|
Optimization of Wireless Networks Infrastructure Using Artificial Intelligence Methods
Šeda, Pavel ; Assoc. Prof. Sergey Andreev, D.Sci. (Tech.) (oponent) ; Araniti, Giuseppe (oponent) ; Hošek, Jiří (vedoucí práce)
The dissertation is focused on optimization of network infrastructures using artificial intelligence algorithms. The growing requirements for network traffic in 5G+ networks require decisive modifications to the network infrastructures. The main aim of this dissertation is to design an optimization model and algorithms for selecting suitable locations for the base station deployment. The proposed optimization model reflects the essential requirements of wireless coverage in today's networks, such as the required capacity (base stations and end-users), existing infrastructure, interference between base stations, or coverage range. Since the model represents an exponential problem that is not possible to solve for larger instances exactly in the available time, it was essential to apply artificial intelligence methods. For the computation the heuristic algorithms were selected and implemented, these are discussed in detail in the dissertation text. The proposed optimization models and algorithms are subsequently verified using suitable simulations for urban, suburban, or rural areas. The practical use of the proposed solution is considered as an additional module to existing tools, recommending the locations to deploy new base stations when the network parameters change (e.g., higher capacity requirements in certain areas), serving as a basis for further practical verification. To conclude, the main contributions are in the design of models extending classical covering problems together with the implementation using modified heuristic algorithms. Including their subsequent verification at instances with hundreds of thousands of nodes and their publication in impact journals and at international conferences.
|
|
Optimization of Wireless Networks Infrastructure Using Artificial Intelligence Methods
Šeda, Pavel ; Assoc. Prof. Sergey Andreev, D.Sci. (Tech.) (oponent) ; Araniti, Giuseppe (oponent) ; Hošek, Jiří (vedoucí práce)
The dissertation is focused on optimization of network infrastructures using artificial intelligence algorithms. The growing requirements for network traffic in 5G+ networks require decisive modifications to the network infrastructures. The main aim of this dissertation is to design an optimization model and algorithms for selecting suitable locations for the base station deployment. The proposed optimization model reflects the essential requirements of wireless coverage in today's networks, such as the required capacity (base stations and end-users), existing infrastructure, interference between base stations, or coverage range. Since the model represents an exponential problem that is not possible to solve for larger instances exactly in the available time, it was essential to apply artificial intelligence methods. For the computation the heuristic algorithms were selected and implemented, these are discussed in detail in the dissertation text. The proposed optimization models and algorithms are subsequently verified using suitable simulations for urban, suburban, or rural areas. The practical use of the proposed solution is considered as an additional module to existing tools, recommending the locations to deploy new base stations when the network parameters change (e.g., higher capacity requirements in certain areas), serving as a basis for further practical verification. To conclude, the main contributions are in the design of models extending classical covering problems together with the implementation using modified heuristic algorithms. Including their subsequent verification at instances with hundreds of thousands of nodes and their publication in impact journals and at international conferences.
|
|
Využití genetických algoritmů v problému pokrytí
Šeda, Pavel
Tato práce pojednává o použití genetických algoritmů a simulovaného žíhání k řešení problému pokrytí. V práci je popsán problém pokrytí a metody, kterými je vhodné problém řešit, se zaměřením na genetické algoritmy. V práci je implementován program v jazyce Java, který umožnuje zadávání parametrů genetického algoritmu a simulovaného žíhání. Uvedené metody jsou testovány na souboru dat většího rozsahu, převzatého z OR-Library a jsou zhodnoceny dosažené výsledky.
|