Page 41 - 2024-Vol20-Issue2
P. 41
37 | Jumaa, Mohamad, Allawy & Mohammed
GTFT node’s behaviour optimization experiments (shown in 2018, the simulation of the ad-hoc network was performed.
Table I) have four control parameters (which represents the The information regarding the parameters of ad-hoc networks
four factors that are mentioned above), where each parameter was gathered using MATLAB and subsequently imported into
has three levels (values), so, L9 orthogonal array is selected. the Minitab DOE in preparation for Taguchi analysis.
The parameters are:
VII. GTFT NODE BEHAVIOUR RESULTS
• F1: Power constraints per node (P). Where (L1= 1,
L2=0.5, L3=0.2). Impact factor levels are taken as initial parameters and en-
tered GTFT according to the OA-designed experiments of the
• F2: Number of packets required to forward (N). Where Taguchi method which are used to evaluate node behaviour
(L1=1000, L2=500, L3=200). performance. Fig. 1 to 9 show the performance of the node.
Based on Taguchi optimization, statistics show that the op-
• F3: power consumption per node (C). Where (L1=0.001, timal case for a node to behave as a relay is in the third
L2=0.0008, L3=0.0003). experiment where the values of ? and ? were one due to the
availability of ideal factors that are: the power of the node at
• F4: Grade of generosity GoG (e). Where (L1=0.08, the highest level, that is, the battery node is fully charged, the
L2=0.05, L3=0.02). number of packets at its lowest level where it is equal to 200,
and the packets that are required to be redirected by the node
When the OA has designated the iterations or trials, experi- consume energy at the lowest level (0.0003 watts), that it is
ments can be done. The trial number is concluded based on mean, the two hundred packets consume 6% of the node’s
the application budget and the complication of experiments. battery power.
For the GTFT node’s behaviour experiments, it may contain 3 The throughput was as high as possible, and it was 200 out
trials, and at finally, each trial has three outcome values which of 200 packets, which means that all packets were redirected.
are: The influencing factor mainly is the number of packets re-
quired to be redirected, accordingly, the level of generosity
• T1: is the measured Throughput. does not effectively affect the behaviour, of the node because
in the third experiment, the level of generosity was at the
• T2: is the measured ?. lowest level, which is 0.02. While most selfish behaviour of
the node was in experiment 8 where the throughput equal to
• T3: is the measured ? . 41 out of 500 packets due to the non-ideal conditions of the
node:
Then, SNR is calculated using (2), (3), (4), (5), and (6). All
of experiments’ calculations are shown in Table II. After • The power of the node is at the lowest level. It is equal
that, the average SNR value per parameter is calculated in to 0.2, meaning that the battery is almost empty and at
accordance with given levels. The calculations for the ? (F1), the level of 20% of the remaining charge, which drives
N (F2), c (F3), and e (F3) are shown in Table III. From the it to selfish behaviour to maintain the level of its ability
DOE of GTFT node behavior, notices that Taguchi Method to stay alive (power on).
is perfect for identifying the best impact factor on the node
behavior from the four given factors. Table IV concluded the • The number of packets was at a high level, equal to 500.
results. Since this article’s objective is to present the problem
of improving GTFT node behaviour as a series of experiment The packets that are required to be forwarded by the node con-
design tests, Taguchi’s method of designing experiments has sume power at the highest level of 0.001 watts, meaning that
been used. The strength of using the Taguchi method in this the 500 packets consume 50% of the energy of the full node
article is measuring design quality using the signal-to-noise battery, in other words, that the node battery is not sufficient
ratio (SNR) and the orthogonal array principle that is used to to redirect the packets. The selfish behaviour of experiment 8
study the influence of several design parameters: number of was at the same level of generosity as that of the ideal cooper-
packets (N), power consumption (c), power constraint of the ative behaviour of experiment 3, equal to 0.02.
node (?), and GoG, respectively. The Taguchi method is a Fig. 1 shows the behaviour of a node for experiment number
robust design approach that uses many insights from statistical (1), and from our observation of the shape, when ordering
experimental design to evaluate and implement process im- more than 200 packets, the ? of the node begins to decrease
provements. The basic principle is to improve product quality dramatically to reach a value of zero when requesting more
by reducing the impact of the causes of the difference without than 1000 packets. From Fig. 2 and at experiment number
eliminating the causes. (2), we notice that the node behaves roughly like experiment
It should be mentioned that the Taguchi DOE was generated
for the L9 OA using Minitab version 20. Using MATLAB