Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Load Balancing and QoS-Aware Network Selection

Scheme in Heterogeneous Vehicular Networks


Noura Aljeri and Azzedine Boukerche
PARADISE Research Laboratory, EECS, University of Ottawa, Canada
nalje094@uottawa.ca, boukerch@site.uottawa.ca

Abstract—With the increasing demands for various wireless Henceforth, an efficient handover scheme is necessitated to
communication technologies and standards, new challenges arise mitigate the vehicle’s mobility and network conditions to en-
in seamless connectivity among different techniques. Mobility hance the handover method. In general, the handover process
management protocols face a new difficulty in vehicular net-
work heterogeneity, from deployment issues to optimal handover is done in three main steps: Network discovery, Handover
management process. However, due to the dynamic environment trigger, and Network selection [3]. The most common trigger-
in vehicular networks, providing the best quality of service is ing condition is the received signal strength (RSS) threshold
a critical issue. Additionally, conventional mobility management value [4], which is the amount of power transmitted by a given
solutions do not consider user’s preferences when selecting the cell, and the selection of the next network is based on the
next points of access. In this paper, we present a load balancing
and QoS-aware handover scheme in Heterogeneous Vehicular best-offered RSS value among neighboring cells. Nevertheless,
Networks (Het-VeNET) in order to choose the least loaded this will lead to network congestion, as vehicles will always
network, while maintaining the high level of required quality of access the maximum offered SINR by nearby cells. Also, the
service. We define the vehicle’s mobility and QoS measurements dependence on one signal feature in selecting the next access
and demonstrate the stated handover process in a vehicular point is not a realistic approach, as it does not consider the
environment. The proposed scheme performance showed a higher
rate of successful handoff and load balance on different cells and required service quality nor the vehicle’s mobility projections
scenarios when compared to benchmark schemes. and network load.
Index Terms—Mobility Management, Handover, QoS, Load In this paper, we propose a network selection scheme
balance, Heterogeneous Vehicular networks for heterogeneous vehicular networks to handle the handover
process and consider QoS parameters while guaranteeing
I. I NTRODUCTION a balanced load on the network cells. We first define the
Over the past few years, we have witnessed significant vehicle’s mobility and QoS measurements to be included in the
growth in wireless communication technologies; each offers mobility management process, in which vehicles and access
an additional contribution to the current standards from low- points contribute to providing the available QoS on cells and
latency to increased bandwidth availability, referred to as preferred quality of service by users. Furthermore, a new
Heterogeneous Networks [1]. In the context of vehicular handover trigger is added to the protocol that does not rely on
networks, providing vehicles with a variety of access tech- the signal strength only but also considers current cell service
nologies to support real-time applications and services is a quality and the surrounding neighbors’ proximity to initiate
promising architecture for the current advances in vehicular the handover process. Thus, the proposed scheme guarantees
technologies. From autonomous driving to traffic monitoring users preferred QoS while maintaining a balanced network.
and safety applications [2], all of which require efficient and The rest of the paper is organized as follows. In Sec-
reliable communications to deliver content information in a tion II, we discuss the related works of mobility management
vehicular environment. However, network heterogeneity comes schemes in vehicular networks. The proposed network model
with a drawback in mobility management between different and measurement functions are derived in Section III. In
interfaces. Section IV, we present and describe the proposed mobility
handover management protocol in each phase. This is followed
Mobility management is concerned with two essential as-
by the simulation experiments and performance evaluation of
pects, handover and location management protocols [3]. The
our model in Section V. We conclude our work and future
later tracks and update the vehicles’ location, while handover
direction in Section VI.
management provides seamless mobility to vehicles when
transferring communication links between two points of ac- II. BACKGROUND
cess. However, with the vehicles’ rapid movement and network
A heterogeneous network offers several wireless access
topology changes, providing smooth handover is a challenge,
technologies and sometimes different types of macro and
especially in heterogeneous vehicular networks (Het-VeNET).
small cells with varying ranges of coverage. Vehicular net-
This paper is partially supported by NSERC, CREATE TRANSIT, and work communications are usually classified into two main
Canada Research Chairs. types: vehicle to vehicle (V2V) communications and vehicle

