MULUS : un réseau social distribué pour une consommation sélective de contenus composites

Amina Chaabane 1
1 LAAS-SARA - Équipe Services et Architectures pour Réseaux Avancés
LAAS - Laboratoire d'analyse et d'architecture des systèmes [Toulouse]
Abstract : With the evolution of new technologies, social networks (SN) are extensively used for data sharing between users. They have two major shortcomings. First, much of the shared content does not interest those users, which unnecessary increases the bandwidth consumption and degrades the SN quality of service. Second, these SNs use a centralized infrastructure managed by a service provider. This latter imposes some constraints related to storage capacity and offered services and can also involve fees for using the SN. In this context, we developed a distributed SN (DSN) based on peer-to-peer (DHT) publish/ subscribe system. It deploys the event service on user devices connected in P2P manner. Our DSN ensures selective data sharing where users receive content generated according to their interests, which are specified using the publish/subscribe paradigm. Our DSN expresses user interests with semantic and composites subscriptions. To address the semantic aspect of subscriptions, we proposed to use a structured domain ontology, which is shared between all users. Then, we proposed an indexing method of this ontology based on prime numbers. This method provides a semantic routing on structured P2P topology (DHT) of our system. To preserve the expressivity of composite interest, we have defined a cube structure to index subscriptions on the DHT nodes. This structure provides filtering of composite events by a simple binary research performed on this cube. Although DHT used in our DSN provides self-organizing nodes, handling data loses still remains a challenge with highly dynamic nodes. Then, we proposed a replication strategy based on nodes availability in order to maximize the availability rate desired by the user. Experimental results using the FreePastry simulator demonstrate the outperformance of our DSN over existing SNs, in terms of scalability, latency and network overhead.
Document type :
Theses
Réseaux et télécommunications [cs.NI]. INSA de Toulouse, 2016. Français
Liste complète des métadonnées

Cited literature [94 references]  Display  Hide  Download

https://hal.laas.fr/tel-01376046
Contributor : Arlette Evrard <>
Submitted on : Tuesday, October 4, 2016 - 10:32:36 AM
Last modification on : Thursday, January 11, 2018 - 2:04:51 AM
Document(s) archivé(s) le : Friday, February 3, 2017 - 2:50:35 PM

Identifiers

  • HAL Id : tel-01376046, version 1

Citation

Amina Chaabane. MULUS : un réseau social distribué pour une consommation sélective de contenus composites. Réseaux et télécommunications [cs.NI]. INSA de Toulouse, 2016. Français. 〈tel-01376046〉

Share

Metrics

Record views

129

Files downloads

228