François Bourdon

Agents mobiles et réseaux pair-à-pair : vers une gestion sécurisée de l'information répartie.

In this paper, we present a robust decentralized platform for data storage, built on a self-adaptive system composed of mobile agents on the top of a peer-to-peer network. A fragmentation redundancy and scattering (FRS) mechanism is used to provide fault tolerance capability, and information availability. To decentralize such a system, each fragment of information is as an autonomous bio-inspired agent capable to choose its own place of storage. We have implemented flocking rules to maintain a swarm of fragments. These local rules allows us to find few fragments to steer the whole flock towards a peer in the network in order to reconstruct a file. Another issue of our system is the optimization of available resources. We show how the flock mobility associated to a pheromone deposit can provide an eficient load distribution, while avoiding suspicious peers.

Repairing Flocks in Peer-to-Peer Networks

In this paper we consider the problem of reliable and robust data storage in peer-to-peer networks. The approach we follow builds a multi-agents system in which documents are split using a (m,n)-erasure code. Each generated fragment is embedded into an autonomous and mobile agent. This mobility gives them, for example, the ability to choose the network area where they want to be hosted. But this motion may have heavy consequences on the system robustness if the initial fragmentation parameters of a document are inappropriate. As a consequence, we focus on the problem of finding the suitable m and n values to apply to a document given the underlying peer-to-peer network properties.

Reducing Correlated Failures Impact in Peer-to-Peer Storage Systems Using Mobile Agents Flocks

This paper introduces MinCor (Minimum of Correlations), a decentralized simulated annealing algorithm designed for the data placement in peer-to-peer networks. Its goal is to reduce the correlated failures impact in such data storage systems. This data placement is realized using a multi-agent system which turns the documents into mobile agents flocks. After a network clustering step where highly correlated peers are regrouped together, the flocks executing MinCor are able to find a placement minimizing the number of agents on the same clusters. This placement is obtained in a decentralized way thanks to the environment exploration capabilities of the flocks. A set of experiments are performed on this system in presence of correlated failures. They show that, in practice, the expected placement is well obtained. They also show that, flocks using the MinCor algorithm suffer less simultaneous losses in presence of correlated failures than a mere random placement.

Data mobility in peer-to-peer systems to improve robustness

In this paper, we present the design of a robust decentralized peer-to-peer (P2P) platform for data storage. We use Fragmentation Redundancy and Scattering (FRS) mechanism on a file to provide fault tolerance capability, and information availability. To build a fully decentralized system, we consider each fragment of information as an autonomous bio-inspired agent capable to choose his own place of storage (to move from a peer to another peer). To reconstruct a file we have to gather a subset of fragments (defined by a minimal threshold to reconstruct the entire document), thus we have implemented flocking rules to maintain a swarm of fragments. These simple local rules allows us to find just one fragment to steer the whole flock towards a peer in the network. Another issue is the use of all network capabilities. We show in this paper how the flock mobility (based on pheromones) can provide the load distribution, while avoiding suspicious peers.

Searching Flocks in Peer-to-Peer Networks

Storage in peer-to-peer networks must be reliable and dependable. This reliability is partially dependent of data placement. In some approaches, information pieces are moving in the network to find a placement that optimizes different criteria. But this mobility may be a drawback for the localization of those pieces. In this paper, we propose to measure the impact that mobility of network objects has on their localization.