978-1-7281-5089-5/20/$31.00 ©2020 IEEE

Authorized licensed use limited to: University of New South Wales. Downloaded on August 09,2020 at 02:09:52 UTC from IEEE Xplore. Restrictions apply.
to infrastructure (V2I) communications [5]. Recently, new movement trend and signal strength, among other parameters,
additional types are introduced, including vehicle to network to derive a network selection. However, not many works
(V2N), and vehicle to pedestrian (V2P), all under the umbrella consider user preferences and network load parameters in their
of V2X communication standards. In this work, we are more decision making.
focused on V2I communication standards. The most common
communication standard for vehicular communication is the III. N ETWORK M ODEL
Dedicated Short-Range Communication (DSRC), also known In our model, we consider a vehicular network comprised
as IEEE802.11p [6], introduced by the IETF group. However, of different Radio Access Technologies (RATs) such as LTE,
the advances of vehicular applications and services come as a WiMAX, and WiFi. The mobile terminals’ ability to switch
shortcoming for DSRC moderate latency and short-range. between different RATs is called the vertical handover pro-
The DSRC protocol is the standard for vehicular communi- cess [15]. Vehicles and cells use the standard interface card
cation, but many research studies examined the use of cellular MIH to facilitate the transitions between different access
technologies, such as LTE-Advanced, to support vehicular technologies. The vertical handover process is considered
network applications [7]. Since the new trend of wireless net- expensive and usually suffers from long and hard handover
work architecture includes various radio access technologies, delays, in which vehicles drops communication link between
a standard interface is needed to carry communication and itself and the current access point then initiate the next
transition between them. The IETF group introduced a mobil- handover process. Therefore, data links are disrupted, and
ity management solution, named Media Independent Handover packets are lost until a new connection is established. These
(MIH), referred to as IEEE802.21 [8], to provide seamless issues have led many researchers to consider other alternatives
mobility between RATs. The MIH interface defines two main to facilitate and provide a seamless handover process between
entities, Point of Service (PoS) and Point of Attachments different technologies [16].
(PoA’s). The later holds a set of available access technologies
in the network, while the former maintains communication A. Problem Definition
between the network and vehicles. Three primary services In vehicular networks, it is essential to select the most
exist in the MIH interface, namely Event service, Command suitable network among the available wireless access technolo-
service, and Information service [9]. gies such as WiFi, Cellular, or WiMAX. The selection of the
Most of the related work studies addressed issues with the network should take into account several metrics to provide
handover process through each phase, the handover trigger, the optimal performance, including user preference, available
and network selection. Omheni et al. [10] proposed a vertical bandwidth, and network load. In a conventional handover, ve-
handoff protocol that monitors the communication link’s signal hicles will always choose the best available service regardless
strength quality using two threshold values. The received of their velocity nor preferences. Many solutions worked on
signal strength (RSS) is defined as the amount of power from enhancing the selection process by including vehicles’ trajec-
a received signal. However, a single point of measurement tory, speed, and distance to access point. However, selecting
may cause handover failures if not correctly estimated. Other the best-desired quality of service among multiple cells while
studies use Intelligent and fuzzy-based solutions that may maintaining balance load on access points is yet to be fully
require a considerable amount of time for computations. explored.
The network selection methods can be categorized into The general process of any handover scheme involves three
multiple attributes decision, intelligence-based, and function- main steps, discovery, trigger, and network selection. Several
based [11]. The multiple attributes decision making considers scenarios in which the handover process may fail and cause
various characteristics from either the network or terminal either ping-pong effect or link failure [17]. The handover
devices, such as cell coverage, cost, load, latency, user pref- trigger may be too early, or too late. If the handover is
erences, or application requirements. Ndashimye et al. [12] triggered too early, then resources will be allocated too soon,
proposed a fuzzy-logic network selection scheme for V2I and the vehicle does not connect to the target cell. If its
communications. A self-evaluating strategy is used by vehicles too late, the handover will not be completed by the time
to estimate the best available network. Each vehicle measures the vehicle leaves the current serving cell, and service will
the dwelling time of cells using their location and direction be disrupted until the transfer is completed. Timely triggered
to choose the best option. However, their work does not handover does affect the overall performance of the handover
take into account any QoS parameters and users’ preferences. process. In our scheme, we trigger the handover process when
Some research studies investigated handover management in the vehicle estimated link duration time with the current access
terms of energy efficiency. A proposal by Nguyen et al. [13], points exceeds a specific threshold value. Assuming that the
presented an energy-function based network selection scheme vehicle desired QoS will remain the same throughout the
for heterogeneous networks. The selection strategy is based on simulation time.
the terminal user and using several measurements to derive a As for the network selection, any parameters are involved
decision such as power consumption gain and link quality. in the decision making of the selection process, either from
Furthermore, a vertical handoff decision scheme proposed the network measurements (RSS, SINR, bandwidth, data rate,
in [14], which is based on a decision tree model. They use cost, etc.) or from the user measurements (velocity, location,

