Multi-MNO Predictive-QoS for Vehicular Applications

Prachi Mittal and Tim Leinm

uller

Denso Automotive Deutschland GmbH, Germany

Keywords:

Connected Vehicles, Predictive Quality of Service, Seamless Mobility, Multiple Mobile Network Operators.

Abstract:

There are more and more ‘connected’ vehicles on the streets, and they run increasingly more safety critical

applications. To meet the connectivity requirements of these vehicles, network providers need to not only

ensure the quality of service (QoS) but also to predict any upcoming changes in the QoS and inform the

vehicle(s) about it. This concept is called predictive-QoS (P-QoS) and is being heavily discussed in various

organizations, e.g. 3GPP, 5GAA.

To allow a seamless service to vehicles, some issues such as handling multiple mobile network operators

(MNOs), while roaming for example, need to be addressed. For example, if prediction about QoS is available

for multiple MNOs simultaneously for a speciﬁc area, this could be beneﬁcial for the vehicle in selecting an

MNO for further operation in speciﬁc scenarios, e.g. roaming, driving through an area where the current MNO

is predictive to have poor QoS.

In this paper, we introduce an entity, that takes the QoS prediction about multiple MNOs and makes decision

about how to manage the connectivity in a vehicle, e.g. selecting a set of MNOs for further connectivity

including a preference for each, making an “MNO usage timeplan” based on the QoS comparison etc.

1 INTRODUCTION

Connectivity is becoming increasingly important for

automotive applications that serve many functions in

vehicles such as infotainment, safety, automation etc.

The number of vehicles that rely on connectivity is

on a constant rise. Therefore, the reliability as well

as predictability of the whole connectivity ecosystem,

especially the communication technology itself, is a

major consideration for vehicle manufacturers. That

is why, the automotive and telecommunications in-

dustry have been placing emphasis on increasing the

reliability and coverage of the connectivity and addi-

tionally, on predicting any upcoming degradations or

changes to better prepare the vehicles as well as the

entire system.

Communication in vehicles comes in many forms.

This includes the traditional cellular mobile con-

nectivity utilizing 3GPP (3rd Generation Partnership

Project) technologies 2G, 3G and 4G LTE. During the

past decade, vehicles now also include direct connec-

tivity, known as vehicle-to-everything (V2X) which

enables them to communicate with other road partici-

pants (other vehicles, pedestrians etc).

For cellular mobile connectivity, prediction of the

‘quality of service’ (P-QoS) has been an active topic

of research and discussions for past several years,

in the context of 5G and beyond technologies. It

is addressed in public-funded research projects in

EU, such as 5G-NetMobil (Hofmann et al., 2019)

and 5G-CroCo (Hetzer et al., 2019), as well as in

organizations like 5GAA (5G Automotive Associa-

tion) (5GAA, 2020) and 3GPP (3GPP, 2019d; 3GPP,

2019c; 3GPP, 2019a; 3GPP, 2019b).

Typically, a mobile network operator (MNO) has

all control over which services are provided to a net-

work user and how are they provisioned. The roam-

ing scenarios, i.e. when a user of MNO1 is being

served by MNO2, are well deﬁned in terms of techni-

cal (how is the connectivity handled technically in the

respective networks) and business solutions (roaming

charges, if any). But for vehicular network users,

the roaming scenarios are not easily implemented.

Firstly, the movement between the coverage areas of

different MNOs is fast and frequent, and secondly, the

applications can be highly demanding (e.g. in terms of

network resources) or even safety critical.

In this paper, we focus on the problem of manag-

ing P-QoS in an multi-MNO environment. We intro-

duce a novel multi-MNO approach to P-QoS provi-

sion that utilizes QoS predictions about multiple cel-

lular mobile networks (managed by different MNOs)

at the same time. The approach makes decision about

how to manage the connectivity in a vehicle, e.g. se-

Mittal, P. and Leinmüller, T.

Multi-MNO Predictive-QoS for Vehicular Applications.

