Information Technology Reference
In-Depth Information
The first part of this general metric is used to collect all the transaction
information and the second part is used to compensate for the effect of the
community context factors on the final reputation result. This reasonably
effective model successfully collects all the useful information of the P2P
environment.
One major drawback of models like the ones mentioned above is that they
tend to concentrate on the relationship between a node and its direct neigh-
bors. Usually, the path to a desired resource, known as the resource path, is
obtained from numerous peers and their neighbors; such a resource path
affords the most useful information that could be employed during future
searches. Thus, we try to make an efficient use of this resource chain; the
entire resource chain is strengthened if the end of the chain provides reliable
service and it is weakened in the event of an unsuccessful download. Our
model is based on this reasoning.
4.2 The Approach
The major aspects of our approach are that our model is a best effort service
and that it does not include the guaranteed service feature. In a P2P network,
our model always tries to choose the most suitable node from among a set
of possible peers for the purpose of downloading. In our approach, the reli-
ability of a node is judged by considering credibility of the node. We present
the following scenario to indicate the chain of events when a node enters a
new P2P community.
Internet protocol (IP) and media access control (MAC) addresses provide
a peer with a unique identity. Therefore, as a node enters a P2P network,
it broadcasts or multicasts connection requests and waits for replies from
the rest of the peers. After a sufficient number of replies are received, the
node can choose neighbors from among the peers that have replied to the
node's connection requests. This choice is made on the basis of IP addresses
that is, neighbors across sufficiently different IP address ranges are chosen
so that peers from different IP groups can be accessed. Since making this
choice is just the beginning of the search process, it is possible that there
are malicious users among the neighbors chosen. This is taken care of by
the fact that a node's neighbor table can be updated to add new neighbors
and delete malicious nodes on the basis of subsequent transactions. We will
discuss this in the update credibility. Network neighbors can be obtained
as explained above. Later, the desired resource can be found by sending out
search queries to the node's neighbors and the neighbors' neighbors. These
search paths, for example, the path from the node that sends out the request
to the node at which the desired resource is found, can sometimes comprise
several long chains. Therefore, the best chain from among such candidate
