Information Technology Reference
In-Depth Information
more efficient. Structured P2P systems often use Distributed Hash Tables
(DHTs) as a substrate. In a Distributed Hash Tree approach such as Chord,
Pastry, and Tapestry, a global identification scheme for the peers is exploited
in order to decide what part of the search space the peer is associated with
[25]. On the other hand, unstructured overlay networks organize peers in a
random graph in a flat or hierarchical manner (e.g., using a Super-Peer layer)
and use flooding or random walks or expanding-ring Time-To-Live (TTL)
search, and so on, for the graph to query content stored by overlay peers.
Each peer visited will evaluate the query locally on its own content, and
can therefore support complex queries, which can be difficult to achieve in
structured networks.
A particular issue involved in the access of multimedia through P2P net-
works is how to create an efficient way to access the huge personal multime-
dia database with maximum security. Yang [26] focuses on three systems
(symbolic query on symbolic database, monophonic acoustic query on sym-
bolic database, and polyphonic acoustic query on polyphonic acoustic data-
base). He proposes a music indexing framework known as the Music-Audio
Characteristic Sequence Indexing System or MACSIS. Two protocols are used
with replicated databases separated in two different conditions. For the first
one, each query is processed by one response node at a time. For the second
one, each query is processed by several nodes simultaneously. On the other
hand, the protocol for general P2P networks tends to be divided into two
phases (a pre-search phase and a search phase). By using Yang's protocol,
a system throughput can be improved and waste reduced. Both (replicated
database and generic) protocols will involve query setup, query processing,
result generation, and process interruption. The advantage of the proposed
protocols is not limited to the music search domain. They can also be applied
to many other domains as long as the algorithm remains suited to certain
conditions. The disadvantage of the proposed protocols is that they assume
all nodes in the P2P network are cooperative, and will follow the specified
protocols without deviating from them.
Klampanos et al. [27] focus on full text searching and retrieval of docu-
ments and explore information retrieval (IR) approaches in P2P environ-
ments. They propose an architecture for IR over large semicollaborating P2P
networks. They define a network where peers have to cooperate in order to
perform IR without the need to share any detailed information with the rest
of the network. Lu [28] explores more on content-based retrieval in P2P net-
works that use a hybrid P2P architecture and focuses on the resource selection
conducted by directory nodes. For content-based retrieval in P2P networks,
both the retrieval accuracy and the efficiency of query routing are impor-
tant. The authors therefore measure the performance of different resource
selection and document retrieval algorithms in hybrid P2P networks by the
retrieval accuracy and efficiency of query routing. Most of the algorithms that
have been used in their experiments require a query matching rule, which
defines the number of query terms that need to be matched for a query.