DOI: 10.5220/0010525606930697

In Proceedings of the 7th International Conference on Vehicle Technology and Intelligent Transport Systems (VEHITS 2021), pages 693-697

ISBN: 978-989-758-513-5

693

lecting a set of MNOs for further connectivity includ-

ing a preference for each, making an “MNO usage

timeplan” based on the QoS comparison etc. This ap-

proach puts the vehicle’s interest at the centre of the

P-QoS provision. Apart from the general problems

of multi-MNO scenarios stated above, providing P-

QoS can be especially tricky. This is because it ad-

ditionally requires knowledge of the cellular network

performance and the capability of making predictions

of high reliability. While one MNO may be capable

enough to assess its own network and make predic-

tions about the QoS, it may not be ﬁt to do the same

for networks operated by other MNOs.

The remainder of this paper is organized as fol-

lowed. In the next section, we brieﬂy discuss related

work. We introduce our Multi-MNO P-QoS approach

in Section 3, followed by a discussion of advantages

and issues in Section 4. Finally, we present conclu-

sions and an outlook to future work in Section 5.

2 RELATED WORK

The P-QoS research has focused on many different

aspects of P-QoS.

Making prediction for end-user mobility is one

approach that can be taken to managing and/or pre-

dicting connectivity (and subsequently QoS) require-

ments for a mobile network. The authors of (Fazio

et al., 2017) demonstrate how this mobility predic-

tion can be used for in-advance bandwidth reserva-

tions in order to achieve service continuity, avoiding

call droppings during active sessions. In a similar vein

of work (Abdalla and Arifﬁn, 2019), Abdalla et al.

discuss a scheme to make more accurate predictions

for end-user mobility and corresponding future cell.

Both works lay the groundwork for one of the most

important elements of P-QoS, namely prediction of

mobility and thus connectivity of an end-user.

The authors of (Hofmann et al., 2019) evaluated

QoS adaptation for platooning scenario. They re-

ported improved planning towards an optimum efﬁ-

ciency for the platooning application in trucks when

they are informed, in advance, of the possible upcom-

ing changes in end-to-end latency and packet delivery

rate.

In (Hincapie. et al., 2019), the authors focused on

the software implementation of P-QoS in a vehicle.

They introduced a collaborative and distributed QoS

manager in the vehicle that seeks to increase and di-

versify the sources of information for P-QoS parame-

ters and the selection of further conﬁguration settings

for the vehicular connectivity.

Torres-Figueroa et al. apply various machine

learning mechanisms to recorded data sets of mul-

tiple MNOs for generating respective QoS predic-

tions (Torres-Figueroa et al., 2020). They deem the

prediction results to not be sufﬁcient for highly reli-

able safety-critical and propose a combined network-

and user-based P-QoS prediction to achieve more re-

liable predictions.

Closest to the scope of our paper is the research

done within the 5G-CroCo project where the re-

searchers deﬁned multiple deployment options for a

P-QoS service (5G-CroCo, 2021). This approach al-

lows the vehicle navigate a multi-MNO environment

but in each deployment option, the vehicle is, in ef-

fect, served by a single MNO at a time.

To the best of our knowledge, no research has been

done on the speciﬁc scenario of managing P-QoS for

multiple MNOs at the same time by a single entity in

order to cherry-pick the best available QoS for a user.

3 MULTI-MNO P-QoS

MANAGEMENT

3.1 Multi-MNO Environment

In a typical connected vehicle scenario, there a num-

ber of vehicles on the roads that have network con-

nectivity provided by a set of MNOs. Different MNO

networks have different capabilities, coverage, data-

trafﬁc at any given time, and different roaming and

service-level agreements (SLAs) with other MNOs.

