Seamless Handover Scheme for MEC/SDN-Based Vehicular Networks

With the recent advances in the fifth-generation cellular system (5G), enabling vehicular communications has become a demand. The vehicular ad hoc network (VANET) is a promising paradigm that enables the communication and interaction between vehicles and other surrounding devices, e.g., vehicle-to-vehicle (V2V) and vehicle-to-everything (V2X) communications. However, enabling such networks faces many challenges due to the mobility of vehicles. One of these challenges is the design of handover schemes that manage the communications at the intersection of coverage regions. To this end, this work considers developing a novel seamless and efficient handover scheme for V2X-based networks. The developed scheme manages the handover process while vehicles move between two neighboring roadside units (RSU). The developed mechanism is introduced for multilane bidirectional roads. The developed scheme is implemented by multiple-access edge computing (MEC) units connected to the RSUs to improve the implementation time and make the handover process faster. The considered MEC platform deploys an MEC controller that implements a control scheme of the software-defined networking (SDN) controller that manages the network. The SDN paradigm is introduced to make the handover process seamless; however, implementing such a controlling scheme by the introduction of an MEC controller achieves the process faster than going through the core network. The developed handover scheme was evaluated over the reliable platform of NS-3, and the results validated the developed scheme. The results obtained are presented and discussed.


Introduction
Motivated by the high-speed development of the mobile internet and the growing business demand, the fifth-generation cellular system (5G) is supposed to have a low cost and low power consumption and be safe and reliable [1]. Fifth-generation cellular systems support new demands and requirements to enable new use cases and novel services. The transmission rate is supposedly increased up to 100 times the value of the last release of the fourth-generation cellular system (4G), and the peak transmission rate is expected to reach 10 Gbit/s [2,3]. One of the main targets for 5G is to reduce the end-to-end delay to 10 ms for the first commercial releases and to 1 ms (5 ms while moving) to support ultrareliable enables the network operator to introduce new services and provide a new means of service innovation [27].
In this work, we use MEC as part of a vehicular network in which cars can communicate with smart traffic lights, surrounding objects, other vehicles on the road, and other infrastructure. The main contributions of this article can be summarized in the following points: 1.
Designing and developing a reliable system structure for a VANET over a 5G system, with the deployment of MEC technology, that supports V2X and V2V communications.

2.
Designing and developing a novel seamless, fast handover mechanism for the introduced VANET system structure. 3.
Implementing the developed handover algorithm over an MEC server. 4.
Conducting a performance evaluation of the proposed system and the developed handover mechanism.

