The usage of peer-to-peer networks often follows a query-response paradigm, where users initiate searches which then have to be routed over the network efficiently. This paper proposes an approach for self-organizing information distribution in peer-to-peer networks that inverts this paradigm: data objects actively travel through the net to those nodes for which they are relevant. The underlying mechanism is rooted in the principles of Swarm Intelligence and relies on the dissemination of artificial pheromones, where each pheromone represents one particular relevance criterion that applies to a given data object. Data objects leave trails of these pheromones at each node they visit and move along gradients of pheromone concentration to regions in which they are relevant.
[1]
Lars Schmidt-Thieme,et al.
Emergence of Spontaneous Order Through Neighborhood Formation in Peer-to-Peer Recommender Systems
,
2008,
ArXiv.
[2]
Marco Dorigo,et al.
The ant colony optimization meta-heuristic
,
1999
.
[3]
Ronaldo Menezes,et al.
On the implementation of SwarmLinda
,
2004,
ACM-SE 42.
[4]
Gerti Kappel,et al.
Using taxonomies for content-based routing with ants
,
2007,
Comput. Networks.
[5]
Riccardo Poli,et al.
New ideas in optimization
,
1999
.
[6]
Mitchel Resnick,et al.
Turtles, termites, and traffic jams - explorations in massively parallel microworlds
,
1994
.