Authorized licensed use limited to: University of New South Wales. Downloaded on August 09,2020 at 02:09:52 UTC from IEEE Xplore. Restrictions apply.
Fig. 1: Heterogeneous Vehicular Networks Scenarios.

and battery power, and preference) or both. However, relying Working on the same previously described analogy, in which
entirely on the network measurements may lead to network small-cells are not to be considered when the vehicles’ mobil-
congestion, as the selection will always be to access the best ity is high unless there are no other options in its proximity.
cell (i.e., Always Best Connection (ABC) concept). Taking The direction and distance of vehicles will determine the
into account the user preferences and the load balance on cells eligible access points that can provide the vehicle with a
is a necessity to overcome the drawbacks of conventional han- reasonable residual time in its range, assuming the vehicle
dover decision schemes. In Figure 1, we demonstrate several will maintain its direction and speed using the link duration
scenarios in which a vehicle may be facing in heterogeneous time estimation as follows
vehicular networks. Generally, vehicles that are moving in a R − |dist(xi , aj )|
relatively high-speed are better to be connected with broad LDT = (1)
vi
range cells, named macrocells, and vehicles that are parking or
moving in low-speed are connected with smallcells. The use of where dist(.) is the euclidean distance between the vehicle xi
smallcells will help in offloading traffic load from macrocells, and an access point aj , R is the communication range of the
in addition to being within close proximity to vehicles to access point, and v is the current velocity of the vehicle.
provide more reliable services. Nonetheless, in some cases, After that, a list of the possible points of connection that
vehicles are looking for a specific quality of services among guarantees a duration time above λ is generated. Assuming
available cells, as seen in Figure 1 second scenario. In such that more than one access point option is present for the
cases, the decision making will consider not only the vehicle’s vehicle, the next selection strategy will be based on the user
mobility but also the cell’s available quality of service that preference for the Quality of Service.
meets the users’ preferred options. In later sections, we discuss C. QoS Measurements
the type of quality of services that can be measured in the
network cell and terminal devices. Furthermore, in many cases, The handover process is usually triggered because of three
using the QoS measurements may cause network congestion main reasons: i) to reduce the load on macrocells; ii) a vehicle
when many handover requests are going toward the same is seeking for a better QoS in terms of cost or bandwidth; iii)
access point. or a vehicle is on the border of the current serving cell’s
communication range and requires a new registration. The
B. Mobility Measurements selection mechanism of a desired quality of service cell works
Each cell periodically transmits a broadcast packet to ad- as follows. Each cell is set with a QoS parameter and vehicles
vertise itself and its available services. Vehicles maintain a that search for point of access to switch to, for a better QoS
neighboring list of every cell in its proximity, which contains in term of cost, latency, or bandwidth are the triggers for the
the cell id, location, available bandwidth, cost, advertisement handover process. The vehicle defines its preferences through
timer, and load capacity. The mobility of vehicles is an desired throughput, delay, and loss.
essential factor in the network selection process, especially We define the QoS vector to include several parameters in
in Het-VeNET. Therefore, we set the mobility attributes as both vehicles and access points. The vehicle’s QoS vector
the first set of measurements that need to be considered when maintains the user preferences in terms of the minimum
the handover process is triggered. According to the vehicles’ bandwidth and its weight value β. Each access point will
velocity, direction, and location, the decision process will start also maintain its QoS vector with the available bandwidth and
eliminating the undesirable access points in its neighborhood. usage cost, which is shared with vehicles in its communication