Background and Related Works
Recently, the VANET has attracted many researchers in the academic and industrial fields [28,29]. The VANET is one of the main use cases of 5G networks that has been announced, and since the first commercial version of the 5G system was released, the research in the area of VANETs has become more attractive [30,31]. Many existing works in the literature consider VANETs; however, in this section, we consider only the works that are most relevant to our proposed study in order to highlight the novelty of our work.
The handover process is one of the main critical processes in vehicular communication systems, since it should be done in an efficient, seamless way [32]. This is crucial and presents a challenge for designing VANETs, due to the high mobility of vehicles [33,34]. There are few works that consider the handover process in VANETs or the development of a handover algorithm for VANETs. We considered the recently developed handover schemes for VANETs.
In [35], the authors introduced a novel network structure for a VANET based on MEC technology. The proposed structure achieved high system scalability and high latency efficiency. However, the handover mechanism considered was that of a traditional cellular system, essentially the LTE-V handover mechanism. The authors did not consider developing a new handover mechanism for the developed VANET structure. This would affect the overall performance and create conflict, since MEC servers were not considered in the handover process. In our proposed work, the MEC paradigm is deployed in the VANET structure and represents a main part of the suggested handover algorithm. This makes the handover process more seamless and achieves many other benefits, as introduced later, in Section 3.
In [36], the authors developed a novel network structure for vehicular communications based on SDN technology. The developed system met the ITS requirements and the specifications that had been announced. The challenge of the handover scheme was addressed by developing a fog-shaped cell that reduced the rate of the handover process among RSUs. The work considered the throughput and latency as the key performance indicators. The developed system achieved a higher latency and throughput than other existing systems.
In [37], the authors developed an SDN-based handover scheme for vehicular networks. SDN was used to maintain the transport layer communication during the handover process. The VANET deployed the IEEE 802.11p and enabled V2X communications over an unlicensed spectrum. The simulation results validated the system in terms of throughput. However, our proposed system involves implementing the handover scheme over MEC servers, which achieves less latency. Moreover, the time required to migrate V2X tasks is considered, which reduces the probability of task blocking during the handover process and achieves a higher system availability.
In [38], scientists employed mobile edge computing (MEC) and software-defined network (SDN) technologies in the building of the network to modify the overall net-work performance under the conditions of high traffic density. They also developed an algorithm of D2D-based clustering to solve the problem of disconnected vehicles in dead zones and offload the network in regions with high traffic. By using the aforementioned technologies, the authors could achieve a performance of up to 74% gains in terms of task-blocking probability.
In [22], the authors proposed an advanced handover algorithm by adopting SDN in the building of the network architecture, and they explained the advantages of using SDN in the network control to administer the data transfer in VANETs. Hence, the connection of the transmission layers could be fixed when the handover occurred. This achieved a high-quality improvement based on the vehicle's mobility and link rates.
In [39], the authors focused their studies on the latest developments in data-offloading techniques in VANETs.
Specifically, they investigated the communication models between vehicles and infrastructure and classified the technique into several categories: data offloading through vehicle-to-vehicle communication, data offloading through vehicle-to-everything communication, and data offloading through vehicle-toinfrastructure communications.

MEC/SDN Vehicular Network Structure
In this section, the considered structure of the VANET is introduced. Figure 1 presents the developed MEC/SDN vehicular network that runs over the 5G system. The system enables V2V and V2X communications, as vehicles can communicate with each other and with distributed roadside units (RSUs). Each vehicle deploys an onboard unit (OBU) whose structure is presented in Figure 2. This embedded unit facilitates the implementation and execution of the required communication standards and interfaces. J. Sens. Actuator Netw. 2022, 11, x FOR PEER REVIEW 4 of 16 V2X tasks is considered, which reduces the probability of task blocking during the handover process and achieves a higher system availability.
In [38], scientists employed mobile edge computing (MEC) and software-defined network (SDN) technologies in the building of the network to modify the overall network performance under the conditions of high traffic density. They also developed an algorithm of D2D-based clustering to solve the problem of disconnected vehicles in dead zones and offload the network in regions with high traffic. By using the aforementioned technologies, the authors could achieve a performance of up to 74% gains in terms of taskblocking probability.
In [22], the authors proposed an advanced handover algorithm by adopting SDN in the building of the network architecture, and they explained the advantages of using SDN in the network control to administer the data transfer in VANETs. Hence, the connection of the transmission layers could be fixed when the handover occurred. This achieved a high-quality improvement based on the vehicle's mobility and link rates.
In [39], the authors focused their studies on the latest developments in data-offloading techniques in VANETs. Specifically, they investigated the communication models between vehicles and infrastructure and classified the technique into several categories: data offloading through vehicle-to-vehicle communication, data offloading through vehicle-toeverything communication, and data offloading through vehicle-to-infrastructure communications.

