Towards an SDN/NFV-based multi-tenant network and cloud testbed for end-to-end 5G services Raul Mu˜ noz, Josep Mangues-Bafalluy, Nikolaos Bartzoudis, Ricard Vilalta, Ricardo Mart´ ınez, Ramon Casellas, Nicola Baldo, Jos´ e Nu˜ nez-Mart´ ınez, Manuel Requena-Esteso, Oriol Font-Bach, Miquel Payar´ o, Pol Henarejos, Ana Perez-Neira. Centre Tecnol` ogic de Telecomunicacions de Catalunya, CTTC Av. Carl Friedrich Gauss 7, 08860 Castelldefels (Barcelona), Spain Email: raul.munoz@cttc.es Abstract—5G has a main requirement of highly flexible, ultra- low latency and ultra-high bandwidth virtualized infrastructure in order to deliver end-to-end services. This requirement can be met by efficiently integrating all network segments (radio access, aggregation and core) with heterogeneous wireless and optical technologies (5G, mmWave, LTE/LTE-A, Wi-Fi, Ether- net, MPLS, WDM, software-defined optical transmission, etc.), and massive computing and storage cloud services (offered in edge/core data centers, and even, distributed in network nodes). This paper introduces the preliminary architecture aiming at integrating three consolidated and standalone experimental in- frastructures at CTTC, in order to deploy the required end-to- end top-to-bottom converged infrastructure pointed out above for testing and developing advanced 5G services. The exist- ing experimental facilities cover complementary technologies from terminals to radio access, aggregation/core and cloud, and are: the GEDOMIS R testbed (LTE/5G PHY testbed), the EXTREME Testbed R (wireless HetNet and backhaul, edge data- center,a d distributed computing nodes), and the ADRENALINE Testbed R (packet aggregation and optical core network, core data-center). Two uses cases addressing Fixed Mobile Conver- gence (FMC) developed in ADRENALINE and EXTREME, and virtual mobile network function splitting and deployment, involving EXTREME and GEDOMIS, are also presented. I. I NTRODUCTION Software Defined Networking (SDN) [1] and Network Functions Virtualization (NFV) [2] architectures are the key enablers to federate heterogeneous experimental facilities and to integrate both network and cloud resources to offer ad- vanced end-to-end 5G services upon multi-domain heteroge- neous networks and distributed data centers (DC). SDN and NFV have emerged as the most promising can- didates to improve network programmability and dynamic adjustment of the network resources. SDN is defined as a control framework that supports the programmability of network functions and protocols by decoupling the data plane and the control plane, which are currently integrated vertically in most network equipment. SDN proposes a logically central- ized architecture where the control entity (SDN controller) is responsible for providing an abstraction of network resources through Application Programming Interfaces (API). One of the main benefits of this architecture resides on the ability to perform control and management tasks of different wireless and wired network forwarding technologies (e.g., packet/flow switching or circuit switching) by means of the same network controller. The OpenFlow protocol is the most commonly deployed protocol for enabling SDN. It offers a logical switch abstraction, mapping high-level instructions of the protocol to hide vendor-specific hardware details, which mitigates inter- operability issues commonly found in multi-vendor deploy- ments [3]. This abstraction enables SDN to perform network virtualization, that is, to slice the physical infrastructure and create multiple co-existing network slices (virtual networks) independent of the underlying wireless or optical technology and network protocols. In a multi-tenant environment, these virtual networks can be independently controlled by their own instance of SDN control plane (e.g., virtual operators) [4]. The notion of NFV relates to deploying network functions that are typically deployed in specialized and dedicated hardware servers, as software instances (named virtual network func- tions - VNF) running on commodity servers in data-centers (DCs) (or in general, in computing distributed throughout the network) through software virtualization techniques. Ex- amples of VNFs include IP network functions such as load balancers, firewalls, security or Authentication, Authorization and Accounting (AAA), LTE/EPC network functions, such as Mobility Management Entity (MME), Serving Gateway (SGW), and PDN Gateway (PGW), or transport network functions [5] [6]. 5G poses stringent communication and control/management requirements. In addition to the expected 1000x increase in traffic volume and much lower latency to be handled in the data plane, on the management side, it will require the collection of a huge amount of data generated at the terminals, sensors, machines, nodes, etc., that will be transported through networks to data-centers in order to be processed (Big Data) and make the proper decisions (Cognition). Additionally, the recent rise of NFV services will also require investments in cloud/DC [7]. Originally, cloud computing services have been offered in centralized DCs. However, there is a general trend to spread the DCs to the edge of the network in order to reduce services’ latency to the end user. The extension of cloud computing and services to the edge of the network is known as fog computing, and it will lead to an exponential growth on the inter-data center traffic requiring high-bandwidth and low-latency 5G networks to interconnect them and offer global end-to-end cloud services. Current Distributed/Federated cloud