Implementation and Simulation of Handover Techniques to Guarantee

Service Continuity through Microservices at Edge

Luigi Bitonti and Mauro Tropea

Dimes Department, University of Calabria, Via P. Bucci 39/c, 87036 Rende (CS), Italy

Keywords:

Automotive, Edge Computing, Collision Avoidance.

Abstract:

Edge computing is a mesh network of micro data centers, capable of processing and storing critical data locally,

and of transmitting all data received and / or processed to a central data center or to a cloud storage repository.

With this paradigm, data, applications and computational services are brought from the cloud servers to the

margins of the network. To divide their computational power, virtual hosts are distributed within a single

physical host through virtualization. One of the virtualization techniques that is spreading is containerized

programming. In this paper, the concepts just mentioned will be applied to a use case, to ensuring continuity

of service to vehicles traveling on a fast-moving road and which are managed by Edge technology. Two

handover algorithms supported by microservices at Edge are proposed and performance assessed.

1 INTRODUCTION

A cellular telecommunication network is a communi-

cation network that has the last link wireless and is

distributed geographically in areas called cells, each

served by at least one ﬁxed-location transceiver called

Base Transceiver Station (BTS). The division of ter-

ritory into cells allows to guarantee the service to a

large number of users with a single BTS that operates

the frequency reuse. When a mobile terminal is on the

edge of a cell, it starts a control mechanism in order

to check a signal with a greater power by a different

BTS. If the mobile terminal ﬁnds a new BTS with a

stronger power, it performs a procedure called han-

dover (Salihin et al., 2018). There are also four basic

techniques for handover: 1) Received signal strength

(RSS); 2) RSS with threshold; 3) RSS with hystere-

sis; 4) RSS with hysteresis and threshold. Based on

the handover metrics mentioned above, the decision

about how and when to switch the interface to which

network will be made.

Edge computing (Khan et al., 2019) has gained

considerable popularity in academic and industrial

sector in the last few years. It serves as a key en-

abler for many future technologies like 5G, Inter-

net of Things (IoT) (Shah and Yaqoob, 2016), aug-

mented reality and vehicle-to-vehicle communica-

tions by connecting cloud computing facilities and

services to the end users. And, a particular attention

has been addressed to security aspects of these tech-

nologies (Santamaria et al., 2019; Santamaria et al.,

2018). The Edge computing paradigm provides low

latency, mobility, and location awareness support to

delay-sensitive applications. The cell permanence

time and the mobile analysis for predictive services

are important aspects in wireless networks and they

are object of study by research communities, such as

it possible to view in (De Rango et al., 2008), (Fazio

et al., 2017), (Frnda et al., 2013).

In this work, we have considered as scenario a

Vehicular Ad-Hoc Network (VANET) (Ghori et al.,

2018) where the mobile device supports simultane-

ous connection to multiple access points. Some stud-

ies show the importance of prediction in handover

mechanism for guaranteeing QoS such as (Fazio et al.,

2016). During the handover process, there is no inter-

ruption of the service, given that the device remains

connected to at least one access point. Very studied

aspects for the ad-hoc network are routing issues as

it is possible to view in (Zhou et al., 2006), (Fotino

et al., 2007), (Socievole et al., 2011), and also for

VANET is an important topic as it is possible to view

in (De Rango et al., 2009), (Fazio et al., 2013). More-

over, in the considered scenario the support of edge

computing (Pan and McElhannon, 2017; Khan et al.,

2019) has been considered in order to permit the com-

munication in the vehicular environment taking into

account the latency, that can play an important role for

avoiding vehicles accidents. In this work, as edge de-

vice has been chosen the cellular BTS that is placed at

Bitonti, L. and Tropea, M.

Implementation and Simulation of Handover Techniques to Guarantee Service Continuity through Microservices at Edge.