MEC/SDN Vehicular Network Structure
In this section, the considered structure of the VANET is introduced. Figure 1 presents the developed MEC/SDN vehicular network that runs over the 5G system. The system enables V2V and V2X communications, as vehicles can communicate with each other and with distributed roadside units (RSUs). Each vehicle deploys an onboard unit (OBU) whose structure is presented in Figure 2. This embedded unit facilitates the implementation and execution of the required communication standards and interfaces.   The developed structure deploys distributed edge computing with two heterogeneous levels. The first level represents the multiple-access edge computing units connected to the RSUs, which are referred to as the vehicular edge computing units (V-MECs). The second level represents the distributed MEC units connected to the cellular base stations, which are referred to as the cellular MECs (M-MECs). Both levels of these distributed edge units are introduced to provide computing resources near to vehicles to achieve higher latency efficiency, higher reliability, and higher system availability. Moreover, the introduction of MEC units facilitates the implementation of the developed networking algorithms. In particular, the V-MEC implements our developed handover scheme, resulting in a higher execution efficiency than that of the traditional systems.
The introduction of distributed edge units as presented in Figure 1 facilitates the achievement of the required quality of service (QoS) of V2X applications. The communication between the OBU and the V-MEC reduces the communication latency, increases the throughput, achieves a higher availability and reliability, and increases the spectral efficiency.
The network deploys SDN technology for the control and management scheme. The considered core network deploys a control plane with multiple SDN controllers. The dataplane of the SDN deploys a distributed scheme of OpenFlow switches that receive their forwarding tables and their updates from the control scheme over the OpenFlow interface. The introduction of SDN technology is intended to provide the required network flexibility and an efficient management scheme for the network. For the developed handover scheme, the introduction of SDN achieves a seamless process, since the TCP connection is maintained.

Novel Seamless Handover Scheme for MEC/SDN Vehicular Networks
When a vehicle comes to the intersection between two coverage regions of neighboring RSUs, the vehicle should perform a seamless, fast handover process. Setting up the handover process and selecting the optimum time to turn from one coverage region of an RSU to the other coverage region is a challenge, due to vehicles' mobility. This section presents the developed seamless, fast handover scheme for the considered vehicular network structure introduced in the previous section.
The quality of the communication process depends on the received signal power, i.e., the received signal-to-interference noise ratio (SINRRx), which should not be decreased lower than a threshold level. The main steps of the developed handover scheme are presented in Figure 3; Figure 4 shows the signaling flow of the developed handover scheme. The first step is the initialization phase, when the vehicle triggers the handover according to the gathered information. These data contain parameters gathered by the different com- The developed structure deploys distributed edge computing with two heterogeneous levels. The first level represents the multiple-access edge computing units connected to the RSUs, which are referred to as the vehicular edge computing units (V-MECs). The second level represents the distributed MEC units connected to the cellular base stations, which are referred to as the cellular MECs (M-MECs). Both levels of these distributed edge units are introduced to provide computing resources near to vehicles to achieve higher latency efficiency, higher reliability, and higher system availability. Moreover, the introduction of MEC units facilitates the implementation of the developed networking algorithms. In particular, the V-MEC implements our developed handover scheme, resulting in a higher execution efficiency than that of the traditional systems.
The introduction of distributed edge units as presented in Figure 1 facilitates the achievement of the required quality of service (QoS) of V2X applications. The communication between the OBU and the V-MEC reduces the communication latency, increases the throughput, achieves a higher availability and reliability, and increases the spectral efficiency.
The network deploys SDN technology for the control and management scheme. The considered core network deploys a control plane with multiple SDN controllers. The dataplane of the SDN deploys a distributed scheme of OpenFlow switches that receive their forwarding tables and their updates from the control scheme over the OpenFlow interface. The introduction of SDN technology is intended to provide the required network flexibility and an efficient management scheme for the network. For the developed handover scheme, the introduction of SDN achieves a seamless process, since the TCP connection is maintained.