A multi-MNO environment, such as the one de-

picted in Figure 1, presents a number of challenges

when a connected vehicle is to be served. One such

challenge is to provide a service that runs seamlessly

across network coverage boundaries. This is espe-

cially difﬁcult to achieve for users with high mobil-

ity, such as vehicles traveling at a highway. The dif-

ﬁculty of this challenge is further escalated when the

road segment traveled by the said vehicle crosses over

national boundaries and thus requiring not only inter-

MNO SLAs but also international agreements and/or

requirements.

Through the network connectivity, the vehicles are

not only connected to the serving MNO but also to

other entities in the cloud, e.g. OEM backend, third

party service providers. It is this connectivity that we

intend to leverage for the idea presented in this paper,

through a so-called ”Communication Decision Entity

(CDE)”.

VEHITS 2021 - 7th International Conference on Vehicle Technology and Intelligent Transport Systems

694

MNO1

Traffic Participant

Mobility

MNO2

Traffic Participant

Coverage / National

Boundary

Figure 1: Typical Inter-MNO Mobility Scenario.

Communication Decision Entity (CDE)

Storage

Multi-MNO P-QoS

Data Collection

Common Data

Formatting

Data Analysis

MNO Performance

Comparison

Connectivity

Decision

Vehicle Application

Requirements Data

Collection

Figure 2: Functions of Communication Decision Entity.

3.2 Communication Decision Entity

(CDE)

The “Communication decision entity (CDE)” acts as

aggregator of P-QoS related information from multi-

ple MNOs. It then analyzes the P-QoS data to make

decisions about the network connectivity for a vehicle

or a whole ﬂeet of vehicles.

3.2.1 CDE Functions

The CDE accomplishes its objectives by performing

following functions (shown in Figure 2) –

• Collect P-QoS data about multiple MNOs (multi-

ple embodiments are possible as described later):

this data includes the current provided QoS of

each MNO and any predictions about it, e.g. pre-

dicted degradation in the bandwidth in a speciﬁc

timeslot due to a network maintenance event

• Obtain application requirements of the vehicle:

these requirements can be expressed in terms of

bandwidth, (tolerable) latency, reliability etc

• Bring the prediction data of multiple MNOs into

a common format, if required

• Gather insights about connectivity both spatial

and temporal, e.g. the expected QoS in a certain

geographic area in a certain time-period (say 5 to

6 o’clock in the evening on a weekday) is below

the requirements of a safety application running

on a vehicle

• Consider which parameters to compare and pro-

duce a quantitative output from the comparison.

The parameters are chosen based on applica-

tion(s) requirements in a vehicle. An example

parameter for comparison can be latency and the

quantitative output can be an index, e.g. a 0.8

and 1.1 for MNO1 and MNO2 respectively could

mean that the MNO1 provides a latency of 0.8

times the latency required by the application and

MNO2 provides 1.1 times the latency.

• Make decision about further connectivity, e.g.

– Which MNO to select when roaming to make

sure that an application can continue to run in a

vehicle even when the parent MNO is not avail-

able any more (hence the roaming on another

MNO)

– Which MNO to select when current MNO’s

performance falls below a certain threshold

– A timeplan of usage such as use MNO1 for

the next 10 minutes and then switch to MNO2.

This is to (1) ensure seamless execution of an

application in a vehicle, (2) ensure an efﬁcient

usage of MNO resources, and (3) possibly keep

the cost to a minimum (due to roaming agree-

ments and corresponding charges)

Multi-MNO Predictive-QoS for Vehicular Applications

695

3.2.2 CDE Conﬁgurations

The CDE can reside in the backend (e.g. a third party

service provider or an MNO’s core network) or in the

vehicle itself. Each approach comes with its own set

of advantages.

• Backend-based CDE: can make decisions tailored

to one vehicle but also for an entire ﬂeet (e.g. an

OEM-backend-based CDE making decisions for

all connected vehicles of the said OEM).

• Vehicle-based CDE: offers an advantage of more

efﬁcient decisions that are highly tailored to that

vehicle and has an additional advantage of making

decisions even when there is no connectivity (e.g.

based on historical data).

