Information Technology Reference
In-Depth Information
to frequent itemsets. However, the numbers of events that are sent to RPs related to
frequent itemsets are less compared with the numbers of events that are sent to the
other category of RPs. As a result, the total matching times on RPs are even.
3. It reduces the event publication cost. DMPSS groups subscriptions that do not
contain frequent itemsets onto a certain number of RPs. As a result, the numbers of
events that are sent to RPs for matching are reduced, and the overheads of nodes for
message transmission are reduced correspondingly.
The remainder of this paper is structured as follows. Section 2 introduces the
background and key technologies that are used in DMPSS. Section 3 presents the
proposed DMPSS scheme. In Section 4, we evaluate the performance of DMPSS by
simulations. And Section 5 offers some concluding remarks.
2
Background
2.1
System Definition
Similar to most of the previous publish/subscribe systems, DMPSS is defined as
S
=
{
A
1
,
A
2
,...,
A
n
}
,where
A
i
represents an attribute. Each attribute is described
by a tuple
[
name
:
type
,
min
,
max
]
. A subscription is a conjunction of predicates
over the attribute values, i.e.,
sub
p
m
,where
p
i
specifies a constant
value or a range for an attribute using common operators (
=
p
1
∧···∧
,etc.).
For example, a subscription expressed on the attribute
A
1
and
A
2
may be of the form
sub
=
,>,≥,<,≤,
=
=(
a
<
A
1
<
b
)
∧
(
c
<
A
2
<
d
)
. An event is a set of equalities on attributes, e.g.,
e
. In this paper, if the value of an attribute is not set in an event,
we suppose it is NULL.
In this paper, the attribute set that appears in a subscription
sub
is defined as
Att r
sub
, and that appears in an event
e
is defined as
Att r
e
. For example, when
sub
1
=(
=
{...,
A
i
=
v
i
,...}
a
<
A
1
<
b
)
∧
(
c
<
A
2
<
d
)
and
e
1
=
{
A
1
=
v
1
,
A
2
=
v
2
,
A
3
=
v
3
}
,
Att r
sub
1
=
{
. For ease of description, a subscription
sub
or an
event
e
contains attribute set
att r
means
att r
A
1
,
A
2
}
and
Att r
e
1
=
{
A
1
,
A
2
,
A
3
}
Att r
e
.Anevent
e
matches a subscription
sub
if and only if each predicate of
sub
is satisfied by the
value of corresponding attribute contained in event
e
.Thatistosay,ifanevent
e
matches a subscription
sub
,
Att r
sub
⊆
⊆
Att r
sub
or
att r
⊆
Att r
e
.
2.2
DHT-Based Publish/Subscribe System
The DHT overlay network is a class of decentralized systems in the application
level that partition ownership of a set of objects among participating nodes, and can
efficiently route message to the unique owner of any given object [1]. The classic
DHT overlay networks include Chord [11], CAN [12], Tapestry [13], Pastry [14],
etc. Each object or node in a DHT overlay network is assigned an ID (i.e. key)
according to a consistent hash function. An object is stored in a node whose ID
closest or immediately succeeds to the object's ID.
Search WWH ::
Custom Search