Due to the increasing number of heterogeneous user applications requiring Internet connection to operate properly (e.g., real-time video streaming, electronic sports, eXtended Reality (XR), etc.), the network requirements to sustain these huge amounts of traffic are becoming stricter and stricter. For this reason, now more than ever researchers from academia, industry and standards organizations are challenged to design, propose and introduce innovations to the next generation of mobile networks. To do so, a combination of mathematical models, simulations and empirical studies is fundamental to come up with solutions that consider every aspect of nowadays telecommunication challenges. In this thesis, not only do we use these tools to contribute to the advancement of the state of the art in different telecommunication domains, but we also propose open source alternatives that the community can benefit from. In the context of vehicular networks, we developed MilliCar, an ns-3 module for Vehicle-to-Vehicle (V2V) mmWave networks, which features a detailed implementation of the sidelink PHY and MAC layers based on NR V2X specifications. Through extensive simulation campaigns, we showed how vehicular communications can benefit from the use of mmWave frequencies, despite the numerous challenges associated to this part of the spectrum. We also proposed a new entity to orchestrate teleoperated vehicles, integrated at the Radio Access Network (RAN) level and operating in the sub-6 GHz frequency bands, that implements Predictive Quality of Service (PQoS) functionalities with the support of an Artificial Intelligence (AI) framework. Moreover, we focused on indoor mmWave networks, presenting a mathematical model for the scheduling of periodic requests, and a cross-layer policy (based on both experimental measurements and simulations) to optimize the performance of the Transmission Control Protocol (TCP). In addition, we studied VR applications as a use case for next-generation networks. Specifically, we collected traces from a realistic setup, analyzed their temporal correlation, and distilled two prediction models for future video frame size, which can be instrumental in the creation of dynamic resource allocation algorithms. Finally, we showed how AERPAW, the NSF-funded testbed for advanced wireless research, can be used for the end-to-end full-stack performance evaluation of networks in realistic scenarios, providing a reference for future research on UAV communication.
La crescita del numero di applicazioni che richiedono una adeguata connessione ad Internet (ad esempio, streaming di video in tempo reale, videogiochi online, eXtended Reality (XR), etc.), ha portato ad un progressivo inasprimento dei requisiti minimi necessari a sostenere questi nuovi flussi di traffico. Di conseguenza, ora più che mai la sfida per i ricercatori nel campo delle telecomunicazioni consiste nel progettare e proporre soluzioni adeguate a supporto delle reti mobili di nuova generazione. Per raggiungere questo obiettivo, l’utilizzo congiunto di modelli matematici, simulazioni e studi empirici è fondamentale per considerare ogni aspetto delle sfide da affrontare. Sfruttando questi strumenti, in questa tesi presenteremo soluzioni per lo studio di reti wireless in contesti eterogenei, mettendole a disposizione di tutta la comunità. Nell’ambito delle reti veicolari abbiamo sviluppato MilliCar, un modulo plug-in per il simulatore di rete Network Simulator 3 (ns-3), con l’obiettivo di studiare l’impatto dell’utilizzo di frequenze millimetriche per comunicazioni Vehicle-to-Vehicle (V2V) attraverso un’estesa campagna di simulazioni. Spostando il focus sulle bande di comunicazione sotto i 6 GHz, abbiamo proposto un algoritmo di Artificial Intelligence (AI) con l’obiettivo di implementare una strategia di Predictive Quality of Service (PQoS) in uno scenario di guida remota. Per uno studio a 360◦ , l’impatto della comunicazione ad alte frequenze è stato studiato anche in contesti indoor. In questo caso, abbiamo portato un duplice contributo: (i) un modello matematico per l’allocazione di risorse a flussi di traffico periodici per reti WiFi, e (ii) una strategia cross-layer per l’ottimizzazione delle prestazioni del protocollo TCP. Ci siamo poi concentrati su applicazioni di realtà virtuale, monitorando il traffico sulla rete generato con un setup realistico, ed analizzandone la correlazione temporale. Di conseguenza, abbiamo ottenuto modelli predittivi che possono essere integrati in algoritmi di allocazione delle risorse. Per concludere, abbiamo dimostrato come AERPAW, un testbed finanziato dalla National Science Foundation (NSF) situato a Raleigh, Carolina del Nord, può essere utilizzato per studiare i molteplici aspetti dell’utilizzo dei droni per la comunicazione in zone sia rurali che urbane.
End-to-end Evaluation and Optimization of the Next Generation of Mobile Networks / Drago, Matteo. - (2023 Feb 17).
End-to-end Evaluation and Optimization of the Next Generation of Mobile Networks
DRAGO, MATTEO
2023
Abstract
Due to the increasing number of heterogeneous user applications requiring Internet connection to operate properly (e.g., real-time video streaming, electronic sports, eXtended Reality (XR), etc.), the network requirements to sustain these huge amounts of traffic are becoming stricter and stricter. For this reason, now more than ever researchers from academia, industry and standards organizations are challenged to design, propose and introduce innovations to the next generation of mobile networks. To do so, a combination of mathematical models, simulations and empirical studies is fundamental to come up with solutions that consider every aspect of nowadays telecommunication challenges. In this thesis, not only do we use these tools to contribute to the advancement of the state of the art in different telecommunication domains, but we also propose open source alternatives that the community can benefit from. In the context of vehicular networks, we developed MilliCar, an ns-3 module for Vehicle-to-Vehicle (V2V) mmWave networks, which features a detailed implementation of the sidelink PHY and MAC layers based on NR V2X specifications. Through extensive simulation campaigns, we showed how vehicular communications can benefit from the use of mmWave frequencies, despite the numerous challenges associated to this part of the spectrum. We also proposed a new entity to orchestrate teleoperated vehicles, integrated at the Radio Access Network (RAN) level and operating in the sub-6 GHz frequency bands, that implements Predictive Quality of Service (PQoS) functionalities with the support of an Artificial Intelligence (AI) framework. Moreover, we focused on indoor mmWave networks, presenting a mathematical model for the scheduling of periodic requests, and a cross-layer policy (based on both experimental measurements and simulations) to optimize the performance of the Transmission Control Protocol (TCP). In addition, we studied VR applications as a use case for next-generation networks. Specifically, we collected traces from a realistic setup, analyzed their temporal correlation, and distilled two prediction models for future video frame size, which can be instrumental in the creation of dynamic resource allocation algorithms. Finally, we showed how AERPAW, the NSF-funded testbed for advanced wireless research, can be used for the end-to-end full-stack performance evaluation of networks in realistic scenarios, providing a reference for future research on UAV communication.File | Dimensione | Formato | |
---|---|---|---|
Tesi_definitiva_Matteo_Drago.pdf
accesso aperto
Descrizione: Tesi_Definitiva_Matteo_Drago
Tipologia:
Tesi di dottorato
Dimensione
20.12 MB
Formato
Adobe PDF
|
20.12 MB | Adobe PDF | Visualizza/Apri |
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.