You are on page 1of 7

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/278675919

Quality of Experience in the Multimedia Internet of Things: Definition and


practical use-cases

Conference Paper · June 2015


DOI: 10.1109/ICCW.2015.7247433

CITATIONS READS

41 1,456

2 authors:

Alessandro Floris Luigi Atzori


Università degli studi di Cagliari Università degli studi di Cagliari
33 PUBLICATIONS   411 CITATIONS    256 PUBLICATIONS   14,903 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

CFP: Special Issue "Big Data Analytics in the Internet of Things: Architectures, Emerging Trends and Open Challenges" View project

Smart State Estimation exploiting IoT architectures View project

All content following this page was uploaded by Alessandro Floris on 18 June 2015.

The user has requested enhancement of the downloaded file.


Quality of Experience in the Multimedia
Internet of Things: definition and practical use-
cases
Alessandro Floris, Luigi Atzori
Department of Electrical and Electronic Engineering (DIEE), University of Cagliari
Cagliari, 09123, Italy
alessandro.floris@diee.unica.it, l.atzori@ieee.org

architecture is analysed to understand which QoE influence


Abstract—In this paper, a first approach for evaluating the factors have to be considered in relevant application scenarios.
Quality of Experience (QoE) for IoT (Internet of Things) Secondly, we introduce the concept of Multimedia IoT (MIoT)
applications is presented. Firstly, a layered IoT architecture is and define a layered QoE model aimed at evaluating and
analysed to understand which QoE influence factors have to be
combining the contributions of each influence factor to
considered in relevant application scenarios. Secondly, we
introduce the concept of Multimedia IoT (MIoT) and define a estimate the overall QoE in MIoT applications. Finally, we
layered QoE model aimed at evaluating and combining the present a vehicular MIoT application that has been used to
contributions of each influence factor to estimate the overall QoE conduct subjective quality assessments to verify the
in MIoT applications. Finally, we present a vehicular MIoT performance of the proposed approach.
application that has been used to conduct subjective quality In section II, a background on IoT is provided. In section III
assessments to verify the performance of the proposed approach.
the Multimedia Internet of Things (MIoT) is defined, whereas
Index Terms—Quality of Experience, Internet of Things.
in section IV the proposed layered QoE model for MIoT
I. INTRODUCTION applications is presented. In section V, a MIoT vehicle
application for conducting experimental evaluations is
T HE Internet of Things (IoT) is a world-wide network of
interconnected objects uniquely addressable, based on
standard communication protocols. The IoT extends the
proposed. Section VI and VII discuss the experimental
evaluations and relevant results, respectively. Finally, section
conventional concept of the Internet as an infrastructure VIII concludes the paper.
network reaching out to end-users’ terminals, to a concept of II. BACKGROUND
interconnected objects forming pervasive computing
environments. IoT evolved in the last years, involving at first The Internet of Things (IoT) refers to a network of
Radio-Frequency IDentification (RFID) technology, then interconnected objects which are able to acquire information
from the physical world and to make this information
sensors, actuators and Wireless Sensor Networks (WSN), up
available on the Internet. It is possible to basically identify
to the connection of any object, anywhere and anytime.
three generations of IoT on the basis of the involved
As each telecommunications service, also IoT applications
technologies [1]. The first generation was developed around
have to be evaluated, in order to measure the quality provided the RFID tags, which were typically used for monitoring,
to the end-users who use the application. While a few studies logistics and tracking applications. The second generation was
proposed QoS evaluation models for IoT applications, to the enhanced by sensors and actuators, which permitted to acquire
best of the authors’ knowledge no studies take on the QoE various physical characteristics from the real world and
evaluation of IoT applications. In telecommunications, the brought to the birth of WSNs [2]. The third and present IoT
Quality of Experience (QoE) aims at evaluating how end-users generation is mostly developed around the association of a
subjectively perceive the quality of an application or a service. Virtual Object (VO) to each Real World Object (RWO),
Being user-centric, the QoE provides a more holistic which is then part of the IoT, with the aim of virtualizing its
understanding of the system’s influence factors than functionalities. Everyday physical objects become “smart”
technology-centric measures such as the Quality of Service objects and are integrated within the global resulting cyber-
(QoS). physical infrastructure [3]. Furthermore, the information and