DOI: 10.5220/0009971002870293

In Proceedings of the 10th International Conference on Simulation and Modeling Methodologies, Technologies and Applications (SIMULTECH 2020), pages 287-293

ISBN: 978-989-758-444-2

287

roadside and then it can assist the vehicles that mov-

ing in the road. The BTSs long the road are numer-

ous and then, they can collaborate each other in order

to guarantee service continuity to the mobile devices,

avoiding any form of interruption. In this work, a so-

lution has been proposed to guarantee the continuity

of the service in an automotive environment managed

with Edge Computing technology (Klas, 2017). This

technology was chosen because of its low latency and

high reliability properties in respect to Cloud architec-

ture (Pan and McElhannon, 2017). Docker containers

were used to develop the servers to make applications

light and (Doan et al., ). In this way, if the servers

do not need to use all their resources to satisfy the re-

quests, they could also be used by other clients. The

servers have been grouped in a Cluster, to be man-

aged efﬁciently. Kubernetes was used to orchestrate

the Cluster containers and to make the system robust

(Bachiega et al., 2018). To ensure the continuity of

the service, two handover algorithms have been im-

plemented. The ﬁrst algorithm is based on the thresh-

old RSS technique, the second one adds Cell Breath-

ing Mechanism. The proposed mechanisms have been

implemented in Java, taking advantage of libraries

and tools adapted to the context.

The rest of this paper is organized as follows: Sec-

tion 2 presents the proposal of guaranteeing service

continuity to vehicles on the road managed by Edge

technology. In Section 3, a description of the sim-

ulator and the implementation details are given. In

Section 4, we describe the numerical results. Finally,

Section 5 concludes the paper.

2 PROPOSED SOLUTION

The roadside BTSs communicate with each others

thanks to the transport infrastructure backbone. By

taking advantage of 5G networks, latencies are re-

duced to about 15 ms (Erel-

Ozc¸evik and Canberk,

2019). In this work, a smart BTS equipped with a pro-

cessing unit has been considered able to perform data

elaboration and then, to make decisions analyzing ob-

tained results on the basis of the proposal algorithm.

In this way, the BTS is also able to evaluate if it has to

propagate messages arriving from previous BTSs to-

wards the next BTSs or to stop forwarding because it

is not necessary. To apply this behavior the following

considerations can be made: all vehicles registered

with the BTS are tracked; if the last vehicle needs to

be notiﬁed, then the message must be propagated to

the adjacent BTS; vice versa, if the last vehicle is at

a distance from the event such that the warning is not

necessary, then it is useless to forward the message to

the adjacent BTS, because surely the vehicles under

its range will be far enough, and then it is not neces-

sary the use of the edge.

To implement handover processes it was neces-

sary to create two servers that work in parallel, able

to simulate two different BTSs. This was applied con-

sidering three virtual machines able to form the Ku-

bernetes cluster: one machine has been used as a mas-

ter; other two as worker nodes. The virtual machines

were hosted on a normal PC, as it is possible to view

in Figure 1.

Figure 1: Architectural Scheme.

An urban mobility simulator, SUMO, was used to

generate data from vehicles (Lim et al., 2017). SUMO

is an open source software that allows to create vari-

ous trafﬁc scenarios. This simulator was used to have

the most realistic data possible. It is also possible

to interface the simulator with the Java programming

language, thus making it easier to collect vehicle data

at runtime. The simulator runs on an other PC differ-

ent to the one hosting containers, as it is possible to

view in Figure 1.

For the sake of simplicity, it is assumed that all

cars travel in one direction. To facilitate the simula-

tion and make it faster, the two servers are supposed

to be at a distance of 100 meters from each other and

their coverage range is 90 meters. In this way, it is

possible to know what happens during handover since

the vehicles remain in the area covered by both nodes

for a sufﬁciently long time, see Figure 2.

Figure 2: Considered Test Environment.

In Figure 2, it is possible to note three different zones:

1) zone covered by ﬁrst server; 2) zone covered where

a mobile device can be covered by both ﬁrst and sec-

ond server, here takes place the handover mechanism;

3) zone covered by only second server.

SIMULTECH 2020 - 10th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

288

2.1 Docker and Kubernetes

Docker is an open source platform that allows to

develop and to run applications (doc, 2019). With

Docker it is possible to keep applications separate

from the hosting infrastructures, to make software

distribution possible quickly and easily using a struc-

ture called container. Kubernetes is a portable, ex-

tensible and open source platform useful for manag-

ing workloads and services developed with containers

(kub, 2019). An important concept in Kubernetes is

the cluster. It is a set of machines hosting container-

ized applications, where at least a machine is a mas-

ter node (controlling and managing operations) and

another one is a worker node (run the master instruc-

tions). The smallest unit that constitutes a Kubernetes

application is represented by the Pod, that represents

the processes running on the cluster.

