The evolution of mobile radio networks has been consistently driven by the need to increase data rates to satisfy the demands of an increasingly connected society. With the advent of 5th generation (5G), the focus has shifted from merely enhancing throughput to enabling a diverse range of advanced communication services, including mission-critical applications, high-bandwidth services, and Ultra Reliable and Low Latency Communications (URLLC) communications. Among these, Vehicle-to-Everything (V2X) communication stands out, supporting essential applications such as collision avoidance, cooperative driving, and traffic optimization. These services impose stringent requirements on latency, reliability, and service continuity, exceeding the capabilities of traditional mobile networks. Meeting these demands requires a fundamental rethinking of network management paradigms. In this context, Self-Organizing Network (SON) has emerged as a key technology, providing autonomous configuration, optimization, and self-healing functionalities. The SON framework reduces operational complexity and enables real-time adaptation to dynamic network conditions. Its optimization and healing functions are particularly critical in 5G and V2X networks. When integrated with intelligent control frameworks such as Open RAN (O-RAN), SON unlocks unprecedented programmability and visibility, enabling tailored solutions for the unique challenges of 5G and V2X. This dissertation investigates the synergy between 5G, V2X, and SON running on top of O-RAN, laying the foundation for highly responsive and resilient mobile networks capable of supporting the most demanding use cases of the connected world. On this premise, we first highlight the characteristics of O-RAN, a Radio Access Network (RAN) architecture overhaul that offers unprecedented data collection and control capabilities, in line with the SON requirements, making it a primer for the widespread dissemination of the SON paradigm. In the initial phase of this disseration, we advanced the functions of SON-O-RAN within 5G networks by proposing a novel self-optimizing SON framework aimed at enhancing the Quality of Service (QoS) of user services exhibiting conflicting performance requirements. As a case study, we focused on the coexistence of two challenging services, namely bandwidth-intensive Heavy-Hitterss (HHs) and low-latency Multicast/Broadcast Services (MBS) services. In such a challenging use case, it is impossible to devise a single SON solution that entirely addresses the complexity of this situation; thus, the synergy of different SON functions is essential. In this context, we propose seamlessly using load-balancing and Resource Allocation (RA) SON functions to tackle the problem. Specifically, the proposed RA algorithm is based on Multilevel Queue Scheduling (MLQ) and is capable of guaranteeing the QoS of low-latency MBS services while simultaneously maintaining a high QoS for high-bandwidth HHs. An additional novelty of this solution lies in the use of the X2 interface to broadcast MBS data to neighboring cells, thereby minimizing inter-cell dissemination latency. Furthermore, a Deep-Neural Network (DNN)-based detection mechanism is tailored for HHs within a disaggregated 5G network architecture, and a heuristic load-balancing SON function is proposed, leveraging HH-guided handovers to further enhance the QoS of HHs services. Building on the substantial benefits of SON applications within 5G networks, the second phase of this doctoral dissertation was devoted to extending SON to V2X networks and developing advanced SON functions for this domain. Our contributions are both theoretical and experimental. Specifically, we examine the architectural extensions required to integrate V2X network into O-RAN, thereby enabling O-RAN to deploy SON functions over them, eventually highlighting the extensive opportunities that arise from this integration. Initially, nodes' —Connected and Autonomous Vehicles (CAVs)— mobility as an unexplored aspect in the O-RAN architecture is identified and analyzed. Subsequently, the use of the cellular infrastructure, along with the ubiquitous Below 6GHz (sub-6GHz) frequencies, is proposed as a feasible solution to relay O-RAN control messages to the CAVs and backwards. To foster practical experimentation, we present a simulation framework based on Network Simulator 3 (ns-3) and O-RAN that allows testing the V2X integration in a realistic environment. It is the first open-source V2X-O-RAN testbed. Using this framework, we addressed two open issues in V2X networks. The first involves the application of SON to mitigate link blockages through a self-healing Traffic Relaying solution, a critical mechanism for ensuring reliable communication in high-mobility scenarios. We provide a theoretical formulation of the problem and present extensive simulation results demonstrating the effectiveness of this approach. The second focuses on using self-optimization paradigm of SON to develop a novel RA scheme for V2X networks implemented on top of O-RAN. The theoretical model of the problem is outlined, and comprehensive simulation results illustrate the performance gains achieved by our solution.
