Ανάλυση και σχεδίαση πυκνά δομημένων ασυρμάτων δικτύων
Analysis and design of ultra-dense wireless networks
View/ Open
Keywords
Ασύρματα δίκτυα ; Στοχαστική γεωμετρία ; Κυψελωτά δίκτυα ; D2D επικοινωνίες ; Cellular networksAbstract
The proliferation of wireless devices and the introduction of novel applications and
services towards the realization of the internet of things (IoT) impose serious challenges
for the cellular network operation. It is widely recognized that addressing
these challenges will require a revolutionary (rather than evolutionary) upgrade of
current cellular networks. The major enabler of this upgrade will be ultra dense networking,
i.e., operation under extreme infrastructure densification levels, motivated
by the need for an aggressive spatial frequency reuse towards providing the high
quality of service requested by an extremely large number of user devices. At the
same time, this extreme density of devices strongly motivates the introduction of
new channel access methods (modes), in addition to the traditional uplink/downlink
transmissions, with the most prominent example being that of direct device-todevice
(D2D) communications. These two features of the future network (densification
and new access methods), although, in principle, able to provide performance
benefits, also introduce significant challenges that were not present in the previous
cellular network generations.
This thesis investigates various aspects of the operation of future ultra dense cellular
networks using analytical tools from stochastic geometry. Regarding conventional
downlink transmissions, the impact of an increasing infrastructure density on user
performance is investigated. It is shown that infrastructure densification alone is
able to provide considerable gains, however, when supported by a carefully designed
intra-cell resource allocation (scheduling) scheme. Even though performance
is monotonically increasing with increasing infrastructure density, densification beyond
the level where the number of access points is comparable to the number of
users provides only moderate gains and, therefore, may not be of interest in practice.
The potential of D2D communications as a technique that is beneficial to the overall
system performance is investigated based on the new notion of operational region of
D2D communications, which identifies the range of system conditions (in terms of
parameters such as density of access points and user devices) where D2D communications
are beneficial. The operational region is identified in closed form for various
combinations of mode selection, spectrum sharing, and channel access schemes.
Analysis reveals that the introduction of D2D communications enhances the performance
of the conventional (non-D2D-enabled) cellular network under all operational
conditions as long as a distance-based mode selection scheme is adopted and D2D
transmissions are performed on the same bandwidth as cellular transmissions with
appropriate power control. The critical issue for future cellular networks of characterization of cross-mode interference,
i.e. interference experienced by a receiver operating on a certain access mode
(e.g., cellular downlink) due to transmissions corresponding to other modes (e.g.,
D2D) is examined. By adopting a general system model, a characterization of intercell
cross-mode interference statistics is analytically obtained that is applicable for
many operational scenarios of interest, ranging from conventional uplink/downlink
to cross-cell D2D transmissions, with interference generated by transmissions of different
mode(s). The analysis allows for an efficient computation of performance
measures of interest such as coverage probability and can serve as an analytical
baseline for sophisticated design of, e.g., resource allocation algorithms.
As the extreme infrastructure densification motivates the management of the network
by multiple competing operators, a techno-economic investigation of the system
is performed under the assumption of network partitioning over multiple operators
offering their services to self-interested users. By combining tools from stochastic
geometry and game theory, it is shown, among others, that users enjoy maximum
quality of experience in a market where multiple operators are competing with comparable
networks in terms of density and transmission technology. Although increasing
infrastructure density is always beneficial from an engineering perspective
(e.g., higher user rates), it is shown for the tractable case of two competing operators
that stable market conditions cannot be achieved above a certain density
threshold, suggesting a different market model (e.g., based on partial cooperation
among operators).