Furthermore, multiple conﬁgurations for the CDE

are envisioned.

• Conﬁguration 1 (Figure 3): the CDE collects the

QoS prediction data directly from MNOs.

• Conﬁguration 2 (Figure 4): the CDE collects the

data from a 3rd party server which in turn collects

it from the MNOs.

• Conﬁguration 3 (Figure 5): the CDE collects the

data from an MNO which provides not only its

own data but also the data about other MNOs.

Each conﬁguration offers a set of advantages and

disadvantages. For instance, in conﬁguration 1 and 2,

the MNOs provide P-QoS data about their respective

networks to a third party entity (that they trust). This

frees them for performing vehicle related.

3.2.3 CDE Working Example

The following example illustrates functioning of the

CDE as proposed in this paper. A vehicle is driv-

ing automated while the CDE (in the vehicle or in

the backend) receives a prediction about decline in

the QoS of the primary mobile network (let’s call it

MNO1). Based on the predicted QoS of other mo-

bile networks (let’s call them MNO2, MNO3, . . . ),

the CDE determines that based on the connectivity re-

quirements of the automation application running on

the vehicle, it is better to switch to MNO3 after a cer-

tain period of time (say 5 minutes) and then remain

on MNO3 for 15 minutes before switching back to

MNO1.

4 DISCUSSION

The multi-MNO P-QoS management we proposed in

this paper offer multiple advantages. Firstly, the ve-

hicle gets the best service possible based on QoS

MNO 1

MNO 2

…

MNO x

Connectivity Decision

Entity (CDE)

Vehicle

Figure 3: CDE Conﬁguration 1.

QoS Server (3

Party)

MNO 1

MNO 2

…

MNO x

Connectivity Decision

Entity (CDE)

Vehicle

Figure 4: CDE Conﬁguration 2.

MNO1

MNO 2

MNO 3

…

MNO x

Connectivity Decision

Entity (CDE)

Vehicle

Figure 5: CDE Conﬁguration 3.

prediction data of multiple MNOs operating in its

range. This is highly beneﬁcial and even necessary

for safety-critical and/or high-demanding (in terms of

communication network resource usage) applications

such as remote driving. Secondly, there is a natural

’load-balancing’ among multiple MNOs, i.e. instead

of a large number of vehicles overloading one MNO

(e.g. either because they are all contractees of the said

MNO or because the said MNO has SLAs with their

original MNO(s)), the vehicles choose or are assigned

the best possible service at any given time. Thirdly,

connectivity decision can be made not only one single

vehicle but for a whole ﬂeet of vehicles. This is espe-

cially useful if CDE is operated by a vehicle OEM or

a ﬂeet owner such as a logistics company.

The advantages are of course not without their dis-

advantages. There are many roadblocks for the pro-

posed approach. From a technical point of view, the

very ﬁrst demerit is having to process a much larger

amount of data, which in turn may have been for-

matted differently (e.g. formulated using different

KPIs) by different MNOs. Secondly, the proposed

approach may increase the frequency of handovers.

This not only means more complex roaming agree-

ments among the MNOs but also the additional over-

heads in terms of amount of control data and the more

frequent periods of service outage (if any). Concern-

VEHITS 2021 - 7th International Conference on Vehicle Technology and Intelligent Transport Systems

696

ing the business aspects of this approach, for example,

the MNOs need to share their QoS prediction data ei-

ther to each other or to a third party. Additionally,

an MNO might not be forthcoming with our proposed

idea due to data trust or privacy concerns. Even if this

could be resolved by SLAs or similar, the agreements

might be quite complicated, and so will be the service

tariffs and business models.

5 CONCLUSIONS

Making predictions about the performance and hence

the available QoS for a cellular mobile network is a

challenging task because a network serves thousands

of users at a time and the users have varying mobil-

ity and network usage. This task becomes even more

complex if one tries to make similar predictions about

a multitude of such networks because each network

has its own characteristics (e.g. capacity, topology,

maintenance schedule, roaming agreements) as well

as its own set of users. Thus, providing predictive-

QoS in a multi-MNO environment is very challeng-

ing.

