Information Technology Reference
In-Depth Information
gives up in its absence. Due to its complexity (Michlmayr, 2007) solutions proposed to SQRP
typically limit to special cases.
Hyper-Heuristic_AdaNAS (HH_AdaNAS) is an adaptive metaheuristic algorithm, which
resolves SQRP (Hernandez, 2010). This algorithm was created from AdaNAS (Gómez et al.,
2010). The
high-level algorithm
is formed by HH_AdaNAS, which use as solution algorithm
AdaNAS, that is inspired by an ant colony and the set of
low-level heuristics
are included in
the algorithm called HH_TTL. The goal of hyperheuristic HH_TTL is to define by itself in
real time, the most adequate values for time to live (TTL) parameter during the execution of
the algorithm. The main difference between AdaNAS and HH_AdaNAS are:
when applying the modification of the TTL and the calculation of the amount of TTL to be
allocated. In the Figure 8 we show HH_AdaNAS is form by AdaNAS + HH_TTL.
SURVIVAL RULE
SURVIVAL RULE
Tables
Learning:
pheromone table
τ and tables
D
,
N
y
H.
Tables
Learning:
pheromone table
τ and tables
D
,
N
y
H.
+
+
SQRP
SQRP
SQRP
AdaNAS
AdaNAS
REPOSITORY
REPOSITORY
REPOSITORY
E
v
a
l
u
a
t
i
o
n
F
u
n
c
t
i
o
n
E
v
a
l
u
a
t
i
o
n
F
u
n
c
t
i
o
n
Tables Learning:
pheromone table τ
hh
and
tables η
.
Tables Learning:
pheromone table τ
hh
and
tables η
.
Tables Learning:
pheromone table τ
hh
and
tables η
.
Tables Learning:
pheromone table τ
hh
and
tables η
.
NODE
NODE
NODE
D
o
m
a
i
n
B
a
rr
i
e
r
D
o
m
a
i
n
B
a
rr
i
e
r
Hyper-Heuristic
HH_AdaNAS
Hyper-Heuristic
HH_AdaNAS
Hyper-Heuristic
HH_AdaNAS
Hyper-Heuristic
HH_AdaNAS
P2P-NETWORK
P2P-NETWORK
P2P-NETWORK
HH_TTL
HH_TTL
HH_TTL
HH_TTL
H1
H1
H1
H1
H2
H2
H2
H2
H3
H3
H3
H3
H4
H4
H4
H4
Z
Z
x
x
H1
: To increase TTL in 1 unit.
H2: Maintains TTL
constant
.
H3 = Falls TTL in 1 unit.
H4 = Falls TTL in 2 units.
H1
: To increase TTL in 1 unit.
H2: Maintains TTL
constant
.
H3 = Falls TTL in 1 unit.
H4 = Falls TTL in 2 units.
H1
: To increase TTL in 1 unit.
H2: Maintains TTL
constant
.
H3 = Falls TTL in 1 unit.
H4 = Falls TTL in 2 units.
H1
: To increase TTL in 1 unit.
H2: Maintains TTL
constant
.
H3 = Falls TTL in 1 unit.
H4 = Falls TTL in 2 units.
TOPOLOGY
TOPOLOGY
TOPOLOGY
Fig. 8. HH_AdaNAS is form by AdaNAS + HH_TTL.
Data structures of HH_AdaNAS
HH_AdaNAS inherited some data structures of AdaNAS, as the pheromone table τ and the
tables
H
,
D
and
N
. Besides the data structures of the high level metaheuristics, are the
structures that help to select the low-level heuristic these are the pheromone table τ
hh
and
the table hiperheuristic visibility states η. All the tables stored heuristic information or
gained experience in the past. The relationship of these structures is shown in Figure 9.
When HH_AdaNAS searches for the next node, in the routing process of the query, is based
on the pheromone table τ and tables
D
,
N
y
H
; these tables are intended to give information
on distant
D
,
H
is a table that records the successes of past queries and number of
documents
N
which are the closest nodes that can satisfy the query. In the same way, when
HH_TTL chooses the following low level heuristic, through data structures τ
hh
and η. The
memory is composed of two data structures that store information of prior consultations.
The first of these memories is the pheromone table τ
hh
which has three dimensions, and the
other memory structure is the table hiper-heuristic visibility states η, which allows the hiper-