Authorized licensed use limited to: University of New South Wales. Downloaded on August 09,2020 at 02:09:52 UTC from IEEE Xplore. Restrictions apply.
Algorithm 1 Network Selection Method points in their proximity. The list will be periodically updated,
Require: B, (c) ,Listn , cAP and the past lifetime advertisement will be deleted. If a
1: procedure H ANDOVER T RIGGER cell’s entry already exists in the list, then it will be updated
2: while Every T time do accordingly. Similarly, each cell maintains a list of the vehicles
3: Update vehicle’s link duration time LDT with cAP
4: Get cAP RSSI
connected to it during a specific time, this help determines the
5: if LDT < thresh1 or RSSI < thresh2 then cell’s load. The early discovery of nearby cells is essential to
6: Select Network() reduce the amount of time in which the vehicle is required to
7: end if make a decision.
8: end while
9: end procedure B. Handover Trigger Phase
10: procedure S ELECT N ETWORK
11: Eliminate undesired APs from Listn As mentioned in earlier sections, the handover trigger is
12: New short list L initiated due to the vehicle’s losing signal strength from the
13: get vehicle’s QoS vector (B, β) current serving cell and requires a new connection. The time
14: for each AP in L calculate Eq (2) of trigger is crucial to avoid any handover failure or ping-
15: get maximum APi function value
16: if APi == cAP then
pong events. In our method, each vehicle will periodically
17: continue measure the link duration time and the received signal strength
18: else between itself and the current serving cell. If the estimated
19: Start handover process with APi link duration time is below a specific threshold value or the
20: end if signal strength weakens below another threshold value, then
21: end procedure
the handover initiation is triggered, and the vehicle will begin
selecting the next cell and execute the handover process. In
another situation, if the vehicle was not able to find the desired
range. According to the desired QoS and the available set of level of QoS from the current AP’s region, then the handover
access point’s information with the vehicle, an AP selection process can be triggered if such a cell is available.
can be determined by the total QoS function as follows
C. Network Selection Phase
(b, c) = β(b) − (1 − β)(c) (2)
The important aspect of our proposed scheme, is the net-
where (b) is the bandwidth function, (c) is the cost work selection strategy, using the neighboring cell’s table, a
function, and β is the weight value on the access point’s vehicle will start the selection process according to the derive
bandwidth between 0 ≤ 1. The bandwidth and cost functions QoS attributes mentioned earlier in Equation (2). The selection
are calculated by of networks involves the vehicles desired QoS preference
 and the available cells’ QoS states. After a handover trigger
1 if b ≥ B
(b) = is initiated, vehicles will need to find another alternative
0 if b < B before they lose communication with its current AP. Each
where B is the user minimum bandwidth preference. There- vehicle maintains a list of neighboring access points that have
fore, the QoS function of each available access point can be been collected either by listening to broadcast messages from
derived by the previous equation to determine the highest nearby access points or by sending a solicitation message to
access point’s function value for the vehicle to connect to, request information.
in alignment with their desired preference. The network selection method presented in Algorithm 1, in
which an elimination process will first begin by removing the
IV. T HE PROPOSED L OAD BALANCING AND Q O S- AWARE undesired access points using the vehicle’s current mobility
N ETWORK S ELECTION S CHEME measurements. If the vehicle exhibits a high-speed mobility
The main goal of our proposed scheme is to select the least trend, then smallcells will not be used in the selection process,
loaded access point while maintaining a high level of service with the assumption that vehicles will always be presented
quality. The handover process can be described in three phases, with at least one macrocell and one smallcell. In another case,
network discovery, handover initiation, and network selection. vehicles with low-speed mobility are preferred to use small-
In the following, we define each stage separately according to cells unless the desired QoS is not available in nearby small-
our proposed solution. cells; therefore, macrocells will be considered the alternative.
A new shortlist is established, with each cell assigned a QoS
A. Network Discovery Phase value according to the derived QoS equation and the user’s
Each cell or Access Point (AP) will periodically send preference vector. Then, select the best access point’s state
advertisement messages to nearby vehicles. An advertisement among all to be the chosen network for handover. However,
packet contains the cell’s ID, cell’s location, QoS vector in case there is more than one access point with the desired
(i.e., available bandwidth and cost), ads lifetime, and average QoS value, a load balancing mechanism is initiated to select
network load. Upon receiving the AP advertisement message, the least loaded cell. Therefore, maintaining a distributed data
vehicles will retain a neighboring list of the available access load among access points in an area.