In this paper, we propose a novel approach for

multi-MNO P-QoS management that focuses on pro-

viding a vehicle with best possible QoS from mul-

tiple MNOs operating within vehicle’s range. This

approach also ensures a more uniform data-trafﬁc

load distribution among the MNO networks. This

is achieved by introducing a dedicated ’Communica-

tion Decision Entity’ (CDE) for a vehicle or ﬂeet of

vehicles that collects QoS relevant data for multiple

MNOs and makes the connectivity decisions for the

vehicle(s). On one hand, this approach acts in the

best interest of the vehicle(s), on the other hand, it

can alleviate over-loading problems for the involved

MNOs. We envision multiple conﬁgurations for CDE

to suit different network conﬁgurations or business

paradigms.

In future, we plan to provide simulation-based ev-

idence for this approach. We also intend to work

on analyzing the challenges of adopting such an ap-

proach in real-life environment.

REFERENCES

3GPP (2019a). TR 23.287, V16.0.0 (2019-09): Architec-

ture enhancements for 5G System to support Vehicle

to-Everything (V2X) services.

3GPP (2019b). TR 23.288, V16.1.0 (2019-09): Architec-

ture enhancements for 5G System to support network

data analytics service.

3GPP (2019c). TR 23.786, V16.1.0 (2019-06): Study on

architecture enhancements for 3GPP support of ad-

vanced V2X services.

3GPP (2019d). TS 22.186, V16.2.0 (2019-06): Service re-

quirements for enhanced V2X scenarios.

5G-CroCo (2021). Deliverable D3.2 - Intermediate E2E,

MEC and Positioning Architecture. Technical report,

5G-CroCo.

5GAA (2020). 5GAA White paper: Making 5G Proactive

and Predictive for the Automotive Industry. Accessed:

2021-02-04.

Abdalla, S. E. and Arifﬁn, S. H. S. (2019). The inte-

grated history based prediction for future networks. In

2019 International Conference on Computer, Control,

Electrical, and Electronics Engineering (ICCCEEE),

pages 1–5.

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

diction and qos enhancement in new generation cel-

lular networks with mobile hosts: A survey on dif-

ferent protocols and conventional/unconventional ap-

proaches. IEEE Communications Surveys Tutorials,

19(3):1822–1841.

Hetzer, D., Muehleisen, M., Kousaridas, A., and Alonso-

Zarate, J. (2019). 5G Connected and Automated Driv-

ing: Use Cases and Technologies in Cross-border En-

vironments. In Proceedings of European Conference

on Networks and Communications (EuCNC), pages

78–82.

Hincapie., D., Saad., A., and Jiru., J. (2019). Collabora-

tive and Distributed QoS Monitoring and Prediction:

A Heterogeneous Link Layer Concept towards always

Resilient V2X Communication. In Proceedings of the

5th International Conference on Vehicle Technology

and Intelligent Transport Systems - Volume 1: VE-

HITS,, pages 601–608. INSTICC, SciTePress.

Hofmann, F., Mahdi, A. H., Brahmi, N., Wegmann, B.,

Muehleisen, M., Petry, F., Mudriievskyi, S., El As-

saad, A., Jornod, G., Franchi, N., and Fettweis, G.

(2019). 5G NetMobil: Pathways Towards Tactile Con-

nected Driving. In Proceedings of 2nd IEEE 5G World

Forum (5GWF), pages 114–119.

Torres-Figueroa, L., Schepker, H. F., and Jiru, J. (2020).

QoS Evaluation and Prediction for C-V2X Commu-

nication in Commercially-Deployed LTE and Mobile

Edge Networks. In Proceedings of 91st IEEE Vehicu-

lar Technology Conference, VTC2020-Spring.

Multi-MNO Predictive-QoS for Vehicular Applications

697