Information Technology Reference
In-Depth Information
Peer 1(
P
1
) shares files in three categories:
c
1
,
c
2
,
c
3
where it shares maximum number
of files in category
c
1
, followed by category
c
2
and so on. On the other hand, Peer 3
(
P
3
) shares maximum number of files in category
c
2
as it's the most popular among
the categories chosen by it, followed by
c
4
and so on.
(3)
Query initiation model
: The peers usually query for files that exist on the net-
work and are in the content category of their interest [8]. In each cycle of simulation,
active peers issue queries. However number of queries a peer issues may vary from
peer to peer, modeled by
Poisson
distribution as follows. If
M
is the total number of
queries to be issued in each cycle of simulation and
N
is the number of peers present
in the network, query rate
M
is the mean of the
Poisson
process. The expres-
λ
=
N
e
−
λ
λ
K
sion
gives the probability that a peer issues
K
queries in a
p
(#
quries
=
K
)
=
K
!
cycle. The probability that a peer issues query for the file
f
c,r
depends on the peer's
interest level in category
c
and rank
r
of the file within that category.
(4)
Trust management engine
: The trust management engine helps a peer to com-
pute trust rating of other peer from past transaction history as well as recommendation
from its neighbor. It allows a peer to join the network with default trust level and
gradually build its reputation by providing good files to other peers. In the proposed
scheme, each peer maintains a
least recently used
(LRU) data structure to keep track
of recent transactions with almost 32 peers at a time. Each time peer
i
downloads a
file from peer
j
, it rates the transaction as positive (
tr
ij
=1) or negative (
tr
ij
= -1) de-
pending on whether downloaded file is authentic or fake.
1
∑
is the fraction
S
=
tr
ij
ij
TD
of successful downloads peer
i
had made from peer
j
, where
TD
is the total number of
downloads. Peer
i
considers peer
j
as trustworthy if
S
>
0
.
5
, and malicious if
.
S
<
0
ij
ij
If
≤
ij
S
, peer
i
considers peer
j
as
average trustworthy
. Peer
i
seeks recom-
mendations from other peers about peer
j
if the information is not locally available.
0
<
0
.
5
3.2 The Proposed Trust-Aware Algorithm
The proposed scheme uses three steps for its operation: (i) search, (ii) trust checking
and (iii) topology adaptation. These steps are discussed in the rest of this section.
3.2.1 Search
The
time to live
(TTL) bound search is used which evolves along with topology. At
each hop, query is forwarded to a fraction of neighbors, the number of neighbors is
decided based on the local estimate of network connectivity, known as probability of
community formation,
Prob
com
,. For node
x
,
Prob
com
is defined as in (1):
degree
(
x
)
−
initial
_
degree
(
x
)
(1)
Prob
com
(
x
)
=
initial
_
degree
(
x
).(
edge_limit
−
1
When
Prob
com
for a node is low, the peer has the capacity to accept new community
edges. Higher the value of
Prob
com
, lesser the neighbors choose to disseminate queries.
As simulation proceeds, connectivity of good nodes increases and reaches a maximum
value. So, they focus on directing queries to appropriate community which may host
Search WWH ::
Custom Search