In this paper, a first approach for evaluating the QoE for application storage and computing have been moved to the
IoT applications is presented. Firstly, a layered IoT cloud.
A. IoT Architecture
This work has been partially funded by the Regione Autonoma della Currently, there is no reference IoT architecture common to
Sardegna, P.O.R. FESR 2007-2013 6.2.1.a, within the project SocialMobility the different proposed platforms. Still, one representative
(C.U.P. F25C10001420008).
model for the third generation is the iCore solution drafted in packet switched unreliable link by direct current motors along
[4]. From this a five-layer model can be extracted: with gear boxes that enable movement in several degrees of
1. Physical and communication layer: refers to the freedom. The experiment required users to direct a fixed laser
sensing devices of the IoT that are the RWOs; attached to the fixed arm’s grabber towards a set of targets
2. Virtualization layer: creates the VOs which virtualize located on the vertical plane at fixed distance from the arm.
the functionalities of the associated RWOs; The experimental factors were the average one way delay, the
3. Aggregation layer: has the task to combine various packet loss, and the number of degrees of freedom. From the
VOs in order to create Composite Virtual Objects subjective quality assessment results the authors defined a
(CVOs) capable of providing a determined service QoE model that estimates the QoE in function of the
that a single VO cannot accomplish; considered experimental parameters. In [7], the authors
4. Servitization layer: creates a link between a product defined the Cognitive Internet of Things (CIoT), a new
and a set of remote functionalities that augment the network paradigm where physical and virtual things or objects
capabilities and the properties of the product; are interconnected and behave as agents, with minimum
5. Applications layer: accounts for the various IoT human intervention. The CIoT framework includes a service
applications, exporting all the system’s evaluation layer to evaluate the services and feedback the
functionalities to the final user. evaluation result to the cognition process. This layer has to
B. IoT Applications measure three different qualities: i) Quality of Data (QoD),
A huge number of IoT applications have been developed for which aims to evaluate the quality of sensed data, the process
different objectives and services. Generally, as in [1, 2, 3, 5], of data acquiring and the possible data distribution at the
IoT applications are classified on the basis of the domain of perception/sensing stage; ii) Quality of Information (QoI),
utilization. The main domains are: Healthcare, Personal and which concerns the information that meets a specific user’s
social, Smart home/Smart building, Smart City, Smart grid need at a specific time, place, physical location, and social
and smart metering, Smart environment, Transportation, Smart setting, climbing the protocol stack through the applications
business/inventory and logistics, Security and surveillance. layer to a postulated user layer; iii) Quality of Experience
For some of these applications the QoE management is not (QoE), which is evaluated from factors in four levels: Access,
of interest, as the user is not always the recipient of the Communication, Computation, and Application.
services. Indeed, if the application output is received by
another system, then the performance of the application can be
evaluated through a classical QoS-based approach. In fact, the
objective of the QoE modelling activity is to have total control
of the delivered service as perceived by the end-user to
optimize the quality while controlling the utilized resources of
each system component (in the IoT platform for the interest of
this work). Additionally, in this work we are interested in the
evaluation of the QoE for applications where the multimedia
content has a major role.
QoE requirements can be very different with respect to the
considered IoT application domain; furthermore, QoE
requirements can also be different among IoT applications
belonging to the same IoT domain. This is also shown in Fig.
1, where we show the distribution of IoT applications which
are representative for each IoT domain. These are placed on
the picture on the basis of the type of information they manage Fig. 1 Classification of IoT applications on the basis of the type of
(multimedia or non-multimedia content), and of the type of information they manage (multimedia or non-multimedia content) and of the
end-user orientation (person or system oriented). Person type of recipient (person or system oriented).
oriented IoT application means that people have a While [7] considered both QoS and QoE parameters, [8, 9]
fundamental rule on the evaluation of the IoT application, only focused on QoS evaluation. In fact, the authors defined
since they are the end-users who benefit from the information an IoT architecture composed of three layers: sensing,