L’evoluzione delle reti radio mobili è stata costantemente guidata dalla necessità di aumentare i tassi di trasmissione dati per soddisfare le esigenze di una società sempre più connessa. Con l’avvento della quinta generazione (5G), l’obiettivo si è spostato dal semplice incremento della velocità di trasmissione all’abilitazione di una vasta gamma di servizi di comunicazione avanzati, tra cui applicazioni mission-critical, servizi ad alta larghezza di banda e Comunicazioni Ultra Affidabili e a Bassa Latenza (URLLC). Tra questi, la comunicazione Vehicle-to-Everything (V2X) si distingue per il supporto ad applicazioni essenziali come l’evitamento delle collisioni, la guida cooperativa e l’ottimizzazione del traffico. Questi servizi impongono requisiti stringenti in termini di latenza, affidabilità e continuità del servizio, superando le capacità delle reti mobili tradizionali. Soddisfare tali esigenze richiede una profonda revisione dei paradigmi di gestione della rete. In questo contesto, la tecnologia Self-Organizing Network (SON) è emersa come elemento chiave, fornendo funzionalità di configurazione autonoma, ottimizzazione e self-healing. Il framework SON riduce la complessità operativa e consente un’adattabilità in tempo reale alle condizioni dinamiche della rete. Le sue funzioni di ottimizzazione e healing sono particolarmente cruciali nelle reti 5G e V2X. Integrato con framework di controllo intelligenti come l’Open RAN (O-RAN), il SON sblocca un livello senza precedenti di programmabilità e visibilità, consentendo soluzioni su misura per le sfide uniche del 5G e del V2X. Questa dissertazione indaga la sinergia tra 5G, V2X e SON eseguito su O-RAN, ponendo le basi per reti mobili altamente reattive e resilienti, capaci di supportare i casi d’uso più esigenti del mondo connesso. Su tale premessa, si evidenziano innanzitutto le caratteristiche di O-RAN, una revisione dell’architettura della Radio Access Network (RAN) che offre capacità di raccolta dati e controllo senza precedenti, in linea con i requisiti del SON, rendendola la base ideale per la diffusione del paradigma SON. Nella fase iniziale di questa dissertazione, sono state avanzate le funzioni SON-O-RAN all’interno delle reti 5G proponendo un nuovo framework SON auto-ottimizzante volto a migliorare la Qualità del Servizio (QoS) di servizi utente caratterizzati da requisiti prestazionali contrastanti. Come caso di studio, ci si è concentrati sulla coesistenza di due servizi particolarmente sfidanti, ossia, i servizi Heavy-Hitters (HHs), ad alta domanda di banda, e i servizi Multicast/Broadcast (MBS) a bassa latenza. In tale scenario complesso, non è possibile ideare una singola soluzione SON che affronti interamente la complessità del problema ed è quindi essenziale la sinergia di differenti funzioni SON. In questo contesto, si propone l’uso integrato delle funzioni SON di load-balancing e di allocazione delle risorse (Resource Allocation, RA) per affrontare il problema. L’algoritmo RA proposto, basato su uno schema di Multi-Level Queue Scheduling (MLQ), è in grado di garantire la QoS dei servizi MBS a bassa latenza mantenendo al contempo un’elevata QoS per i servizi HHs. Un’ulteriore novità della soluzione risiede nell’impiego dell’interfaccia X2 per la trasmissione broadcast dei dati MBS alle celle adiacenti, riducendo così la latenza nella disseminazione inter-cella dei pacchetti data. Inoltre, un meccanismo di rilevamento basato su Reti Neurali Profonde (Deep Neural Network, DNN) è stato progettato per identificare i flussi HH all’interno di un’architettura 5G disaggregata, e una funzione SON di load-balancing euristica è stata proposta sfruttando handover guidati dagli HHs per migliorare ulteriormente la QoS dei relativi servizi. Basandosi sui significativi vantaggi delle applicazioni SON nelle reti 5G, la seconda fase di questa dissertazione è stata dedicata all’estensione del SON alle reti V2X e allo sviluppo di funzioni SON avanzate per tale dominio. I contributi forniti sono sia teorici che sperimentali. In particolare, vengono esaminati gli adattamenti architetturali necessari per integrare le reti V2X in O-RAN, consentendo a quest’ultimo di distribuire funzioni SON su di esse, e vengono evidenziate le ampie opportunità derivanti da tale integrazione. È stato inoltre individuato e analizzato il tema della mobilità dei nodi — veicoli connessi e autonomi (Connected and Autonomous Vehicles, CAVs) — come aspetto ancora inesplorato nell’architettura O-RAN. Successivamente, viene proposta l’utilizzazione dell’infrastruttura cellulare, insieme all’impiego delle frequenze al di sotto dei 6 GHz (sub-6GHz), come soluzione praticabile per instradare i messaggi di controllo O-RAN verso i CAV e viceversa. Per favorire la sperimentazione pratica, viene presentato un framework di simulazione basato su Network Simulator 3 (ns-3) e O-RAN che consente di testare l’integrazione V2X in un ambiente realistico. Si tratta del primo testbed open source V2X-O-RAN. Attraverso tale framework, vengono affrontati due problemi aperti delle reti V2X. Il primo riguarda l’applicazione del SON per mitigare i ostacoli del collegamento tramite una soluzione di self-healing basata sul Traffic Relaying, un meccanismo cruciale per garantire comunicazioni affidabili in scenari ad alta mobilità. Viene fornita una formulazione teorica del problema, accompagnata da estesi risultati di simulazione che dimostrano l’efficacia dell’approccio. Il secondo si concentra sull’impiego del paradigma di self-optimization del SON per sviluppare un nuovo schema di allocazione delle risorse per le reti V2X implementato su O-RAN. Il modello teorico del problema viene delineato e i risultati di simulazione confermano i guadagni prestazionali ottenuti grazie alla soluzione proposta.