2.2 Vehicle Trafﬁc Simulator

Through the TraaS library it was possible to interface

SUMO with Java (sum, 2019), (tra, 2019), (Artal-

Villa et al., 2019). Eclipse was used as a development

environment. For vehicle conﬁguration and simula-

tion, XML language ﬁles have been written. For each

vehicle, the IDentiﬁer (ID) number, speed and posi-

tion are taken. These parameters are formatted ac-

cording to a speciﬁc code, so as to be made legible

by the receivers. When the vehicle is in an area cov-

ered by both BTSs, the vehicle sends the message

to both, thus simulating the mechanism of macro-

diversity. Macro-diversity is a mechanism introduced

in UMTS networks through which it is possible to be

connected simultaneously to more than one cell (Ges-

bert et al., 2010). In any case, although the vehicle is

connected to both stations, only one is used to warn it

with warning messages. The choice on which the two

stations should have the control is made by the sta-

tions themselves and the algorithm, shown later, will

explain it. When the vehicle no longer receives the

signal from a station, it communicates this condition

to the one whose signal it is currently receiving, in

order to warn it of the new condition. In this way, a

soft handover was carried out: the vehicle passes un-

der control of a new station without remaining always

connected to the network. The continuity of the ser-

vice is thus guaranteed. The Constrained Application

Protocol (CoAP) was used to communicate with the

servers.

3 SOFTWARE

IMPLEMENTATION

The servers were implemented in Java language and

use the CoAP protocol (cal, 2019) to communicate

with the vehicles or other servers. For the manage-

ment of handover, two paradigms were used: a RSS

with threshold technique and a combination of RSS

with threshold and cell breathing. An explanation of

the type of messages that can be generated by the sys-

tem and a description of both algorithms will now be

given. Messages can be classiﬁed into ﬁve different

typologies:

• Type 0: it is the most used type of message. It

is a periodic message that each vehicle sends to

its reference server to update it on its speed and

position;

• Type 1: this message is sent from one server to

another upon it is detected a particular event;

• Type 2: it is sent from the vehicle to the server

for notifying handover end. The vehicle switches

from the old server to the new one;

• Type 3: it is a message that arises from the receipt

of a type 2 message. It is used to notify to the old

server that a vehicle is disassociated from it (the

old server removes it from its database);

• Type 4: it is the only message used in the second

handover implementation for communicating to a

server that its coverage range is decreasing.

Thanks to the exchange of these messages the soft

handover can be used and then a vehicle can pass

from a coverage area served by a BTS to the new area

served by another BTS able to manage the mobile de-

vice. In the following the two approaches, with a RSS

threshold and with cell breathing mechanism are de-

scribed and compared in order to show the behavior

of the proposal algorithms.

3.1 BTS with RSS Threshold

Mechanism

In this scenario, the server continually receives mes-

sages from vehicles that are under its coverage. To be

initialized, the server must know the location/position

of the neighbor server, its IP address, the port on

which to send the data and the range of coverage. Tak-

ing into consideration Figure 2, it is assumed that if

the vehicle is inside zone 2, it is in a location where

it can receive a signal from the radio base station 1

whose power is above of the acceptable threshold. In

zone 3, the signals from server 1 can no longer be

processed. The mechanism adopted for performing

Implementation and Simulation of Handover Techniques to Guarantee Service Continuity through Microservices at Edge

289

Figure 3: Flowchart of a server with RSS control mecha-

nism.

handover is based on RSS technique, a mechanism

well known in cellular networks (Tinkhede and In-

gole, 2014). When a vehicle receives a Type 0 mes-

sage, the vehicle’s position is saved. In this way the

server is aware of the presence of this vehicle under

its coverage. When a vehicle is in an alert condition,

such as a sudden braking, the server will notify all the

vehicles under its coverage and in the vicinity of the

vehicle that caused the alert generation. In the case

of server 2, it will send a notiﬁcation only to vehicles

in zone 3. Server 1, on the other hand, takes care of

vehicles in zone 1 and zone 2. Server 2 also sends a

Type 1 message to server 1 to inform it that a vehi-

cle under its coverage has made an emergency stop.