and contents provided by the IoT application. On the other networking and application; in each layer, a QoS monitoring
hand, system oriented IoT application refers to those module manages the resource allocation in function of the
applications which automatically acquire, control and manage measured QoS metrics. In particular, the sensing layer
data in order to do specific tasks and take specific decisions. accounts for the QoS parameters of the sensing networks and
In these applications human intervention and evaluation are devices such as information accuracy, sensing precision,
not essential. energy consumption, etc. The network layer accounts for the
C. Quality evaluation in the IoT QoS parameters of the networks through which the IoT
Not many studies exist concerning QoE evaluation for the services are provided, such as bandwidth, delay, throughput,
IoT. In [6], the authors focused on the perceived quality in coverage, etc. Finally, in the application layer the QoS metrics
actuators connected to the IoT. They developed a test-bed of web services are considered, such as the service perform
which consisted of a electro-mechanical arm controlled over a cost, the perform time, the load, the reliability, etc. Although
QoS parameters are important on the performance evaluation has to determine if the person recorded by the camera is
of an IoT platform, they should be considered in function of authorized to enter to that place. Once the software identifies
the quality perceived by the end-user and not only as a the person, a command is sent to the door actuator to open the
fulfilment of Service Level Agreements. door. Otherwise the door stays closed and the access is denied.
Fig. 2 (green) illustrates the second scenario. Various IoT
III. MULTIMEDIA INTERNET OF THINGS
objects measure some medical parameters of a patient (e.g.
The emerging categories of IoT objects tend to be mobile, temperature, pulse, pressure, etc.) which are collected by the
multi-sensorial and smart, such as wearable sensors, IoT platform. This application has the task to present the status
smartphones, and smart vehicles, bringing also to an increase of the patient in a multimedia way, for example by using
of multimedia content in the IoT. Multimedia content refers to graphs, animation, alarms, etc. Finally, Fig. 2 (orange) shows
a combination of two or more different media contents such as the last case. Some surveillance cameras record images and
text, audio, image, video, etc. In section II we defined the IoT audio of a place. This multimedia information is collected by
as a network of interconnected objects which are able of an IoT application and presented in a multimedia way (videos,
acquiring some information from the physical world and make audios) in order to provide a remote security control service.
this information available on the Internet. Consequently,
Multimedia IoT (MIoT) can be considered as a “network of IV. LAYERED MODELLING FOR IOT APPLICATIONS
interconnected objects capable to acquire multimedia contents In the Qualinet White Paper on Definitions of Quality of
from the real world and/or present information in a Experience [10], the QoE is defined as “the degree of delight
multimedia way. We also define multimedia objects as or annoyance of the user of an application or service. It
“objects capable to acquire multimedia contents from the results from the fulfilment of his or her expectations with
physical world, being equipped with multimedia devices such respect to the utility and/or enjoyment of the application or
as cameras and microphones”. service in the light of the user’s personality and current state”.
In the MIoT, three scenarios can be distinguished on the Accordingly, in the ITU-T Rec. P.10 [11], QoE is defined as
basis of the use of the multimedia content: “the overall acceptability of an application or service, as
1. Multimedia as IoT input: the multimedia content is perceived subjectively by the end user.”
acquired by multimedia objects and it is used by an IoT
Such definitions of QoE are valid for general
application to provide a determined service;
telecommunications applications and services, therefore they
2. Multimedia as IoT output: IoT objects acquire signals,
can be also extended to IoT applications. However, currently
data and information (non-multimedia content) that are
presented in a multimedia way by an IoT application; there is no reference model for evaluating the QoE of IoT
3. Multimedia as IoT input and output: the multimedia applications, probably due to the lack of a reference IoT
content is acquired by multimedia objects and it is architecture and to the different requirements of IoT
presented in a multimedia way by an IoT application. applications belonging to different domains of utilization.
Therefore, on the basis of the MIoT definition presented in
section III, we propose a layered QoE model capable of
evaluating the QoE of MIoT applications in the scenarios of
multimedia as IoT output and multimedia as IoT input and
output, in which people are involved as the end-users of the
multimedia content.
