Towards xURLLC strategies with programmable pipelines and network coding

• Autor(a) principal: Dias, Eurico Osório Marques
• Data de Publicação: 2022
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10773/37892
• topic: xURRLC; Programmable pipelines; Software-defined networks; P4 programming; In-band telemetry; Wireless backhaul; Network coding; 5G networks; Heterogeneous networks; MmWave network; Source routing; Network path optimization

The strong increase in the traffic volume and sudden changes in recent years have led to the evolution of carrier-grade network architectures, culminating in the current 5G generation. Increasing cell density due to traffic evolution towards xURLLC represents a new investment in the backhaul infrastructure expansion. Thus, fiber cable deployments become unsustainable in terms of urban site availability and monetary costs. This poses a use-case opportunity for line-of-sight wireless links, especially on mmWave technologies. However, the technology is lossy, while also suffering from loss of line-of-sight, making it unreliable. The work in this dissertation focuses on mitigating those problems in mmWave links, as part of 5G backhaul infrastructures and further generation of cellular communications. This is done through two different perspectives. The first perspective addresses the backhaul reliability with adaptive forwarding of 5G QoS flows, based on link quality assertion. To achieve this, a monitoring solution based on In-band Network Telemetry was designed, based on programming pipeline implementation. In the data plane, L2 forwarding and bridging, as well as SR-MPLS routing was implemented, using a forwarding pipeline in P4, that incorporates the telemetry model. Regarding the control plane, an ONOS application was written to manage the pipeline execution, while also receiving information from N1 and N2 5G logical interfaces, from UE and PDU session contexts. With this, labeled path routes and data flows are created and optimized, adapting flow paths to link quality conditions, enforcing 5QI constraints, and achieving xURLLC services. The experiments performed have shown that degraded links are avoided, and flow paths meet QoS conditions. The second perspective is performed at the transport layer, where improvements over mmWave channels are made through the use of an Adaptive and Causal Random Linear Network Coding (AC-RLNC) algorithm, on top of PHY and MAC layer coding schemes, to achieve xURLLC services. The solution is then compared to RLNC rateless codes. For this purpose, a simulator was developed for the algorithm execution, using a real mmWave channel data transfer model. Data point collection is performed regarding its normalized throughput, and means and maximum in-order delivery delay measurements. Results have demonstrated that it is possible to obtain ultra-reliable and low latency services with adaptive network coding above mmWave links.