Authorized licensed use limited to: University of New South Wales. Downloaded on August 09,2020 at 02:09:52 UTC from IEEE Xplore. Restrictions apply.
In Figure 3a, an illustration of the average number of
successful handovers that satisfies the user’s QoS preferences
over the simulation time has shown the high accuracy rate
of the proposed scheme in comparison to the conventional
protocol. It is worth noting that the original protocol satisfies
the required QoS almost 50% of the time when 20% of vehi-
cles request certain Quality of Service, and this is merely ran-
domized depending on the available access points. Figure 3b
portrays the average handover latency of the proposed load-
balancing and QoS-aware handover process and a conventional
RSS-based method over the simulation time. The traditional
and QoS-aware handover schemes both produced resembling
handover latencies with an average of 18ms. This indicates
that the addition of QoS network selection and load balancing
scheme does not affect the performance of the handover
Fig. 2: Ottawa Urban environment map. process. Moreover, the handover latency is relatively low and
thus reduces the chances of prolonged service disruption.
TABLE I: Simulation parameters.
In terms of the efficiency of the network selection, we set a
Parmeter name Parameter value
corresponding node to transmit data packets between a service
Network cells smallcells / macrocells
provider and vehicles and evaluate the data packet’s end-to-
Number of cells 3 smallcells - 2 macrocells
Transmission range (m) 250 / 1000
end delay over the simulation time in Figure 3c. The proposed
Number of vehicles 100 QoS-based scheme reduces the packet delivery delay in com-
Simulation environment Urban parison to the RSS-based handover management scheme by
Simulation time (s) 500 an average of 4%. The reason behind the low difference in
reduction is because the simulation setup is relatively small
in size; a more dense network will show the impact of our
load balancing and QoS-aware handover. The network load
V. P ERFORMANCE E VALUATION
in each cell is presented in Figure 4a, with 20% of vehicles
In this section, we present the performance evaluation requesting a quality of service that is offered by the smallcells.
of our protocol compared to the benchmark Mobile IPv6 The results show more than 60% of the load on one cell
protocol. The simulation of infrastructure-based heterogeneous in a conventional handover process. This is due to the fact
vehicular networks is done in NS-2. We use SUMO generated that the handover process is always selecting the best cell for
mobility traces of Ottawa’s urban scenario, as illustrated in vehicles in terms of received signal value and not considering
Figure 2, to evaluate the proposed scheme. In Table I, we the network’s available bandwidth and load. On the other hand,
summarize our simulation parameters, with macrocells and the proposed protocol has shown a more balanced network
smallcells that provide a transmission range of 250m and load, which utilizes nearby access points that offers similar
1000m, respectively. QoS performances.
To evaluate our proposed load balancing and QoS-aware
handover scheme, we tested the handover scheme in various A comparison in the network load is presented in Figures 4b
cases. In our set of tests, we assume that only 20% of vehicles and 4c over different percentages of vehicles requesting a
require specific QoS for their connection according to the certain quality of service, from 10% to 50% of the vehicles
application being used on the vehicle. We gathered the results requesting QoS that is offered by two smallcells AP1 and
after several runs with an average of 95% confidence interval. AP2 and one macrocell Cell 1. As noticed in Figure 4b,
The evaluation is based on several metrics, including the the conventional handover method increases the load on the
following macrocell with an average of 33% of the requests, as the
• Load on cells: measures the average load on each access number of requesting vehicles increase. Comparatively, the
point, which is the average number of vehicles that are proposed protocol load-balanced mechanism distributes the
being served by a specific cell during a step in time. load among the other two smallcells with an average of 21% in
• Average packet loss: measures the amount of service both APs and 16% on cell 1. Moreover, the high load on AP3
packet over the cumulative number of sent packets. is due to the central location of the access point, even though
• Handover latency: defines the total time required to com- Cell-2 offers the same QoS as AP3. Since our protocol only
plete the handover process from trigger to completion. balance the load when QoS is established, the load on other
• QoS success rate: measures the number of successful cells was not considered. Also, AP3 attracts more vehicles
handovers between vehicles and the desired QoS over when no QoS parameter is introduced or no available access
the total number of handoffs. points in the region that offers the desired quality of service.

Authorized licensed use limited to: University of New South Wales. Downloaded on August 09,2020 at 02:09:52 UTC from IEEE Xplore. Restrictions apply.
a: QoS success rate. b: HO latency. c: End-to-End delay.
Fig. 3: Comparison between the proposed QoS-aware HO and RSS-based HO/

a: 20% Vehicles request QoS in AP1, AP3. b: QoS-based HO without LB. c: LBQoS-based HO.
Fig. 4: Network load over different QoS percentages.