Conceptually, we mainly based on the layered QoE model
proposed in [12] for evaluating the QoE of a multimedia
communication system. In fact, layered approaches are widely
used in networking for their functionality, such as the TCP/IP
and OSI networking models. Within each layer, a well-defined
set of network services are provided, which are in turn
implemented utilizing the services provided by the previous
layers. At the lower layers, services tend to be simple and
efficient. As successively higher layers are considered,
functions become gradually more complex. Following this
Fig. 2 Orange: scenario of an IoT application that presents in a multimedia layered approach, each layer must be defined together with its
way information acquired by multimedia objects. Red: scenario of an IoT area of competence. Furthermore, each layer must be able to
application using information acquired by multimedia objects. Green: scenario be combined with its upper and lower layers in order to build a
of an IoT application that presents in a multimedia way information acquired
by IoT objects (non-multimedia). model in which the outcome of a layer can be interpreted and
The first scenario is represented by the use-cases depicted then gradually enhanced by higher layers. For such reason,
in Fig. 2 (red), where a camera records some images of people interfaces between layers should be accurately defined.
who want to enter in a place where the entrance is allowed Therefore, each layer of the proposed QoE model analyses
only to authorized persons. The images recorded by the a specific QoE domain (set of factors), so that the overall
camera are then sent to an IoT platform where an application quality can be computed as a combination of all domains. The
is deployed that consists in an identification software which proposed model is meant to be as general as possible, so to be
valid for, or at least adaptable to, any scenario. perceived by the end-users is also influenced by the way the
contents are presented to the user, the application interface, the
degree of interactivity and the usability of the application.
E. Context layer
Finally, the upper layer accounts for the Context of use,
which concerns many different factors as the device on which
the application is used (technical), the people with which the
end-user consumes the application (social), the application
costs (business), the place where the application is used
(environmental), etc.
V. IOT VEHICLE APPLICATION
In order to conduct experimental evaluations, an IoT
Fig. 3 Layered QoE model for MIoT applications. vehicle application for remote tutoring of driving school
The proposed layered QoE model is shown in Fig. 3. It is lessons has been developed. The objective of this application
composed of five layers: the Physical devices layer, the is to show in a multimedia way and in real-time the state and
Network layer, the Combination layer, the Application layer, the position of the vehicles used for driving lessons, as well as
and finally the Context layer. a video of the roads travelled by these vehicles. In this way,
the vehicles used by the driving instructors for teaching
A. Physical devices layer driving lessons to the students can be remotely controlled.
The Physical devices layer accounts for the quality provided Fig. 4 shows the framework developed for collecting the
by the the physical objects. Such quality depends on objective vehicle information needed by the IoT vehicle application,
measures of the objects’ performance that influence the QoE which is composed of an Arduino Mega 2560 [13], the Telit
perceived by the user. On the basis of the type of object, these UE910-EUR and SL869 modules [14], and a camera. The
measures can include QoS and/or Quality of Data (QoD) Arduino Mega 2560 is an open-source electronics platform
parameters. For example, the performance of a GPS receiver based on easy-to-use hardware and software. It is connected to
that acquires the position of an object or a person can be the On-Board Diagnostic interface (OBD-II) of the vehicle by
determined by measuring the accuracy and precision of the using an OBD-II adapter. In this way, the Arduino is able to
acquired position, which is the QoD. On the other hand, the acquire vehicle parameters, such as vehicle speed, engine
quality of a video sequence recorded by a camera can be revolutions per minute, oil temperature, etc. The Telit SL869
evaluated by both QoD and QoS parameters. For instance, the module [14] is provided with a Global Positioning System
QoD accounts for the image resolution whereas the QoS (GPS), and it acquires the position coordinates of the vehicle
considers the video coding distortion. with current date and time. The Telit UE910-EUR module
receives the vehicles parameters from the Arduino via Serial
B. Network layer
Peripheral Interface (SPI), and the position coordinates of the
The Network layer accounts for the performance of the vehicle from the SL869 via serial interface; therefore, being
network infrastructure on which the IoT platform is based. In provided with 3G connection, it sends the received vehicle
