SmartP2P

1
SmartP2P

• A Multi-objective Framework for Finding Social
  Content in P2P Smartphone Networks

• Andreas Konstantinidis, Christos Aplitsiotis and
  Demetrios Zeinalipour-Yazti
• Department of Computer Science, University of
  Cyprus
• Department of Computer Science and Engineering,
  Frederick University Cyprus

Query Routing Trees (QRTs)
Multi-Objective Optimization

• QRTs are optimized to
• minimize energy during search
• minimize response time during search
• maximize recall of the user query.

Pareto-Front (PF) a set of non-dominated
solutions.
Optimizer (MOEA/D)
SmartP2P Framework

• Search Techniques
• Centralized Search objects and tags are all
  uploaded on S.
• Distributed Random Search objects and tags are
  all stored in-situ. Query node downloads from S a
  random QRT and conduct a P2P search.
• SmartP2P Search objects and tags are all stored
  in-situ. Query node downloads from S a PF and
  selects a QRT to conduct a P2P search.

SmartLab (A Programming Cloud of Smartphones)
Datasets / Scenarios MSS-1 GeoLife (mobility)
DBLP (social) GeoLife is a real dataset by
Microsoft Research Asia, includes 1,100
trajectories of a human moving in Beijing over a
life span of two years (2007-2009). DBLP is a
real dataset by the DBLP Computer Science
Bibliography website, includes over 1.4 million
publications in XML format. MSS-2 Pics 'n'
Trails (mobility and social) Real dataset
composed of GPS traces of a user moving in Tokyo
(2007-2008) and a collection of photos, tagged by
the location taken and a short description.
SmartP2P Prototype on SmartLab
http//smartlab.cs.ucy.ac.cy/
Experimental Results SmartP2P Evaluation on
SmartLab
Acknowledgements This work was supported by the
last authors Startup Grant, Funded by the
University of Cyprus