Novel Seamless Handover Scheme for MEC/SDN Vehicular Networks
When a vehicle comes to the intersection between two coverage regions of neighboring RSUs, the vehicle should perform a seamless, fast handover process. Setting up the handover process and selecting the optimum time to turn from one coverage region of an RSU to the other coverage region is a challenge, due to vehicles' mobility. This section presents the developed seamless, fast handover scheme for the considered vehicular network structure introduced in the previous section.
The quality of the communication process depends on the received signal power, i.e., the received signal-to-interference noise ratio (SINR Rx ), which should not be decreased lower than a threshold level. The main steps of the developed handover scheme are presented in Figure 3; Figure 4 shows the signaling flow of the developed handover scheme. The first step is the initialization phase, when the vehicle triggers the handover according to the gathered information. These data contain parameters gathered by the different communication layers, e.g., the transport and network layers, such as the signaling parameters, the spectrum parameters, the mobility, and the throughput. Once the received signal strength decreases below an initial threshold level (SINR Rx-th-1 ), the vehicle turns from the steady state to the inspection state. In the inspection state, the vehicle calculates the expected time to reach the next threshold level of received power (SINR Rx-th-2 ), at which the handover process should start.
vehicle through the targeted base station at the estimated time. When reaching the threshold level SINRRx-th-2, the connection with the targeted MEC server should be set up, and the third phase of the handover mechanism, i.e., the execution state, starts.
In the execution state, the handover process should be performed, and the vehicle should establish a connection to the targeted coverage region, while the connection to the currently serving coverage region should be disabled. This process should be carried out seamlessly for the end-user, with no discontinuity of the communication process. This requires migrating the current tasks of the vehicle from the current MEC server to the targeted MEC server. This takes a length of time that should be calculated in the second phase, i.e., the inspection state, as the threshold power SINRRx-th-2 is allocated according to this length of time. During the migration process, the vehicle starts sending information to the targeted RSU. After the process ends, the connection with the old RSU is ended through the MEC server. This should be done before the received power reaches the third threshold level (SINRRx-th-3), which represents the critical received signal, after which the packet reception will be affected.  In the inspection state, the OBU communicates the MEC server of the serving base station to set up a connection with the MEC server of the targeted RSU. The serving MEC server receives information on the vehicle's current mobility and position. The MEC server calculates the expected time to perform the handover process according to the measured data. It then sends a handover request to the targeted MEC server with the expected time. The targeted MEC server starts to set up a connection with the relevant vehicle through the targeted base station at the estimated time. When reaching the threshold level SINR Rx-th-2 , the connection with the targeted MEC server should be set up, and the third phase of the handover mechanism, i.e., the execution state, starts.
In the execution state, the handover process should be performed, and the vehicle should establish a connection to the targeted coverage region, while the connection to the currently serving coverage region should be disabled. This process should be carried out seamlessly for the end-user, with no discontinuity of the communication process. This requires migrating the current tasks of the vehicle from the current MEC server to the targeted MEC server. This takes a length of time that should be calculated in the second phase, i.e., the inspection state, as the threshold power SINR Rx-th-2 is allocated according to this length of time. During the migration process, the vehicle starts sending information to the targeted RSU. After the process ends, the connection with the old RSU is ended through the MEC server. This should be done before the received power reaches the third threshold level (SINR Rx-th-3 ), which represents the critical received signal, after which the packet reception will be affected.
To perform a seamless handover process, we introduce SDN. The SDN controller maintains the connection at the transport layer unaltered, thus avoiding the limitations of the congestion window (cwnd). To perform a faster migration process and am efficient handover process, we introduce the facilities of SDN to the distributed edge computing servers. This can be achieved by our previously developed MEC platform, which deploys an MEC controller with an interface to the SDN controller and implements a lightweight version of the control scheme. To perform a seamless handover process, we introduce SDN. The SDN controlle maintains the connection at the transport layer unaltered, thus avoiding the limitations o the congestion window (cwnd). To perform a faster migration process and am efficien handover process, we introduce the facilities of SDN to the distributed edge computin servers. This can be achieved by our previously developed MEC platform, which deploy an MEC controller with an interface to the SDN controller and implements a lightweigh version of the control scheme. Figure 4 presents the three considered threshold levels deployed for the develope fast handover mechanism. The first threshold level, L1, represents the starting point of th developed handover mechanism, i.e., the start of the inspection state. The second thresh old level, L2, is variable, and it represents the end of the inspection state and the start o the execution phase of the handover process. The handover process must be ended befor the third level, L3, at which the packet reception ratio from the current serving base statio decreases below the required level. Thus, the handover process should be ended befor this point to maintain communication with the required QoS.