this layer, typical QoS parameters capable of measuring the information to the cloud where the IoT platform is developed.
performance of information transmission are considered, such The Telit UE910-EUR module is not equipped with sensors,
as delay, packet loss, jitter, etc. The influence of these but it has the rule of network interface through which the
parameters on the QoE is taken into account. Arduino and the Telit SL869 module are able to send their
C. Combination layer acquired parameters to the IoT platform. Simultaneously, the
The Combination layer includes the Virtual Objects (VO) camera installed into the vehicle cabin records and sends to
and the Composite Virtual Objects (CVO), which are in the the IoT platform the video of the road travelled by the vehicle.
cloud. Each VO communicates directly with the physical The IoT platform is in the cloud and, for this application,
object that it virtualizes and evaluates its quality; such quality includes three VOs and a CVO. Each VO has the task to
depends on the object’s characteristics and should be related to acquire the data provided by the physical object that it
virtualizes and to evaluate its quality. The CVO receives the
the QoE perceived by the user through a mathematical model.
data acquired by the VOs, and analyses and synchronizes the
The CVO is composed of various VOs with the aim of
vehicle and camera data (the images recorded by the camera
combining their functionalities to provide a determined are tagged with date and time), which are then shown in a
service. The CVO is provided with a mathematical model multimedia way to the users on a web application. Precisely,
which combines and weighs the qualities provided by each the web IoT application shows in a web page a video of the
VO in order to evaluate the overall QoE provided to the user. road travelled by the vehicle, a map in which is shown the
D. Application layer vehicle position (highlighted with a moving red circle), and
The Application layer considers the QoE provided by MIoT the speed and revolutions per minute (rpm) of the vehicle. The
applications in terms of control, interactivity, presentation and CVO has also the task to estimate and control the overall
usability. In fact, independently from the quality of the quality provided by the application. It must be able to define a
multimedia content provided by the application, the quality mathematical model which weighs the single contributions
provided by the VOs and estimates the overall QoE provided instance, the speed and rpm indicators suddenly show wrong
to the end user. values whereas the vehicle position is indicated out of the
road. These corruptions simulate data inaccuracy of the
sensing objects. Finally, the speed and rpm values were
presented to the users both textually and graphically (with
gauges).
In the proposed assessment, 24 people, 20 males and 4
females, were asked to rate the 12 TCs. The assessment started
with a pre-test phase, during which the user was provided with
short written instructions and with a brief detailed oral
explanation. Furthermore, some TCs were shown in order to
give each user an overview of the video qualities and data
corruptions involved in the experiment. The device employed
for the subjective quality assessment was a notebook with a
LCD display with a resolution of 1440x900 pixels.
For subjective tests, the specifications given in [15] have
been adopted. Specifically, the single-stimulus Absolute
Category Rating (ACR) method has been adopted, which
Fig. 4 Framework of the IoT vehicle application for remote tutoring of driving
school lessons. consists in presenting the TCs one at a time and allowing for
their independent evaluation on a category scale. The
VI. EXPERIMENTAL EVALUATIONS following five-level ITU scale for rating overall quality has
The experimental evaluations consist on conducting a been used: 5 stars (Excellent), 4 stars (Good), 3 stars (Fair), 2
subjective quality assessment for evaluating the QoE stars (Poor), 1 star (Bad). The parameters which the user had
perceived by the users using the IoT application for remote to consider in the subjective evaluations were: i) the video
tutoring of driving school lessons. The data collected for the quality; ii) the synchronization and accuracy of the vehicle
experimental evaluations were the video sequences of the road position and of the speed and rpm values with respect to the
travelled by the vehicle, and the GPS, speed and rpm data of video; iii) the overall visualization of the vehicle information.
the vehicle. Table I summarizes the Test Conditions (TCs) The assessment has been carried out in a 15 minute-long
which have been chosen to conduct the subjective assessment. session. The IoT application TCs were displayed on the
TABLE I TEST CONDITIONS (TCS). Google Chrome browser in full-screen mode while each
Video bitrate Speed/RPM Speed/RPM subject was wearing earphones. Each TC lasted for 40s. A
TC Video Map
