ALEXANDRIS Konstantinos

Thesis

Mobility Management and Resource Allocation towards 5G Radio Access Networks (RANs)

Mobility management and resource allocation are the key features of the current and next generation cellular systems that enable the users to change seamlessly their cell associations and maintain their service continuity with required quality-of-service (QoS). Software-defined networking (SDN) promises to overcome such adversities in network design by centrally controlling the behavior of the underlying network via an abstraction layer. These mechanisms are applied on those procedures in 5G mobile communications exploiting the holistic knowledge of the network.

 

In the first part of the thesis, we develop and examine the performance of the legacy 3GPP X2 handover (HO) using the OpenAirInterface (OAI) platform to understand not only the mobility management in 4G but also the interplay of system parameters on triggering the HO process towards designing a HO algorithm. As a next step, a load-aware HO decision algorithm for the next-generation heterogeneous networks (HetNets) is proposed considering the users’ service delay and the asymmetrical cells power. Results have shown that the proposed algorithm outperforms the conventional received signal strength (RSS) and distance-based ones. In addition, an SDN architecture is sketched to manage the system decisions resulting in a centralized network coordination.

 

In the second part of the thesis, the importance of users QoS in 5G networks turns our attention to radio resource management under the multi-connectivity setting. A utility-based resource allocation that considers the users QoS requirements is proposed and compared to proportional fairness schemes. Multi-connectivity use case is shown to outperform the legacy 3GPP single connectivity in terms of user satisfaction and aggregated network throughput. In that setup, besides the air-interface limitations, we assume local routing among cells allowing to offload the backhaul/core network. Subsequently, we introduce backhaul capacity constraints leading to results that show multi-connectivity is still superior to the legacy one while retaining users QoS. Finally, we study the performance gain for a multi-connected user under spatio-temporal channel variability (large and small-scale fading) by exploiting the opportunistic scheduling for loaded and unloaded cell scenarios. Extracted results reveal the benefits of UPF utility function with opportunistic scheduling in terms of users QoS.