Estimating Threshold Levels
The main factor in deciding the handover process is the value of the received signa that results in an acceptable packet reception ratio (PRR). Once the value of the receive packet ratio decreases below the critical PRR, the communication is interrupted and th quality of service (QoS) is decreased below the required level. Thus, the handover proces must be performed before the received packet ratio decreases below the critical PRR. Th probability of successful packet reception, Pr,sr, is calculated by Equation (1).  Figure 4 presents the three considered threshold levels deployed for the developed fast handover mechanism. The first threshold level, L 1 , represents the starting point of the developed handover mechanism, i.e., the start of the inspection state. The second threshold level, L 2 , is variable, and it represents the end of the inspection state and the start of the execution phase of the handover process. The handover process must be ended before the third level, L 3 , at which the packet reception ratio from the current serving base station decreases below the required level. Thus, the handover process should be ended before this point to maintain communication with the required QoS.

Estimating Threshold Levels
The main factor in deciding the handover process is the value of the received signal that results in an acceptable packet reception ratio (PRR). Once the value of the received packet ratio decreases below the critical PRR, the communication is interrupted and the quality of service (QoS) is decreased below the required level. Thus, the handover process must be performed before the received packet ratio decreases below the critical PRR. The probability of successful packet reception, P r,sr , is calculated by Equation (1).
where L is the communication losses, P Tx and P Rx are the total transmitted and received power, SINR Rx is the signal-to-noise ratio of the received signal, and SINR Rx-th-3 is the threshold of the received signal-to-noise ratio after which the packets cannot be received successfully. The communication losses can be calculated as the sum of the channel loss and the path loss (PL), which can be calculated as a function of the communication distance (d) using the street canyon model as introduced in [40]. The power spectral density of the considered 60 GHz spectrum is n 0 , and the Bwave is the allocated bandwidth of the wireless access of the vehicular environment (wave). Thus, to guarantee the successful reception of packets, the received signal-to-noise ratio must be higher than the threshold level, SINR Rx-th-3 . This condition is included in the developed handover mechanism, as the handover process should be performed before SINR Rx decreases below the threshold level of SINR Rx-th-3 . During the inspection state, the initial MEC server calculates the time required to migrate the current services of the vehicle, T m . Based on the received data about the vehicle's mobility, V i , and the vehicle's current position, the MEC server calculates the distance corresponding to the T m . This is the required distance a vehicle should travel during the migration process and is referred to as dm. If the distance between the first threshold level, L 1 , to the third threshold level, L 3 , is d, the threshold level L 2 is chosen to achieve d 2 > d m .

Calculating Packet Error Rate
One of the main performance indicators is the packet error rate (PER), which is an indicator for failed transmission. The PER represents the ratio of undelivered packets to the total transmitted packets. It is measured during a sliding window. Figure 5 indicates the methodology of evaluating the PER. The total interval during which the PER is calculated is divided into k windows (ω) of equal time intervals (τ). The total PER of a vehicle v i , PER T vi , is calculated by Equation (3).
where PER ωj vi is the PER of the jth window and can be calculated as the ratio of undelivered packets to the total transmitted packets during the window. PER is an important performance metric, since it indicates two main packet losses; the loss of packets missed during queuing due to the arrival of a newer basic safety message (BSM) and the packet loss in the channel due to collision or weak signal.
where L is the communication losses, PTx and PRx are the total transmitted and received power, SINRRx is the signal-to-noise ratio of the received signal, and SINRRx-th-3 is the threshold of the received signal-to-noise ratio after which the packets cannot be received successfully. The communication losses can be calculated as the sum of the channel loss and the path loss (PL), which can be calculated as a function of the communication distance (d) using the street canyon model as introduced in [40]. The power spectral density of the considered 60 GHz spectrum is n0, and the Bwave is the allocated bandwidth of the wireless access of the vehicular environment (wave). Thus, to guarantee the successful reception of packets, the received signal-to-noise ratio must be higher than the threshold level, SINRRx-th-3. This condition is included in the developed handover mechanism, as the handover process should be performed before SINRRx decreases below the threshold level of SINRRx-th-3. During the inspection state, the initial MEC server calculates the time required to migrate the current services of the vehicle, Tm. Based on the received data about the vehicle's mobility, Vi, and the vehicle's current position, the MEC server calculates the distance corresponding to the Tm. This is the required distance a vehicle should travel during the migration process and is referred to as dm. If the distance between the first threshold level, L1, to the third threshold level, L3, is d, the threshold level L2 is chosen to achieve d2 ˃ dm.

