Information Technology Reference
In-Depth Information
Table 1. Simulation parameters
# of jobs (
N
)
10000
# of workers (
W
)
100
acceptable error rate (
acc
)
0.01 / 0.001
redundancy (
M
)
1
∼
10
sabotage model
predefined colluding
faulty fraction (
f
)
0
∼ f
max
(0.35)
sabotage rate (
s
)
0
∼
1
colluding rate (
c
)
0
∼
1
defection model
random defection
defection rate (
p
d
)
0
∼
0.8
spot-check rate (
q
)
0.1 / 0.2
saboteur knows when it is caught by spot-checking in a system without blacklisting [23,
40]. After having been caught, the saboteur immediately rejoins to the system and returns
incorrect results as a new worker.
3.2.2.
M
-first Voting vs.
M
-Majority Voting
First, We provide comprehensive comparison study of two basic voting methods,
M
-first
voting and
M
-majority voting, to reveal their sabotage tolerance performance and draw-
backs.
0.02
500
M=1
M=2,Majority
M=3,Major
i
t
y
M=2,M-
f
irst
M
=
3,M-first
M=1
M=2,Majority
M=3,Majority
M=2,M-first
M=3,M-first
ε
acc
450
0.015
400
350
0.01
300
250
200
0.005
150
100
0
0
0.2
0.4
0.6
0.8
1
0
0.2
0.4
0.6
0.8
1
s
s
(b) Computation time
T
(a) Error-rate
Figure 5.
M
-first voting vs.
M
-majority voting for sabotage rate
s
(
acc
=0.01
,
f =0.35
,
c =0.1
,
p
d
=0
, random scheduling).
Sabotage rate
s
and fraction
f
Fig.5(a) and Fig.6(a) show error rate of each voting
method for sabotage rate
s
and fraction
f
, respectively.
These figures show that larger