DocumentCode
3331798
Title
Collision Avoidance in Hierarchical Peer-to-Peer Systems
Author
Teo, Yong Meng ; Mihailescu, Marian
Author_Institution
Dept. of Comput. Sci., Nat. Univ. of Singapore, Singapore
fYear
2008
fDate
13-18 April 2008
Firstpage
336
Lastpage
341
Abstract
In a two-level chord-based hierarchical peer-to-peer system, nodes from the first-level overlay network, called supernodes, act as gateways to peers organized in the second-level overlay. Collision occurs when peer-to-peer operations create more than one supernode with the same node identifier in the first-level overlay. Collisions are reported to enlarge first-level overlay by more than 12 times the ideal size and thus increase the lookup path length. Other consequences of collisions include lookup failures even though the lookup resources exist in the network and reduced scalability of the system. In contrast to collision detection and correction techniques, this paper proposes: (i) a collision-free approach for the join and leave operations, and (ii) a new stabilization scheme to recover from second-level overlay node failures. However, avoiding collision due to supernode failures is complex and remains a challenge. Simulation results show that collisions can be avoided and our algorithms improve the overall system performance.
Keywords
peer-to-peer computing; telecommunication congestion control; chord-based hierarchical peer-to-peer system; collision avoidance; collision-free approach; overlay network; stabilization scheme; supernodes; Collision avoidance; Computer networks; Computer science; Costs; Indexing; Mechanical factors; Network topology; Peer to peer computing; Scalability; System performance; collision avoidance; distributed hash tables; hierarchical; peer-to-peer;
fLanguage
English
Publisher
ieee
Conference_Titel
Networking, 2008. ICN 2008. Seventh International Conference on
Conference_Location
Cancun
Print_ISBN
978-0-7695-3106-9
Electronic_ISBN
978-0-7695-3106-9
Type
conf
DOI
10.1109/ICN.2008.110
Filename
4498186
Link To Document