When a Type 1 message is received, the server must

check whether the vehicles under its control are near

the vehicle that has braked sharply. If so, it forwards

a message to these vehicles suggesting them of brak-

ing. In this way, thanks to the collaboration between

the two servers, it was possible to avoid potential ac-

cidents. Upon receiving a Type 2 message, the server

becomes aware that the vehicle is under its control,

since it is in a position such that the signal strength of

the server 1 is below a certain threshold. Therefore,

it sends a Type 3 message to the adjacent server, to

inform of this new condition. The message contains

the ID of the vehicle under handover process. When

a server receives a Type 3 message, it is informed that

a vehicle is no longer under its management. It can

therefore proceed with the removal of the data relat-

ing to it, not more necessary. Hereafter, the server will

no longer exchange messages with this vehicle. The

main issue of this implementation regards no load bal-

ancing functionality. In Figure 3, the ﬂow diagram of

this mechanism is depicted.

3.2 Server with Cell Breathing

Technique

In this section an improvement on the servers func-

tionality is presented in order to enhance the handover

implementation in the system modifying the behav-

ior of the BTSs in the intersected coverage area. The

proposed improvement is based on the cell breath-

ing concept (Thng et al., 2005), mechanism in use in

UMTS networks. The idea is to perform a sort of load

balancing into the BTS stations decreasing the num-

ber of vehicles to manage in a BTS with the conse-

quent increasing into the cell managed by the other

BTS. In the following, some details about the mech-

anism implementation in the considered system will

be explained. In order to consider the new mecha-

nism, three new data structures have been introduced.

The ﬁrst, called ZonaComune vehicles, keeps track

of the number of vehicles present in zone 2, i.e. in

the area where potentially both servers could manage

them. The second one, blackList, indicates those ve-

hicles which, although still in the coverage area, they

no longer need to be managed, as they have been as-

signed to the next server. The third one, VehiclesTo-

Controllare, indicates the vehicles that have been as-

signed to the new server. When a server receives a

message from a vehicle, it veriﬁes that its ID is not

one to be ignored. If so, the message is discarded

without further analysis, since the vehicle is under the

management of another server. Otherwise, it analyzes

the position of the vehicle and it controls if the ve-

hicle can potentially be transferred to another server

by comparing its position with the position of the ad-

jacent server and its range of coverage. This is, the

server becomes aware of the area where the vehicle

is located. If the vehicle is in zone 2, it is added to

the data structure called VehiclesZoneComune. After

having updated the list, the server checks if it has ex-

ceeded the maximum number of connections that it

can accept regarding the mobile devices belonging to

zone 2. If the limit has been exceeded, it virtually de-

creases its range of action. This means that it gives

control of the ﬁrst vehicle of the queue, effectively

relieving its computational load. To ensure that the

vehicle is supervised by the new server, it is neces-

sary to inform this server on the new condition. The

information is sent using the Type 4 message. When a

server receives a Type 4 message, it takes the vehicle

ID contained in that message and it enters this ID in

the appropriate data structure. In this way, the control

of the vehicle goes from one server to the other. So,

the handover was carried out using the Cell Breathing

paradigm. In Figure 4, the ﬂow diagram of the server

that acts with cell breathing paradigm is shown.

SIMULTECH 2020 - 10th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

290

Figure 4: Flowchart of a server with Cell Breathing tech-

nique.

4 PERFORMANCE EVALUATION

In this section, the performance evaluation carried out

through implemented simulator are shown. Different

simulation are performed varying some input param-

eters. The number of vehicles considered in the sim-

ulative campaigns was varied from 1 to 6

A ﬁrst result is referred to the number of pack-

ets managed by two servers that communicate with

the vehicles on the road. In Figure 5, the number of

packets managed without cell breathing paradigm is

shown, while in Figure 6, the number of packets in

the use case of cell breathing approach is depicted. In

the ﬁrst case, the management of vehicles and, there-

fore of the packets to be sent is entirely covered by

the ﬁrst server. In the second case, instead, despite

the vehicle positions are the same, a better load dis-

tribution can be appreciated. This is because some of

vehicles, although still in the range of the ﬁrst server,

are forced to perform handover, so moving under the

management of the second server.

Another interesting analysis can be made on the

time that elapses between the detection of the dan-

ger event and the communication of this event to the

last vehicle of the set. The experiments were con-

ducted using both handover techniques. Considering

only the ﬁrst mechanism, the vehicles are all under

the control of server 1. Ten different simulations are

performed in order to have an average value. The

Figure 5: Packages managed without cell breathing tech-

nique.

Figure 6: Packages managed with cell breathing technique.

