Department of Physics
Library of the School of Science

2024-06-28

2024

Evaggelatos Spyridon

Aris L. Moustakas, Associate Professor, Department of Physics, University of Athens
Hector E. Nistazakis, Professor, Department of Physics, University of Athens
Sergios Theodoridis, Emeritus Professor, Department of Informatics & Telecommunications, University of Athens
George S. Tombras, Professor, Department of Physics, University of Athens
Ioannis Tigelis, Professor, Department of Physics, University of Athens
George Alexandropoulos, Associate Professor, Department of Informatics & Telecommunications, University of Athens
Petros S. Bithas, Assistant Professor, Department of Digital Industry Technologies, University of Athens

Statistical Physics Algorithms for Wireless Communication Networks

English

Statistical Physics Algorithms for Wireless Communication Networks

This thesis is devoted to the application of some well known mathematical tools, developed
in the field of theoretical physics, to the performance analysis of wireless communication
systems. The optimization problems, formulated in terms of the statistical
mechanics framework, are solved via the widely known message passing algorithm.
Specifically, the message passing algorithm was modified and applied to random
networks, modeled as bipartite acyclic graphs. Special emphasis was given to sparse
graphs due to the low probability of presence of cycles and the consequent ease of
converting them into trees, such as Cayley trees. These specific graphs can accurately
represent the properties of wireless networks with random connectivity and serve as a
tool for modeling cognitive radio-enabled IoT (Internet of Things) networks.
In summary, techniques from the theory of spin glasses in statistical physics were
applied to construct iterative algorithms for detecting and locating multiple emitting
sources from secondary wireless sensors. The main goal of this thesis was to address
the problem in a simple semi-analytical way. Additionally, the representation of bipartite
graphs in network topologies was examined for cases where there is no cooperation
between different network elements (primary sources, secondary users). The problem
mentioned above focused on algorithms for message passing between different entities.
The first part of this research, concerning such an architecture, was presented
in [1], where analytical expressions for the density of sources (primary and secondary)
were initially generated in the form of a closed set of equations and solved using the
population dynamics algorithm.
Furthermore, the case of random connectivity between nodes in a Rayleigh fading
channel was studied through extensive simulations. After comparing the two algorithms
in multiple implementations of random two-dimensional networks, their results
were shown to be almost identical. Finally, the convergence speed of the message
passing algorithm was also studied and found to be linear with the number of sources.
In the second part of this study, the detection of sources was examined using
the Replica Symmetry Method, producing results similar to the Bethe free energy
method. The probability of false source detection was derived as a function of two
iterative equations, including cases where sources are out of the range of the secondary
sources/sensors. The work in [2] included analytical expressions for the Replica Symmetry
Method, a comparison of the average communication energy for the message
passing algorithm with a nearest neighbor algorithm, a comparison of the message
passing algorithm with myopic algorithms and a detailed comparison of the message
passing algorithm with other multiple source detection methods. Thus, this study
provided an analytical methodology for detecting the state of multiple sources in a network with secondary users/sensors in the presence of noise.
In the final part of this thesis, the focus is on the study of multi-hop communication
networks with relays in cases of limited infrastructure for potential use of networks with
extended coverage. Specifically, the capacity and the probability of successful signal
transmission were calculated and signal reception models were analyzed using tools
from stochastic geometry [3, 4]. This analysis focused on simultaneous (orthogonal or
coexisting) transmission for asymptotic regions with nodes distributed according to a
Poisson distribution and in regions with low bit error rate.

Science

Statistical Physics, Wireless Networks, Stochastic Geometry, Message Passing Algorithms

Yes

3

Yes

135

156

File access is restricted only to the intranet of UoA.

https://pergamos.lib.uoa.gr/uoa/dl/object/3402158