensor data has to be forwarded to a base station via multi-hop routing, the traffic pattern is highly non-uniform, putting a high burden on the sensor nodes close to the base station. In this paper, we propose some strategies that balance the energy consumption of these nodes and ensure maximum network lifetime by balancing the traffic load as equally as possible. First, we formalize the network lifetime maximization problem then we derive an optimal load balancing solution. Subsequently, we propose a heuristic to approximate the optimal solution and we compare both optimal and heuristic solutions with most common strategies such as shortest-path and equiproportional routing. We conclude that through the results of this work, combining load balancing with transmission power control outperforms the traditional routing schemes in terms of network lifetime maximization.

Computational medical imaging (radiomics) and potential for immuno-oncology

E.Deutsch , N.Paragios , E.Limkin , R.Sun , L.Dercle , E.Zacharaki , C.Robert , S.Reuze , A.Schernberg , C.Ferte , C.Chargari , A.Dirand , A.Alexis


alt
27 views
alt alt alt alt alt
The arrival of immunotherapy has profoundly changed the management of multiple cancers, obtainingunexpected tumour responses. However, until now, the majority of patients do not respond to thesenew treatments. The identification of biomarkers to determine precociously responding patients is amajor challenge. Computational medical imaging is a promising and rapidlygrowing discipline. This new approach consists in the analysis of high-dimensional data extracted frommedical imaging, to further describe tumour phenotypes. This approach has the advantages of being noninvasive, capable of evaluating the tumour and its microenvironment in their entirety, thus characterisingspatial heterogeneity, and being easily repeatable over time. The end goal of radiomics is to determineimaging biomarkers as decision support tools for clinical practice and to facilitate better understandingof cancer biology, allowing the assessment of the changes throughout the evolution of the disease andthe therapeutic sequence. This review will develop the process of computational imaging analysis andpresent its potential in immuno-oncology. Resume : L'arrivee de l'immunotherapie a profondement modifie la prise en charge de multiples cancers, permettant des reponses tumorales jusqu'alors inesperees, meme si une majorite des patients ne repondent pas a ces nouveaux traitements. L'identification de biomarqueurs permettant de cibler les patients repondeurs est un enjeu majeur. L'imagerie medicale computationnelle est une discipline recente et extremement prometteuse. Elle consiste en l'analyse informatique d'images medicales et les traduit en donnees quantitatives complexes. Ces donnees de haute-dimension permettent une caracterisation et une analyse plus en profondeur du phenotype tumoral. L'imagerie medicale computationnelle presente l'avantage d'etre non-invasive, de pouvoir evaluer la maladie tumorale dans sa globalite, et de pouvoir etre repetee dans le temps pour suivre l'evolution tumorale au cours du temps. L'imagerie medicale computationnelle a pour objectif final de determiner des biomarqueurs d'imagerie apportant une aide a la decision medicale et permettant aussi de mieux comprendre la biologie du cancer. Cette revue developpera le processus de l'analyse en imagerie computationnelle, et presentera le potentiel de son utilisation en immuno-oncologie.
alt