Figure 7: Latency without cell breathing technique.

recorded roundtrip time is around 40 milliseconds,

see Figure 7, with a standard deviation of around 5

milliseconds. The time that elapses between the trans-

mission of the message by the vehicle and the pro-

cessing of the same by the server is about 20 millisec-

onds. Moreover, we have observed that the number of

involved vehicles does not contribute to this latency

because the experiments show that cars receive the

danger signal almost simultaneously, with a time dif-

ference of less than one millisecond. Considering in-

stead the second approach, half of the vehicles present

in the area are under the control of server 2; the other

half are under the control of server 1. In this case

the recorded roundtrip time is around 60 milliseconds,

with a standard deviation of around 5 milliseconds be-

cause another hop is added in the messages forward-

ing, see Figure 8.

Implementation and Simulation of Handover Techniques to Guarantee Service Continuity through Microservices at Edge

291

Figure 8: Latency with cell breathing technique.

Figure 9: Handover per time unit.

Finally, the last analysis concerns the number of han-

dovers. The ratio between the handover number and

the employed time has been calculated. In this way

the handover rate per seconds has been obtained. This

value is inﬂuenced by the vehicles speed and their dis-

tance from the vehicle ahead. Proceeding at higher

speeds, the BTS must manage a greater number of

handovers, because the speed with which the mobile

device changes position in the area is greater, see Fig-

ure 9. Considering same speeds, the number of han-

dovers per time unit increases, as the distance be-

tween two vehicles decreases.

5 CONCLUSIONS

In this paper, a solution to guarantee the continuity

of the service in an environment managed with Edge

Computing technology has been proposed. Docker

containers have been used to make applications light

and portable. Kubernetes was used to orchestrate the

Cluster containers and make the system robust. The

results obtained show that, based on the algorithm

used, there may be improvements in the management

of the mobile devices. The use of Cell Breathing fa-

vors the balancing of the computational load, which

becomes a crucial factor if compared to realistic en-

vironments, in which the number of vehicles to man-

age is quite high. Conversely, the latency recorded by

some vehicles increases, which leads to delays in the

delivery of messages.

REFERENCES

(2019). Docker: Empowering app development for devel-

opers. https://www.docker.com/.

(2019). Eclipse californium (cf) coap framework. https:

//www.eclipse.org/californium/.

(2019). Kubernetes: Production-grade container orchestra-

tion. https://kubernetes.io/.

(2019). Sumo - simulation of urban mobility. http://sumo.

sourceforge.net/.

(2019). Traci/traas - sumo documentation. https://sumo.dlr.

de/docs/TraCI/TraaS.html.

Artal-Villa, L., Hussein, A., and Olaverri-Monreal, C.

(2019). Extension of the 3dcoautosim to simulate ve-

hicle and pedestrian interaction based on sumo and

unity 3d. In 2019 IEEE Intelligent Vehicles Sympo-

sium (IV), pages 885–890. IEEE.

Bachiega, N. G., Souza, P. S., Bruschi, S. M., and de Souza,

S. d. R. (2018). Container-based performance eval-

uation: A survey and challenges. In 2018 IEEE In-

ternational Conference on Cloud Engineering (IC2E),

pages 398–403. IEEE.

De Rango, F., Fazio, P., and Marano, S. (2008). Utility-

based predictive services for adaptive wireless net-

works with mobile hosts. IEEE Transactions on Ve-

hicular Technology, 58(3):1415–1428.

De Rango, F., Veltri, F., Fazio, P., and Marano, S. (2009).

Two-level trajectory-based routing protocol for vehic-

ular ad hoc networks in freeway and manhattan envi-

ronments. Journal of Networks, 4(9):866–880.

Doan, T. V., Nguyen, G. T., Salah, H., Pandi, S., Jarschel,

M., Pries, R., and Fitzek, F. H. Containers vs virtual

machines: Choosing the right virtualization technol-

ogy for mobile edge cloud. In 2019 IEEE 2nd 5G

World Forum (5GWF), pages 46–52. IEEE.

Erel-

Ozc¸evik, M. and Canberk, B. (2019). Road to 5g

reduced-latency: A software deﬁned handover model

for embb services. IEEE Transactions on Vehicular

Technology, 68(8):8133–8144.

Fazio, P., De Rango, F., Tropea, M., and Voznak, M. (2017).

Cell permanence time and mobility analysis in infras-

tructure networks: Analytical/statistical approaches

and their applications. Computers & Electrical En-

gineering, 64:506–523.

