However, this feature was not considered in either the APCV or D-Scan method which resulted in the failure to reduce the handover delay in the low average range in response to increases in the number of MNs.
Alternatively, Figure 10(c) presents the average of obtained AP load of AP1 utilizing AHP scheme and APCV and D-Scan methods measured in bits/sec during simulation time.
In percentage form, the AP1 load as presented in Figure 10(b) indicates that the achieved load is the lowest using the AHP scheme followed by D-Scan and APCV methods, respectively.
Figure 7 illustrates the total number of handover decisions triggered by each of the five MNs (successful and failure handovers) employing AHP, APCV, and D-Scan.
It is obvious that proposed AHP scheme performs better than both APCV and D-Scan methods in terms of reducing the total number of handovers.
In contrast, the number of failed handovers in each of 5 MNs using both APCV and D-Scan was 3, 1, 0, 1, and 1 and 3, 2, 1, 2, and 3, respectively.
On the other hand, in APCV method, the MN obtained the handover decisions with APs based on the candidacy value obtained via fuzzy logic regardless of AP's current load factor and its related direction aspect.
The simulation outcome of varying number of failed handovers using proposed AHP scheme in comparison to D-Scan and APCV methods was calculated to demonstrate the probability of failure.
To sum up, the calculated failure probability in MN5 was zero by using AHP scheme, while it were 0.14 with APCV, and 0.6 with Scan method.
Figure 10(d) shows the average MAC-layer delay (measured in seconds), in comparison form between the proposed AHP scheme, D-Scan, and APCV during simulation time.
In contrast, APCV method obtained higher delay compared with AHP scheme, since APCV does not consider any adaptation process or weight vectors in order to improve handover decision making in fuzzy inference systems.