VI. C ONCLUSION [6] D. Jiang, and L. Delgrossi, 2008. IEEE 802.11 p: Towards an interna-
tional standard for wireless access in vehicular environments. In VTC
Heterogeneous vehicular networks ensure high-performance 2008. IEEE, 2036–2040.
efficiency to vehicular applications and services. However, [7] B. Masini, A. Bazzi, and E. Natalizio, 2017. Radio access for future 5G
providing end-users with seamless mobility over various ac- vehicular networks. In 2017 (VTC). IEEE, 1–7.
[8] K. Taniuchi et al.. 2009. IEEE 802.21: Media independent handover:
cess technologies is a challenging task. In this paper, we Features, applicability, and realization. IEEE Comm. Magazine 47, 1
proposed an efficient load balancing and QoS-aware handover (2009), 112–120.
scheme for heterogeneous vehicular networks. Our proposed [9] A. Michalas, A. Sgora, and D. Vergados. 2017. An integrated MIH-
FPMIPv6 mobility management approach for evolved-packet system ar-
protocol succeeds in balancing the load between the different chitectures. Journal of Network and Computer App. 91 (2017), 104–119.
wireless access technologies in comparison to the conventional [10] N. Omheni et. al. 2017. Smart mobility management in 5G heteroge-
handover scheme. Moreover, it guaranteed the vehicles’ QoS neous networks. IET Networks 7, 3 (2017), 119–128.
[11] H.W. Yu and B. Zhang. 2018. A heterogeneous network selection
requirements and distributed load among network entities algorithm based on network attribute and user preference. Ad Hoc
while showed lower packet delivery delays. The proposed Networks, 72, pp.68-80.
scheme was tested on Ottawa urban map and SUMO mobility [12] E. Ndashimye, N. I Sarkar, and S. Ray. 2018. A network selection
method for handover in vehicle-to- infrastructure communications in
traces. In future work, we plan to investigate several mobility multi-tier networks.Wireless Networks (2018), 1–15.
models to evaluate the impact of mobility on the performance [13] Q-T Nguyen-Vuong, N. Agoulmine, and Y. Ghamri-Doudane. 2008.
of our protocol. A user-centric and context-aware solution to interface management
and access network selection in heterogeneous wireless environments.
R EFERENCES Computer Networks 52, 18 (2008), 3358–3372.
[14] H. Wang, C. Fan, C-H. Hsu, Q. Sun, and F. Yang. 2014. A vertical hand-
[1] Z. He, J. Cao and X. Liu. 2016. SDVN: Enabling rapid network off method via self-selection decision tree for internet of vehicles.IEEE
innovation for heterogeneous vehicular communication. IEEE network, Systems. 10, 3 (2014), 1183–1192.
30(4), pp.10-15. [15] J. Márquez-Barja, C.T. Calafate, J.C Cano, and P. Manzoni. 2011. An
[2] T. Kim, A. Hobeika, and H. Jung. 2018. Evaluation of the performance overview of vertical handover techniques: Algorithms, protocols and
of vehicle-to-vehicle applications in an urban network. Journal of ITS. tools. Computer Comm., 34(8), pp.985-997.
22, 3 (2018), 218–228. [16] L. Chen and H. Li. 2016. An MDP-based vertical handoff decision
[3] A. Boukerche , A. Magnano, and N. Aljeri. 2017. Mobile IP Handover algorithm for heterogeneous wireless networks. In 2016 IEEE Wireless
for Vehicular Networks: Methods, Models, and Classifications. ACM Comm. and Networking Conf.. IEEE, 1–6.
Comput. Surv. 49, 4, Article 73 (2017), 34. [17] H-W Yu and B. Zhang. 2018. A heterogeneous network selection algo-
[4] R.C. Luo, and O. Chen. Mobile sensor node deployment and asyn- rithm based on network attribute and user preference.Ad Hoc Networks
chronous power management for wireless sensor networks,IEEE Trans. 72 (2018), 68–80.
on Industrial Electronics, 59 (2012) 2377-2385. [18] Y. Lee, B. Shin, J. Lim, and D. Hong. 2010. Effects of time-to-trigger
[5] S. Chen et. al. 2017. Vehicle-to-everything (V2X) services supported parameter on handover performance in SON-based LTE systems. In
by LTE-based systems and 5G.IEEE Comm. Standards Magazine 1, 2 2010APCC. IEEE, 492–496.
(2017), 70–76.

Authorized licensed use limited to: University of New South Wales. Downloaded on August 09,2020 at 02:09:52 UTC from IEEE Xplore. Restrictions apply.

You might also like