Future wireless and mobile communication systems should provide an appropriate
level of QoE for heterogeneous services. In recent years, technical QoS has been an
important research topic. Current telecommunications systems are mostly driven
by QoS mechanisms, which include a comprehensive set of di_erent parameters.
In IP networks the QoS is provided through the use of various mechanisms, such
as integrated and di_erentiated services. These mechanisms are mostly based on
network parameters, such as delay transfer, delay variation (jitter), packet loss and
bandwidth. The aforementioned mechanisms are rigid, providing little dynamics
and are typically based on the objective QoS metrics. Parameters optimized by
such QoS mechanisms are measured at the packet level and do not directly contribute
to the user perception of quality. Furthermore, these mechanisms do not
include the complete end-to-end communication path, user perception of quality,
previous user's experience, user's expectations and various psychological, sociological
and demographic factors. Currently, it is also not clear to what degree the
usage of di_erent optimization algorithms and QoS mechanisms is e_cient in terms
of perceived improvements and experience from the user's perspective. Therefore,
the ultimate goal of wireless network resource allocation provisioning and of design
of e_cient services is to improve the subjective user's perception. This concept is
very important for end users and internet service providers.
Many de_nitions and specialized models have been proposed in the literature
that provide mapping between QoS and QoE for time-critical real time services. On
the other hand, only few studies exist for the best e_ort interactive applications.
The main problem is that QoE is a complex concept consisting of many subjective
and objective parameters involving di_erent psychological and technical factors that
are not always well understood. Moreover, currently known contextual factors and
mathematical models, which map network QoS parameters and user perception of
quality, do no adequately describe the user's QoE.
In the _rst part of the thesis we focused on the theoretical and experimental
study of developing a connection between the user perception of quality, di_erent
contextual factors and the objective parameters QoS for interactive applications.
One of the most commonly used interactive services in mobile systems is web browsing,
which is likely also the most varied application and is expected to be the
dominant service in future broadband access networks. Focusing on web browsing
can provide valuable insight into a wide range of use cases found in the majority
of modern interactive applications. We conducted two di_erent experiments for
web browsing scenarios, which includes the preparation and planning of the experiments,
selection of the test participants and analysis of the results. The objective
of the _rst experiment is to investigate and mathematically model the inuence of
the network QoS parameters and the previous user experience on the current user
perception of quality. We showed that di_erent response times and previous user
experience signi_cantly a_ect the user perception of web browsing quality. From
the collected data of the _rst experiment we proposed two mathematical models for web browsing quality estimation: the logarithmic model (i.e., psychophysical theory
Weber-Fechner) and exponential model. Furthermore, from the experimental
results we determined the subjective threshold QoE above which users are satis_ed
and estimated the required amount of resources needed to achieve a certain level of
satisfaction.
The result of the second experiment is a novel Markov model QoE. For this
purpose, the probability theory was employed, where the elements of the Markov
model QoE were mathematically determined using a probability theory of discrete
time stochastic process. From the experimental data we determined one-step transition
probability matrix, which represents the inuence of previous user experience
on current perception of quality. The experimental study shows that Markov model
QoE presents a deeper understanding and an important insight into the user
perception of quality for interactive applications. The presented QoE model can
signi_cantly improve the level of user QoE with services in telecommunications systems
if implemented within advanced system design, resource allocation algorithms
and quality assurance procedures.
Wireless communication systems use di_erent channel access methods for shared
medium networks, such as TDMA, FDMA, CDMA, SDMA and OFDMA. OFDM
is a promising technique, which is based on the concept of a multicarrier transmission.
Assuring quality in MU-OFDM systems represents one of the most important
challenges mainly due to complex communication channel that signi_cantly impacts
the transmission quality.
The assurance of quality in the next generation wireless and mobile networks
is very interesting research topic. Many di_erent dynamic QoS resource allocation
mechanisms have been proposed for wide range of services. Optimizing resources
based on the objective QoS metrics represents the main problem of the current
resource allocation methods, since it is not clear to what degree such resource allocation
is e_cient in terms of the perceived improvement and experience from the
user's perspective.
In the second part of the thesis we introduced and developed a novel quality
assurance approach for the downlink of heterogeneous MU-OFDM systems. The
proposed concept provides a novel paradigm shift to a user-centric approach, which
shows great potential for improving user's perceived quality. The main novelty is
explicitly addressing the user's perceived quality for interactive and time-critical
real-time services in terms of QoE. We de_ned the optimization problem for the
MU-OFDM system with the objective of optimizing the human perception of quality
by using the mathematical QoE models. Based on the de_ned optimization
problem we proposed two novel user-oriented low-complexity QoE-aware radio resource
allocation (RRA) algorithms, where resources are allocated by taking into
account the speci_c requirements of each heterogeneous service data ow in terms
of QoE. In the _rst algorithm (i.e., Max-Min MOS), we consider a utility function
aimed at the maximization of the minimum MOS experienced by the users. The
Max-Min MOS algorithm presents the suboptimal solution of the de_ned optimization
problem. The proposed suboptimal algorithm reaches beyond the state-of-the art solutions by presenting practical joint subcarrier and power allocation method
for QoE optimization. The second algorithm (i.e., QoE-OFDM) introduces e_cient
balance between the level of QoE and the spectral e_ciency of the OFDMA system.
Moreover, it also eliminates some issues of the Max-Min MOS algorithm.
The theoretical framework is validated through numerical simulations. The
simulation results of the proposed algorithms are compared with the other
state-of-the-art methods. The results clearly present the e_ciency of the proposed
suboptimal algorithm, where the near to optimal solution is obtained with low
complexity. The numerical results also show that the proposed algorithms achieve
signi_cant increase in the level of QoE for heterogeneous services compared to other
existing methods. We showed that by incorporating the application parameters
and functions describing the user's perception of quality for heterogeneous services
into the RRA algorithms, adaptable allocation of system resources, precise control
of user data rates and appropriate level of QoE can be achieved for each user of
the OFDMA system.
|
