Information Technology Reference
In-Depth Information
10) if
k < , go to step 5, we start next generation in
the chemotactic loop
N
re
11) Elimination-dispersal: for
= with
probability e p , eliminate and disperse each bacterium
(this keeps the number of bacteria in the population
constant). When eliminate a bacterium, we disperse it
to a random location on the optimization domain.
i
1
2
,...,
S
l
<
N
12) if
, then go to step 4; otherwise end.
ed
4
Experimental Evaluations
The proposed BFO-based algorithm for workflow scheduling was implemented in
java programming language on an i3 3.07GHz, 4GB RAM machine running under
windows 7. To measure the performance of proposed BFO-based algorithm, we
compare our algorithm with ant colony algorithm [14] and partial swarm optimization
[15].
The parameter settings for the algorithms are showed in Table 1.
Table 1. Parameter settings for the algorithm (ACO, PSO, BFO)
Algorithm
Parameter name
Parameter Value
ACO
Maximum loop number
50
Ant number
50
History coefficient
1.2
Heuristic coefficient
1
Decay factor
0.1
PSO
Swarm size
50
Self-recognition coefficient
2
Social coefficient
2
Inertial weight
0.9
BFO
Population size
50
Elimination-dispersal steps
2
Reproduction steps
4
Chemotaxis steps
70
Maximum swim steps
4
Step size
0.1
Elimination-dispersal probability
0.25
Attraction depth
0.1
Attraction width
0.2
Repellant depth
0.1
Repellant width
10
Search WWH ::




Custom Search