|ZAMBIANCO MARCO||Cycle: XXXIV |
Tutor: CESANA MATTEO
Advisor: VERTICALE GIACOMO Major Research topic
:Inter-slice and intra-slice resource allocation algorithms for 5G Radio Access NetworkAbstract:
Upcoming 5G systems are going to accommodate a heterogeneous plethora of services ranging from machine-type communication to mobile multi-broadband content delivery. To meet different QoS requirements demanded by each application, the "one-size-fits-all" network paradigm currently employed in modern telecommunications system is no longer suitable to address this new market model. On the other hand, deploying ad-hoc networks for each service would make CAPEX and OPEX unsustainable for networks operators. In this context, inspired by recent advances on Software Defined Networks (SDN) and Network Function Virtualization (NFV), Network Slicing has appeared as a key solution to accommodate these emerging business opportunities in next generation networks. In details, a network slice is defined as set of network functions, and resources to run these network functions, forming a complete instantiated logical network to meet specific network characteristics required by the application. Therefore, different services can be simultaneously multiplexed on top of the same physical network infrastructure by means of suitable virtual networks instancing. In this regard, state-of-the-art has mostly focused on the design of orchestration frameworks to efficiently optimize the deployment of VNFs according to the service level agreement (SLA) but the research on how to achieve such SLAs is lagging. Furthermore, major effort has been dedicated on the analysis of network slicing solutions for the core network as an extension of the well-known virtual network embedding problem widely studied in SDNs. On the other hand, only preliminary results have appeared with respect to slicing of the Radio Access Network and of the radio interface. In fact, being radio bandwidth a scarce resource whose quality is inherently affected by the wireless channel, an efficient slicing policy becomes essential for the sake of slices requirements achievement. More specifically, this investigation will deal with the inter-slice as well as intra-slice resource allocation problem. The former accounts for the coexistence of multiple services (provided by the same network operator or different ones) on the same network infrastructure and aims at ensuring isolation among different slices, whereas the latter accounts for the multi-user diversity within the same slice and aims at enforcing a proper QoS to each slice’s user. In addition, this research will study different approaches to optimize resource efficiency under distributed and centralized decisions by taking into account the new business model upon which network slicing is going to be based. Specifically, a common physical network infrastructure sets some limitations on the information (e.g. network capacity, number of users, latency, network load, etc.) that each virtual operator is willing to share among other competitors due to a potential lowering in the own economic profit. For this reason, in the analysis of novel slicing schemes (either distributed and centralized), a study on which system parameters can be globally employed for the sake of an efficient resource allocation becomes very important to quantify possible tradeoff with respect to operators’ revenue.