Calculating Packet Error Rate
One of the main performance indicators is the packet error rate (PER), which is an indicator for failed transmission. The PER represents the ratio of undelivered packets to the total transmitted packets. It is measured during a sliding window. Figure 5 indicates the methodology of evaluating the PER. The total interval during which the PER is calculated is divided into k windows (ω) of equal time intervals (τ). The total PER of a vehicle vi, PERT vi , is calculated by Equation (3).
where PERωj vi is the PER of the jth window and can be calculated as the ratio of undelivered packets to the total transmitted packets during the window. PER is an important performance metric, since it indicates two main packet losses; the loss of packets missed during queuing due to the arrival of a newer basic safety message (BSM) and the packet loss in the channel due to collision or weak signal.

Performance Evaluation
The developed handover scheme for an MEC/SDN VANET is evaluated in this section. Various simulation scenarios are considered, and the results are presented and discussed.

Simulation Setup
The system was evaluated over the NS-3 platform, with an open-source traffic simulation from the Simulation of Urban Mobility (SUMO) to provide vehicle mobility. The considered vehicles' mobility was that of the highway. The simulation process ran over a machine with an Intel Core i9 processor and 64GB RAM. All the required SDN modules for the SDN controller and OpenFlow switches were included with OpenFlow version 1.3.
The considered simulation topology consisted of the specifications presented in Table 1. A two-lane bidirectional road of 5Km length was considered for the simulation process. The road was served by five RSU units, with one MEC server connected to each RSU. The specifications of the RSUs and the connected MEC servers are presented in Table 1. The vehicles were distributed along the road according to the traffic density shown in Table 1. Three values for the total number of deployed vehicles, N, were considered to investigate the effect of the increase in traffic along the road on the performance of the developed handover scheme. Moreover, three different traffic densities were considered to investigate the effect of the change in the traffic density on the overall network performance. We considered our previously developed offloading scheme introduced in [41] for data offloading in the proposed MEC-based VANET. The offloading parameters are given in Table 1.
Each vehicle in the developed system was assigned a computing task corresponding to the real task's workload. For the performance evaluation, we considered three performance metrics: the packet delivery ratio (PDR), the percentage of blocked computing tasks, and the percentage of failed handover processes from the total number of required handover processes.