Programmable self-organizing radio access networks
GJEÇI, FRANCI
2025/2026
Abstract
The evolution of mobile radio networks has been consistently driven by the need to increase data rates to satisfy the demands of an increasingly connected society. With the advent of 5th generation (5G), the focus has shifted from merely enhancing throughput to enabling a diverse range of advanced communication services, including mission-critical applications, high-bandwidth services, and Ultra Reliable and Low Latency Communications (URLLC) communications. Among these, Vehicle-to-Everything (V2X) communication stands out, supporting essential applications such as collision avoidance, cooperative driving, and traffic optimization. These services impose stringent requirements on latency, reliability, and service continuity, exceeding the capabilities of traditional mobile networks. Meeting these demands requires a fundamental rethinking of network management paradigms. In this context, Self-Organizing Network (SON) has emerged as a key technology, providing autonomous configuration, optimization, and self-healing functionalities. The SON framework reduces operational complexity and enables real-time adaptation to dynamic network conditions. Its optimization and healing functions are particularly critical in 5G and V2X networks. When integrated with intelligent control frameworks such as Open RAN (O-RAN), SON unlocks unprecedented programmability and visibility, enabling tailored solutions for the unique challenges of 5G and V2X. This dissertation investigates the synergy between 5G, V2X, and SON running on top of O-RAN, laying the foundation for highly responsive and resilient mobile networks capable of supporting the most demanding use cases of the connected world. On this premise, we first highlight the characteristics of O-RAN, a Radio Access Network (RAN) architecture overhaul that offers unprecedented data collection and control capabilities, in line with the SON requirements, making it a primer for the widespread dissemination of the SON paradigm. In the initial phase of this disseration, we advanced the functions of SON-O-RAN within 5G networks by proposing a novel self-optimizing SON framework aimed at enhancing the Quality of Service (QoS) of user services exhibiting conflicting performance requirements. As a case study, we focused on the coexistence of two challenging services, namely bandwidth-intensive Heavy-Hitterss (HHs) and low-latency Multicast/Broadcast Services (MBS) services. In such a challenging use case, it is impossible to devise a single SON solution that entirely addresses the complexity of this situation; thus, the synergy of different SON functions is essential. In this context, we propose seamlessly using load-balancing and Resource Allocation (RA) SON functions to tackle the problem. Specifically, the proposed RA algorithm is based on Multilevel Queue Scheduling (MLQ) and is capable of guaranteeing the QoS of low-latency MBS services while simultaneously maintaining a high QoS for high-bandwidth HHs. An additional novelty of this solution lies in the use of the X2 interface to broadcast MBS data to neighboring cells, thereby minimizing inter-cell dissemination latency. Furthermore, a Deep-Neural Network (DNN)-based detection mechanism is tailored for HHs within a disaggregated 5G network architecture, and a heuristic load-balancing SON function is proposed, leveraging HH-guided handovers to further enhance the QoS of HHs services. Building on the substantial benefits of SON applications within 5G networks, the second phase of this doctoral dissertation was devoted to extending SON to V2X networks and developing advanced SON functions for this domain. Our contributions are both theoretical and experimental. Specifically, we examine the architectural extensions required to integrate V2X network into O-RAN, thereby enabling O-RAN to deploy SON functions over them, eventually highlighting the extensive opportunities that arise from this integration. Initially, nodes' —Connected and Autonomous Vehicles (CAVs)— mobility as an unexplored aspect in the O-RAN architecture is identified and analyzed. Subsequently, the use of the cellular infrastructure, along with the ubiquitous Below 6GHz (sub-6GHz) frequencies, is proposed as a feasible solution to relay O-RAN control messages to the CAVs and backwards. To foster practical experimentation, we present a simulation framework based on Network Simulator 3 (ns-3) and O-RAN that allows testing the V2X integration in a realistic environment. It is the first open-source V2X-O-RAN testbed. Using this framework, we addressed two open issues in V2X networks. The first involves the application of SON to mitigate link blockages through a self-healing Traffic Relaying solution, a critical mechanism for ensuring reliable communication in high-mobility scenarios. We provide a theoretical formulation of the problem and present extensive simulation results demonstrating the effectiveness of this approach. The second focuses on using self-optimization paradigm of SON to develop a novel RA scheme for V2X networks implemented on top of O-RAN. The theoretical model of the problem is outlined, and comprehensive simulation results illustrate the performance gains achieved by our solution.| File | Dimensione | Formato | |
|---|---|---|---|
|
PhD_Thesis.pdf
solo utenti autorizzati a partire dal 04/12/2026
Descrizione: PhD Thesis
Dimensione
9.28 MB
Formato
Adobe PDF
|
9.28 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/248057