Some contributions to the management of data in grids

Some contributions to the management of data in grids

Grid access control: Sygn

Data integration

Some contributions to Pervasive/Ubiquitous Systems

PerSE

Architecture of a PerSE Base

PerSE Layers

Context modeling and Context-aware pervasive applications

• Context = ?

• User/Application centric Context

• Context-aware application platform

Chains of trust certificates: if A trusts B and B trusts C then A trusts C… to some level

• Chains of trust certificates: if A trusts B and B trusts C then A trusts C… to some level

• Multi-layered adaptive certificates

• Trust computing and reputation

Authentication and Trust in pervasive systems

Mobile requests

• Information exchange between mobile peers (MANETs, VANETs)

• Trajectory aware geocast request routing

• Construction for each query of an ad hoc overlay of the peers that may participate to the delivery of the requested data

Data replication in small worlds

Replication in ad hoc networks

• Issues: instability, dynamicity, faults/disconnections

• Basic idea: adapt the replication heuristics to the actual usage of the data, e.g.:

• Data highly accessed => replicate
• Rare data and planned disconnection => replicate
• Overloaded peer => replicate most accessed data
• Etc.

• Pre-requisite: access logging + exchange of access data

• Assumption: no central coordination – local decisions

• Question : can local decisions lead to an acceptable (good ?) situation ?

Multimedia Content Adaptation

Semantic protection of multimedia data

• Multimedia => access control is not a binary decision

• Access control <=> content adaptation

• Coherence/Consistence of an adaptation

• Legality of an adaptation

• => Access control = semantic adaptation decision

Semantic clustering of multimedia queries: From query logging to database content and user profile discovery

• Large number of data stores disseminated

• Lack of indexes

• Issues :

• What are the pertinent/interesting data ?
• What are the users interest ?
• How to process a query ?

• Logging of the users’ queries + clustering + communication of clustering info => identification of the community of users’ interests

• Logging of a the queries issued by a specific user + matching to the community clustering => identification of the user’s profile

• Usage: query optimization, data store management, recommendation, social computing, etc.