Simulation Results
Figures 6 and 7 present the recorded PDR for the developed MEC-based handover scheme and the traditional handover scheme without the MEC involvement. Figure 6 provides the PDR for both systems at different traffic densities, with the distance to the RSU. These results were for a VNET with 400 vehicles, i.e., N = 400, and the vehicles' speed was 50 Km/h. Up to 300 m away from the RSU, both handover schemes achieved almost the same PDR; however, after 300 m, the value of the PDR for both systems changed. This change increased after 400 m, since at this distance the handover process had started or was close to starting in the traditional handover scheme. Moreover, with the increase in the traffic density, the PDR decreased due to the higher probability of collisions.
scheme and the traditional handover scheme without the MEC involvement. Figure 6 provides the PDR for both systems at different traffic densities, with the distance to the RSU. These results were for a VNET with 400 vehicles, i.e., N = 400, and the vehicles' speed was 50 Km/h. Up to 300 m away from the RSU, both handover schemes achieved almost the same PDR; however, after 300 m, the value of the PDR for both systems changed. This change increased after 400 m, since at this distance the handover process had started or was close to starting in the traditional handover scheme. Moreover, with the increase in the traffic density, the PDR decreased due to the higher probability of collisions. Figure 7 provides the PDR for both systems, the proposed and the traditional handover scheme, at different vehicle velocities. These results were for a VNET with 400 vehicles, i.e., N = 400, and a traffic density of 0.1 veh/m. Five vehicle velocities were considered, i.e., those given in Table 1. With the increase in the vehicle velocity, the PDR decreased. Up to 50 Km/h the change in the PDR for both systems due to different velocities was small; however, it increased for the 60 and 70 km/h velocities. The developed MEC/SDNbased handover scheme achieved an improvement in the PDR compared to the traditional handover scheme by an average of 47%. Furthermore, the developed scheme achieved a greater improvement in the PDR for a high traffic density and high vehicle velocity compared to the traditional scheme. The percentage improvement in the PDR in the case of high vehicle velocity was 64%.   The second performance metric considered was the percentage of blocked tasks. Three systems were considered for the performance evaluation: the first system represented the traditional VANET with the traditional handover scheme; the second system represented the MEC-based VANET with a traditional handover scheme; and the third system represented the developed MEC/SDN-based VANET with the developed handover scheme. For the three systems, our previously developed offloading scheme, introduced in [41], was used. Figures 8-10 provide the three systems' percentage of blocked tasks in different scenarios. The statistical analysis of the results for each of the three systems is presented in Table 2. This includes the mean and the standard deviation (Std) of the results for each system. Figure 8 provides the average percentage of blocked tasks for each system in three scenarios with different numbers of deployed vehicles. The developed MEC/SDN-based system achieved a better performance in terms of blocked tasks compared with the other  Figure 7 provides the PDR for both systems, the proposed and the traditional handover scheme, at different vehicle velocities. These results were for a VNET with 400 vehicles, i.e., N = 400, and a traffic density of 0.1 veh/m. Five vehicle velocities were considered, i.e., those given in Table 1. With the increase in the vehicle velocity, the PDR decreased. Up to 50 Km/h the change in the PDR for both systems due to different velocities was small; however, it increased for the 60 and 70 km/h velocities. The developed MEC/SDNbased handover scheme achieved an improvement in the PDR compared to the traditional handover scheme by an average of 47%. Furthermore, the developed scheme achieved a greater improvement in the PDR for a high traffic density and high vehicle velocity compared to the traditional scheme. The percentage improvement in the PDR in the case of high vehicle velocity was 64%.
The second performance metric considered was the percentage of blocked tasks. Three systems were considered for the performance evaluation: the first system represented the traditional VANET with the traditional handover scheme; the second system represented the MEC-based VANET with a traditional handover scheme; and the third system represented the developed MEC/SDN-based VANET with the developed handover scheme. For the three systems, our previously developed offloading scheme, introduced in [41], was used. Figures 8-10 provide the three systems' percentage of blocked tasks in different scenarios. The statistical analysis of the results for each of the three systems is presented in Table 2. This includes the mean and the standard deviation (Std) of the results for each system.              The third performance metric was the percentage of failed handover processes compared to the total number of requested handover processes. Figures 11-13 provide the    Figure 8 provides the average percentage of blocked tasks for each system in three scenarios with different numbers of deployed vehicles. The developed MEC/SDN-based system achieved a better performance in terms of blocked tasks compared with the other two system. This performance improvement increased with the increase in the number of vehicles in the network. This was due to the introduction of the MEC units managed by SDN, which provided a path for data offloading. Moreover, the developed handover scheme reduced the probability of task blocking because of the consideration of the migration time of the computing tasks. Figure 9 provides the percentage of blocked tasks for the three systems at different vehicle velocities. These results were for a VANET with 400 vehicles, i.e., N = 400, and a traffic density of 0.1 veh/m. The percentage of blocked tasks increased in all three systems with the increase in the vehicle's velocity. However, the percentage increase in