Fazio, P., Sottile, C., Santamaria, A. F., and Tropea,

M. (2013). Vehicular networking enhancement and

multi-channel routing optimization, based on multi-

objective metric and minimum spanning tree. Ad-

vances in Electrical and Electronic Engineering,

11(5):349–356.

Fazio, P., Tropea, M., De Rango, F., and Voznak, M. (2016).

Pattern prediction and passive bandwidth manage-

ment for hand-over optimization in qos cellular net-

works with vehicular mobility. IEEE Transactions on

Mobile Computing, 15(11):2809–2824.

Fotino, M., Gozzi, A., Cano, J.-C., Calafate, C., De Rango,

F., Manzoni, P., and Marano, S. (2007). Evaluating

energy consumption of proactive and reactive routing

protocols in a manet. In IFIP Conference on Wireless

Sensor and Actor Networks, pages 119–130. Springer.

SIMULTECH 2020 - 10th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

292

Frnda, J., Voznak, M., Rozhon, J., and Mehic, M.

(2013). Prediction model of qos for triple play ser-

vices. In 2013 21st Telecommunications Forum Telfor

(TELFOR), pages 733–736. IEEE.

Gesbert, D., Hanly, S., Huang, H., Shitz, S. S., Simeone, O.,

and Yu, W. (2010). Multi-cell mimo cooperative net-

works: A new look at interference. IEEE journal on

selected areas in communications, 28(9):1380–1408.

Ghori, M. R., Zamli, K. Z., Quosthoni, N., Hisyam, M.,

and Montaser, M. (2018). Vehicular ad-hoc net-

work (vanet): Review. In 2018 IEEE International

Conference on Innovative Research and Development

(ICIRD), pages 1–6.

Khan, W. Z., Ahmed, E., Hakak, S., Yaqoob, I., and Ahmed,

A. (2019). Edge computing: A survey. Future Gener-

ation Computer Systems, 97:219–235.

Klas, G. (2017). Edge computing and the role of cellular

networks. Computer, 50(10):40–49.

Lim, K. G., Lee, C. H., Chin, R. K. Y., Yeo, K. B., and Teo,

K. T. K. (2017). Sumo enhancement for vehicular ad

hoc network (vanet) simulation. In 2017 IEEE 2nd

International Conference on Automatic Control and

Intelligent Systems (I2CACIS), pages 86–91. IEEE.

Pan, J. and McElhannon, J. (2017). Future edge cloud and

edge computing for internet of things applications.

IEEE Internet of Things Journal, 5(1):439–449.

Salihin, S. S., Nissirat, L. A., Noor, R. M., and Ahmedy, I.

(2018). Handover schemes for vehicular ad-hoc net-

works over long term evolution advanced: A survey.

In 2018 International Conference on Computational

Approach in Smart Systems Design and Applications

(ICASSDA), pages 1–7. IEEE.

Santamaria, A. F., Raimondo, P., Tropea, M., De Rango,

F., and Aiello, C. (2019). An iot surveillance sys-

tem based on a decentralised architecture. Sensors,

19(6):1469.

Santamaria, A. F., Tropea, M., Fazio, P., and De Rango,

F. (2018). Managing emergency situations in vanet

through heterogeneous technologies cooperation. Sen-

sors, 18(5):1461.

Shah, S. H. and Yaqoob, I. (2016). A survey: Internet

of things (iot) technologies, applications and chal-

lenges. In 2016 IEEE Smart Energy Grid Engineering

(SEGE), pages 381–385. IEEE.

Socievole, A., De Rango, F., and Coscarella, C. (2011).

Routing approaches and performance evaluation in

delay tolerant networks. In 2011 Wireless Telecom-

munications Symposium (WTS), pages 1–6. IEEE.

Thng, K. L., Yeo, B. S., and Chew, Y. H. (2005). Perfor-

mance study on the effects of cell-breathing in wcdma.

In 2005 2nd International Symposium on Wireless

Communication Systems, pages 44–49. IEEE.

Tinkhede, P. and Ingole, P. (2014). Survey of handover de-

cision for next generation. In International Confer-

ence on Information Communication and Embedded

Systems (ICICES2014), pages 1–5. IEEE.

Zhou, B., Lee, Y.-Z., Gerla, M., and De Rango, F. (2006).

Geo-lanmar: a scalable routing protocol for ad hoc

networks with group motion. Wireless Communica-

tions and Mobile Computing, 6(7):989–1002.

Implementation and Simulation of Handover Techniques to Guarantee Service Continuity through Microservices at Edge

293