(kbps) values visualization rating page was automatically displayed after the end of each
1 Video 1 1500 Sync Sync Gauge TC, where each subject had to select a discrete value between
2 Video 2 1500 Sync Sync Text 1 and 5 indicating the overall quality of the IoT application
3 Video 3 1500 Desync Desync Gauge
just shown. The voting phase was not time-limited and after
4 Video 4 1500 Desync Desync Text
each vote the subjects were asked to confirm their choice.
5 Video 5 1500 Inaccurate Inaccurate Gauge
Upon confirmation, a ten-second pause was imposed, then the
6 Video 6 1500 Inaccurate Inaccurate Text
7 Video 1 700 Sync Sync Gauge
next TC would be automatically activated. Such automatic
8 Video 2 700 Sync Sync Text procedure was iterated for all the 12 TCs. The presentation
9 Video 3 700 Desync Desync Gauge order of the TCs has been randomized for each subject.
10 Video 4 700 Desync Desync Text
VII. EXPERIMENTAL RESULTS
11 Video 5 700 Inaccurate Inaccurate Gauge
12 Video 6 700 Inaccurate Inaccurate Text Mean Opinion Scores (MOS) have been computed for each
TC, together with the 95% confidence interval. MOS results
The parameters chosen to create the TCs are the video
show that video quality does not influence very much the QoE
bitrate, the synchronization and accuracy of the vehicle data, of the IoT application. In fact, being equal the other
and the way the speed and rpm values are presented to the conditions, the MOS concerning TCs with low and high video
user. The video sequences, with resolution of 640x480 pixels, quality have more or less the same value, as shown in Fig. 5.
were encoded at two different bitrates: 700kbps for low video Also data presentation does not influence very much the QoE.
quality and 1.5Mbps for high video quality. TCs with Furthermore, MOS results show that the most annoying TCs
desynchronized data consist in showing desynchronized are those where wrong vehicle data are displayed. Then, the
(delayed with delay of 3s) values of GPS, speed and rpm data accuracy of the data is the most relevant parameter considered
with respect to the video shown on the IoT application. These by the end-users. Also TCs with desynchronized data annoy
corruptions simulate network delay errors and are introduced the end-users, but to a lesser extent. The most evident result is
in a way the user can notice them, for instance when the that from this IoT application the end-users expect the
vehicle is starting or stopping. TCs with inaccurate data displaying of accurate and synchronized vehicle data;
consist in showing randomly wrong values of GPS, speed and otherwise they were annoyed. On the other hand, video quality
rpm data; precisely, the 25% of the values were wrong. For and data presentation have not a great influence on the QoE.
estimate the QoE perceived by the users on the basis of the
QoS and QoD parameters of the objects virtualized by the
VOs and of the QoS of the communication network. In this
case, the CVO can notice that the Rate parameter has
negligible influence on the QoE, and can decide to delete its
contribution from the QoE model to reduce the computation
complexity.
TABLE II VALUES OF THE COEFFICIENTS FOR BOTH THE LINEAR AND
NONLINEAR REGRESSION IN (1)-(3).
Coefficient )*+ ,*+ )-+ ,-+ ).+ ,.+
Value 0.0000 2.4306 0.0521 -2.3438 -0.1528 2.5833
Coefficient )*/+ ,*/+ )-/+ ,-/+ )./+ ,./+
Fig. 5 MOS results with 95% CI comparing low and high video qualities Value 1.0000 2.3276 1.0233 0.2752 0.9644 2.3615
being equal the other conditions.
In order to estimate the quality perceived by the user when VIII. CONCLUSIONS
using the IoT application, the influence of the video coding
In this paper, a first approach for evaluating the QoE for IoT
bitrate (Rate), the data accuracy (ACC) and the network delay applications has been presented. We introduced the concept of
(D) on the QoE have been investigated. The subjective ratings Multimedia Internet of Things (MIoT) and defined a layered
were divided in two datasets: a training dataset composed of QoE model aimed at evaluating and combining the
the ratings provided by the first 12 subjects and a validation contributions of each influence factor to estimate the overall
dataset composed of the ratings provided by the last 12 QoE in MIoT applications. To verify the performance of the
subjects. From these datasets we computed the MOST and the proposed QoE model, we developed a vehicular MIoT
MOSV values, respectively. application that has been used to conduct subjective quality
Thereafter, a linear (l) and nonlinear (nl) regression assessments. The QoE metrics derived from the layered QoE
between the considered corruption parameters and the MOST model are highly correlated with subjective results and allow
results have been computed. The coefficients obtained by to estimate and control the QoE of MIoT applications.
computing these regressions are summarized in Table II, and REFERENCES
are valid for the equations in (1)-(3). [1] L. Atzori, A. Iera, and G. Morabito, “The Internet of Things: A survey,”
Computer Networks, vol.54, no.15, pp.2787-2805, Oct. 2010.
 
