Date Published: April 17, 2019
Publisher: Public Library of Science
Author(s): Khalid M. Hosny, Marwa M. Khashaba, Walid I. Khedr, Fathy A. Amer, He Debiao.
Providing complete mobility along with minimizing the poor quality of service (QoS) is one of the highest essential challenges in mobile wireless networks. Handover prediction can overcome these challenges. In this paper, two novel prediction schemes are proposed. The first, depends on scanning the quality of all signals among mobile station and all nearby stations in the surrounding area, while the second one is based on a multi-criteria prediction decision using both the signal-to-noise ratio SNR value and station’s bandwidth. Moreover, the prediction efficiency is improved by reducing the number of redundant/ unnecessary handovers. The proposed schemes are evaluated using different scenarios with several mobile stations’ numbers, different WLAN access points, LTE-base station number & location, and random mobile station movement manner. The proposed schemes achieved a success rate of 99% with the different scenarios using LTE-WLAN architecture. The performance of the proposed prediction schemes outperformed the performance of the existing prediction schemes in terms of the accuracy percentage.
In the mobile environment, all users simultaneously should get full mobility while preserving the QoS where users ‘mobility significantly affects the QoS. Mobile Station (MS) must update the Base Station (BS) they are to be connected during its user movement. The execution of this process is called “handover”. MS always need to scan its surrounding users and to observe all nearby stations’ parameters such as the quality of the signal or the delay in the packet transfer to be able to execute the handover process accurately. The MS should execute the handover process to keep a continuous data transfer. Therefore, it always compares the signal parameters of the nearby serving stations with a predefined level. If the parameters of the serving station’ signal descend under these predefined values, the handover process begins to guarantee the required QoS .
In this section, all details of the two proposed prediction schemes are presented. The first proposed prediction scheme is called “RSSI-based handover prediction scheme”. The second prediction scheme is called “Multi-criteria handover prediction scheme”.
The area for simulation is divided into four sub-areas as displayed in Fig 7. In each sub-area, different APs and different MSs with various speeds and locations are used in our simulation. We utilized the random waypoint mobility model (RWPMM)  as a movement manner of all MS. By comparing the PRWMM with other mobility models, it is observed that the PRWMM provide the highest level of movement’ randomness which make the simulation very realistic. The average speed of a normal human is 5 km/hour  which equivalent to 1.38 m/s. In the simulation we used a MS speed in a random interval from 0.5 m/s to 3 m/s. The suggested lower limit, 0.5 m/s, is suitable for elderly people while the suggested upper limit is suitable for people with swift movements. The proposed handover prediction schemes are evaluated by using the MATLAB16 within different areas. The positions of all MSs and APs are produced in a random manner.
To ensure the efficiency of the proposed schemes, we compared the performance of the proposed handover schemes with the existing schemes [20, 26, 28, 31–33] in terms of movement manner. The proposed handover prediction schemes, HOP, which utilizes two thresholds, outperformed all existing scheme. The performance of the proposed handover schemes is higher than the performance of the Bellavista et al.  by 30%. Becvar  achieved 45% for three neighboring stations in Manhattan-like road utilization. Becvar et al.  used two independent thresholds. Although they use the same concept, there are two main differences between our proposed schemes and the scheme in . First, the TTP scheme which introduced by Becvar and his co-authors performs handover prediction only in WiMAX networks (Horizontal Handover) while the proposed HOP scheme designed to work with heterogeneous networks (Vertical Handover) as well as homogeneous networks. Second, both LTE and WiMAX networks have fixed locations and fixed certain areas for coverage which reduces the number of handovers. On the other side, the Wi-Fi AP is randomly located and their power are limited and variable which add more challenges to achieve successful handover.
Two novel vertical handover prediction schemes based on channel features are proposed in this paper. The proposed PHO schemes are used with both heterogeneous as well as homogenous networks. Two new thresholds are defined and used in the proposed schemes to minimize the number of redundant handovers. The first proposed scheme is based on monitoring the RSSI values of all nearby stations. It achieves a prediction success with percentage 99% in the predefined surrounding areas using random MS’ movement manner. The second proposed prediction scheme is based on multi-criteria prediction and decision process. It used the SNR, the available bandwidth, the MS’ data rate and the RSSI values. It achieves a success rate 99% in the predefined simulation area.