what-when-how
In Depth Tutorials and Information
≥
ʹ
⎧
⎪
⎪
v
inuite
node i will accept node
j s invitation
i
/
=
tp m
<
v
node i will refu
ʹ
ionship with node j
se node j s invitation or
inuite
(11.18)
i j
,
i
remove the relat
tP
ij
is node
i
's total payoff in the last
m
transactions with node
j
.
v
i
invite
is the invita-
tion threshold of node
i
and
coop
v
=
v
invite
(11.19)
i
i
where
v
i
coop
is the cooperation level of node
i
and can be a constant or variable value,
depending on the requirements of the system.
v
i
accept
is the acceptance threshold of
node
i
and
accept
=
⋅α
v
v
i
invite
(11.20)
i
i
where
α
i
represents the risk node
i
will take with respect to accepting an invitation
that is less cooperative than itself [23].
Reference 23 assumes that when a relationship is established from node
i
to
node
j
, node
j
will also connect to all the neighbors of node
i
. However, when node
j
cuts of the link with node
i
due to reduction of the payoffs, it will not delete
the relationships with the neighbors of node
i
. he simulation of the algorithm
shows that the total payoffs increase as the cooperation level is higher. Based on the
cooperation we can also introduce game theory and evolutionary dynamics in the
process of trust modeling [29].
11.4.8 Subjective Logic-Based Trust Model
Another trust model which also addresses the problem in many other models that
require the explicit and frequent feedbacks or ratings from the nodes to compute
the trustworthiness or reputation is presented in Reference 25. According to the
subjective aspects of trust, in Reference 25 it establishes the trust model by subjec-
tive logic, which includes various logical operators for the opinion combination
such as AND, OR, and NOT. In Reference 25, some nontraditional operators as
recommendation and consensus are emphasized.
Usually one opinion (O) consists of three parts [25]: belief (b), disbelief (d), and
uncertainty (u).
=
{ ,
O b d u
, }
(11.21)
+ + =
∈
[ , ]
(11.22)
b
d
u
1
, { ,
b d u
, }
0 1