  ⋅      
  ⋅  
 (1) [2] J. Gubbi, R. Buyya, S. Marusic, M. Palaniswami, “Internet of Things
(IoT): A vision, architectural elements, and future directions,” Future
 
 ⋅     
 ⋅ 
 2
Generation Computer Systems, vol.29, no.7, pp.1645-1660, 2013.
[3] D. Miorandi, S. Sicari, F. De Pellegrini, and I. Chlamtac, “Internet of
QoE D
a" ⋅ D  K " QoE$ D
K "$ ⋅ a"$
% 3
things: Vision, applications and research challenges,” Ad Hoc Networks,
vol.10, no.7, pp.1497-1516, Sept. 2012.
The equations in (1)-(3) are the models which describe the
[4] R. Minerva et al., iCore, D2.3 Architecture Reference Model, WP2 –
relation between the corruptions and the QoE. By analysing Cognitive Management and Control Framework for IoT, June 2013.
the subjective results we concluded that video quality does not [5] D. Singh, G. Tripathi, and A.J. Jara, “A survey of Internet-of-Things:
influence the QoE. The data regression confirms this result, in Future Vision, Architecture, Challenges and Services,” IEEE World
Forum on Internet of Things (WF-IoT), 2014, Seoul, KR, pp. 287-292.
fact the linear and nonlinear coefficient values for the Rate
[6] J. Arauz, and T. Fynn-Cudjoe, “Actuator Quality in the Internet of
parameter are 0 and 1, respectively. The camera VO must Things,” in Proc. SECON 2013, New Orleans, LA, USA, pp. 34-42.
know the QoE(Rate) metric that accounts for the video [7] Q. Wu, G. Ding, Y. Xu, S. Feng, Z. Du, J. Wang, and K. Long,
quality, whereas the GPS and the vehicle VOs must know the “Cognitive Internet of Things: A New Paradigm Beyond Connection,”
QoE(ACC) and QoE(D) metrics, which take into account the IEEE Internet of Things Journal, vol.1, no.2, pp.129-143, April 2014.
[8] L. Li, S. Li, and S. Zhao, “QoS-Aware Scheduling of Services-Oriented
influence of the data accuracy of the physical devices and of Internet of Things,” IEEE Trans. on Industrial Informatics, vol.10, no.2,
the network delay, respectively. The CVO receives those QoE pp.1947-1505, Feb. 2014.
models from the VOs and defines a composite QoE model [9] R. Duan, X. Chen, and T. Xing, “A QoS Architecture for IoT,” in Proc.
including all the contributions provided by the single of iThings/CPSCom, 2011, Dalian, CN pp. 717-720.
[10] Qualinet European Network on Quality of Experience in Multimedia
components. In this case, the linear and nonlinear composite Systems and Services, “Qualinet White Paper on Definitions of Quality
QoE models are as follows: of Experience,” Version 1.2, 2013.
[11] Recommendation ITU-T P.10/G.100, Amendment 2, “Vocabulary for
   ⋅       ⋅     " ⋅ '  " (4) performance and quality of service,” 2008.
[12] A. Floris, L. Atzori, and G. Ginesu, “Addressing un-Interoperability
   ⋅  
 ⋅  ⋅ 
 ⋅ " ⋅ "
( (5) Issues in QoE Models: Is a Layered Modelling Effective?,” in Proc. ICC
2014, Sidney, AU, pp. 563-568.
Afterwards, the Pearson correlation values achieved [13] http://www.arduino.cc/
between the MOSV values and the quality indexes defined in [14] http://www.telit.com
(4)-(5) have been computed and are 0.91 and 0.86, [15] ITU-T Recommendation P.910, “Subjective video quality assessment
methods for multimedia applications,” 1999.
respectively. The same correlation values are achieved also
without considering the influence of the Rate, as expected.
Being provided with this QoE model, the CVO is able to

View publication stats

You might also like