the number of blocked tasks for the developed scheme was the lowest. Figure 10 shows the percentage of blocked tasks for the three systems at different values of traffic density. The three systems were simulated for 400 vehicles, i.e., N = 400, and a vehicle velocity of 50 Km/h. The percentage of blocked tasks increased with the increase in traffic density in all three systems; however, the developed scheme achieved a higher efficiency in terms of blocked tasks. The developed MEC/SDN-based handover scheme achieved an average percentage improvement in the number of blocked tasks of 51% compared to the first system. Thus, the introduction of MEC in the handover scheme achieved a higher efficiency in terms of blocked tasks, due to the implementation of the handover scheme near to the end user, with less overhead. Moreover, the deployment of SDN with MEC facilitated the operation of the MEC during the handover implementation and the maintenance of the TCP connection.
The third performance metric was the percentage of failed handover processes compared to the total number of requested handover processes. Figures 11-13 provide the records of the handover failure rate for the developed handover scheme and the traditional schemes. These results were for different velocities, traffic densities, and numbers of deployed vehicles. Figure 11 provides the handover failure rate for the two systems at different traffic densities. These results were for a VANET with 400 vehicles and a vehicle velocity of 50 Km/h. The developed scheme achieved better efficiency in performing the handover process. Figure 12 provides the handover failure rate at different velocities, while Figure 13 provides the failure rate at different values of the number of deployed vehicles. The developed handover scheme achieved an average percentage improvement in the failure rate of the handover process of 18%.
J. Sens. Actuator Netw. 2022, 11, x FOR PEER REVIEW 13 of 16 records of the handover failure rate for the developed handover scheme and the traditional schemes. These results were for different velocities, traffic densities, and numbers of deployed vehicles. Figure 11 provides the handover failure rate for the two systems at different traffic densities. These results were for a VANET with 400 vehicles and a vehicle velocity of 50 Km/h. The developed scheme achieved better efficiency in performing the handover process. Figure 12 provides the handover failure rate at different velocities, while Figure 13 provides the failure rate at different values of the number of deployed vehicles. The developed handover scheme achieved an average percentage improvement in the failure rate of the handover process of 18%. Figure 11. Handover failure rate of traditional handover schemes and the developed scheme at different traffic densities.   records of the handover failure rate for the developed handover scheme and the traditional schemes. These results were for different velocities, traffic densities, and numbers of deployed vehicles. Figure 11 provides the handover failure rate for the two systems at different traffic densities. These results were for a VANET with 400 vehicles and a vehicle velocity of 50 Km/h. The developed scheme achieved better efficiency in performing the handover process. Figure 12 provides the handover failure rate at different velocities, while Figure 13 provides the failure rate at different values of the number of deployed vehicles. The developed handover scheme achieved an average percentage improvement in the failure rate of the handover process of 18%. Figure 11. Handover failure rate of traditional handover schemes and the developed scheme at different traffic densities.   records of the handover failure rate for the developed handover scheme and the traditional schemes. These results were for different velocities, traffic densities, and numbers of deployed vehicles. Figure 11 provides the handover failure rate for the two systems at different traffic densities. These results were for a VANET with 400 vehicles and a vehicle velocity of 50 Km/h. The developed scheme achieved better efficiency in performing the handover process. Figure 12 provides the handover failure rate at different velocities, while Figure 13 provides the failure rate at different values of the number of deployed vehicles. The developed handover scheme achieved an average percentage improvement in the failure rate of the handover process of 18%. Figure 11. Handover failure rate of traditional handover schemes and the developed scheme at different traffic densities. Figure 12. Handover failure rate of traditional handover schemes and the developed scheme at different vehicle velocities. Figure 13. Handover failure rate of traditional handover schemes and the developed scheme with different numbers of deployed vehicles. Figure 13. Handover failure rate of traditional handover schemes and the developed scheme with different numbers of deployed vehicles.

Conclusions
The handover process is a challenge in VANETs due to the high mobility conditions and the required user experience. This article introduced an MEC-based VANET structure and a seamless handover scheme for said structure. The handover process is implemented by an MEC server connected to an RSU, when vehicles come to the intersection of two coverage regions of neighboring RSUs. The developed scheme is based on three calculated threshold levels of received signal power in a way that ensures the handover process is performed before the received signal decreases below the reference threshold level that affects the packet reception. The developed scheme achieved a percentage improvement in the PDR of 64% compared to the existing traditional scheme. Moreover, the developed VANET achieved a 51% percentage improvement in blocked computing tasks compared to existing systems. The handover failure rate of the developed handover scheme was lower than the existing scheme by an average of 18%. Our future directions for this work include analyzing and modifying the algorithm for crossroads with overlapped coverage results from multiple RSUs. This includes adding more metrics, such as the RSU load and the vehicle's predicted direction.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.