|dc.description.abstract||In recent years, there has been a drastic growth in wireless traffic, and in the near future, a
majority of wireless traffic will be video-related. How to support the increasing demand of radio
resources from the bandwidth-hungry services has attracted intensive research interests from both
academia and industrial areas. Milli-MeterWave (mmWave) technology and Cognitive Radio (CR)
technology was recently proposed to enhance the wireless network capacity.
However, some challenges are needed to be addressed before one can apply these techniques.
For example, how to adjust coding scheme to the changing wireless network environment to against
the problem of uncertainty of channel condition. Besides, in Cognitive Radio Network (CRN), it
is important to coordinate the transmissions of the primary users and that of the secondary users
so that Cognitive Users (CUs) will not cause unacceptable interference to Primary Users (PUs)
while their utility can be maximized, which also depends on the radio resource allocation for the
CUs. What’s more, in the mmWave network, due to the dynamic channel conditions, how to
optimally coordinate the concurrent transmissions of neighboring links based on their possible
channel conditions, so that the network throughput is can be improved.
In this dissertation work we study how to use adaptive video coding to combat channel uncertainty
in mmWave network and investigate how to optimize the channel allocation in video
streaming over cognitive radio networks.
The first part of this dissertation investigates the problem of streaming uncompressed HD
video over mmWave wireless networks so that the expected mean square error of the reconstructed
video quality isminimized. An adaptive coding scheme that can dynamically adjust to the changing
channel conditions is proposed so that error rate is reduced, and a dynamic interleaving based
transmission strategy is incorporated to avoid busty errors in transmission. Efficient algorithm
with low computational complexity is proposed to solve for the optimal setup.
The second part of this dissertation investigates the problem of video streaming over CRN.
Spectrum sensing and spectrum allocation are optimized such that so that Quality of Experience
(QoE) of CUs are maximized. Due to the non-linearity of the QoE model, the spectrum sensing
problem and the spectrum accessing problem are solved separately and some performance may be
lost. We discuss the case when each CU can sense only one channel each time and the case when
each CU can sense multiple channels each time.
The third part of this dissertation investigates the problem of video streaming over CRN,
where the spectrum sensing, spectrum allocation, and transmit power allocation are jointly optimized
such that so that Quality of Service (QoS) of CUs are maximized, which is significantly
different with the second part. We show that the formulated Mixed Integer NonLinear Programming
problem can be decomposed into two sub-problems without sacrificing optimality, and with
a much lower computational complexity. We analyze the proposed iterative algorithm with respect
to complexity and time efficiency, and derive a performance upper bound.
The fourth part of this dissertation investigates the problem of relay and link selection in a
dual-hop mmWave network aiming at minimizing the Maximum Expected Delivery Time among
all Tx-Rx pairs, while exploiting reflected mmWave transmissions and considering link blockage
dynamics. Due to the NP-hardness of the formulated problem, we develop a Decomposition Principle
to transform this problem into two sub-problems, one for link selection and the other for relay
assignment. We prove that the proposed scheme can achieve an optimality gap of just 1 time slot
at greatly reduced complexity. The proposed schemes are validated with simulations with their
superior performance observed.
The fifth part of this dissertation investigates the problem of user scheduling in multiple hop
mmWave networks, so that the number of time slots needed to serve all user’ traffic demand is
minimized. Channel condition changes over time and at each time slot, given the possible channel
states, the PNC decides the optimal routing path and which users should access the channel at
current time slot, aiming at maximizing the long term utility of the whole network. We propose
a heuristic algorithm with greatly reduced complexity to solve this problem, which first fix the
optimal routing path for a long term and then maximizes the instant network throughput. A simlilar
problem in singel hop mmWave network is also studied and an effective algorithm is proposed. The
performance of the heuristic algorithms is validated with simulations.
In summary, this dissertation aims to improving the QoS/QoE provisioning in emerging wireless
networks by addressing the resource allocation and user scheduling problems. In-depth analysis
and comprehensive results are also provided. Some of the findings may shed light on how to
put emerging techniques into real applications.||en_US