The ADRENALINE testbed® is designed and developed by the CTTC Optical Networks and Systems Department for experimental research on high-performance and large-scale intelligent optical transport networks. It showcases a set of developed technologies:
- SDN/OpenFlow controllers and agents: OpenFlow (OF), as a concrete implementation of Software Defined Networking (SDN), is an open standard (architecture and protocol) for a centralized control plane around the concept of an OF controller. This technology has been enabled in the ADRENALINE testbed in the recent years, as a new centralized Control Plane solution for the Control and Management of packet and heterogeneous optical networks (e.g., fixed- and flexi-grid DWDM transport). The ADRENALINE testbed also includes a dedicated OpenFlow island for packet switched networks and an intra-data centre L2 OpenFlow Network. An integrated PCE/OF controller with proprietary optical extensions has been developed, which communicates concurrently with each of the OF switches (i.e., OF agents) distributed along the network using the OF protocol and upon request.
- Cloud Computing Datacentres: Several small scale datacentres with IT computing, storage and networking resources have been introduced in the ADRENALINE testbed. Also L2 OF switches have been deployed to provide intra-datacentre connectivity. The datacentres can be handled via proprietary tools and extensions upon OpenStack, which is an open source cloud computing solution.
- OpenDaylight (ODL) proprietary extensions for Network Orchestration: ODL is the most important open source SDN controller, which can be used upon OF networks. ODL has been extended to support network orchestration towards different network control paradigms, such as GMPLS/PCE. The introduced extensions allow ODL to support both intra-DC and inter-DC network connectivity.
- Path Computation Element (PCE): ADRENALINE includes a PCE, which is a dedicated network entity responsible for doing advanced path computations. It combines, in a single system, a Path Computation Engine and a standard PCEP protocol interface to access Path Computation Services. The system is composed of a main PCE multi-threaded asynchronous process, running as a PCEP/TCP server in order to accept and process path computation requests from Path Computation Clients (PCC). The PCE serves requests from PCCs, and computes constrained explicit routes (EROs) over the topology that constitutes the optical transport layer. Different architectures can be deployed, including multi-domain path computation with Backwards Recursive Path Computation (BRPC) or with a hierarchical-PCE (H-PCE); multi-layer path computation in MPLS-TP over WSON networks, and stateful PCEs with instantiation capabilities. The PCE can also be used in an OpenFlow network and as a functional element in the scope of network virtualization, in order to virtualize data networks and services.
- Wavelength Switched Optical Networks (WSON): an all-optical Dense Wavelength Division Multiplexing (DWDM) mesh network with two colourless Reconfigurable Optical Add-Drop Multiplexer (ROADM) nodes and two Optical Cross Connect (OXC) nodes provides high-bandwidth, reconfigurable (in space and in frequency) end-to-end lightpaths, transparent to the format and payload of the client signals (e.g., SONET/SDH, Gigabit Ethernet). The transport plane also includes non-intrusive Optical Performance Monitors and optical systems suitable to compensate the impact of physical impairments (e.g., All-Optical Wavelength Converter, Amplified Tunable Dispersion Compensator).
- Network Virtualization (Virtualization): the network virtualization architecture of ADRENALINE is based on virtualizing both the GMPLS control and the WSON data planes. The virtualization of the GMPLS control plane allows creating virtual GMPLS connection controllers that can be used to deploy virtual networks with their own virtualized GMPLS control plane instance. The virtualization of the distributed GMPLS-based control plane relies on the usage of the GNU/Linux kernel KVM virtualization technology and the virtualization capabilities of the Intel ® CPUs (Intel VT). Once virtualized, the same physical machine hosts a configurable number of virtual guests, each replicating an OCC. With respect to the WSON data plane, the optical fiber switches and the controller cards of ROADM nodes support virtualization, being able to handle multiple TCP connections from virtualized OCCs. Finally, a new instance has been introduced so that the tunable transponders can also deal with several TCP connections from virtualized OCCs.
- Network Function Virtualization (NFV) relies upon, but differs from traditional server virtualization techniques such as those used in enterprise IT. A virtualized network function, or VNF, may consist of one or more virtual machines running different software and processes, on top of industry standard high volume servers, switches and storage, or even cloud computing infrastructure, instead of having custom hardware appliances for each network function. End-to-end virtualization services are performed by the Virtual Network Controller (VNC).
- Application-Based Network Operation (ABNO) architecture is implemented in ADRENALINE to perform end-to-end Network Orchestration, regardless the domain, including the Topology Manager (i.e., REST API), Internal architecture based on modular plugins, Network Plugin Manager, PCE plugin, OpenDaylight plugin, and the Provisioning Manager (i.e., PCEP protocol). And from a high-level view, the SDN Integrated IT and Network Ochestrator (SINO) is a centralized system able to coordinate cloud and network service management aspects in modern multi-tenant environments which provides the platform to run user applications and virtualized network functions (VNF Manager).
- A GMPLS Unified Control Plane (GMPLS): Generalized Multiprotocol Label Switching (GMPLS) controllers have been designed and developed for implementing a distributed control plane, enabling the dynamic setup, management and release of data connections. This includes dedicated controllers for the WSON handling, dynamically and in real-time, the resources of the optical node in order to manage the automatic provisioning and survivability of lightpaths as well as an emulated network with up to 74 controllers for exhaustive performance evaluation. Each GMPLS controller executes several collaborative processes and the corresponding control plane protocols: the connection controller, executing the Resource Reservation Protocol–Traffic Engineering (RSVP-TE) signaling protocol, the routing controller, with the Open Shortest Path First–Traffic Engineering (OSPF-TE) routing protocol, the link management agent, with the Link Management Protocol (LMP) along with the Link Resource Manager (LRM) that manages the local resources, the hardware abstraction layers and drivers. The controllers are also manageable via the Simple Network Management Protocol (SNMP).
- Carrier Ethernet Technologies (MPLS-TP/PWE3): ADRENALINE includes a connection-oriented IP/Ethernet Packet Transport Network (PTN), which is based on the Multiprotocol Label Switching–Transport Profile (MPLS-TP) and Pseudo-Wire emulation end-to-end (PWE3) architectures. The introduction of these technologies enabled the evolution of ADRENALINE from a single optical switching layer to dual-layer architecture. Three flexible GMPLS-controlled MPLS-TP/PWE3 nodes, implemented with the Click Modular Router software and with integrated 10Gb/s tunable DWDM transponders, have been deployed to enable flexible packet aggregation and grooming at 10/100/1000 Ethernet, allowing 10 GigE LAN traffic trunks.
- Applications for testbed operation (Operation software): ADRENALINE Network Configurator (ADNETCONF) and ADRENALINE Network Generator (ADNETGEN) are advanced and adapted software applications and tools to allow the rapid operation and maintenance of the testbed, including common tasks such as the configuration and parameterization of the network topology, the generation of client requests modelling the behavior of network customers, or the monitoring, data-mining and statistical processing of obtained results, allowing researchers to obtain numerical performance data and to perform experimental research and quantitative comparative analysis.
- DSP-enabled software defined optical transmission (SDOT): digital signal processing (DSP) has been adopted as a key methodology to develop optical transceivers than can be dynamically adapted to different modulation formats and/or bandwidth occupancies. This technology is referred to as DSP-enabled SDOT. Within ADRENALINE framework, off-line DSP is performed at both the transmitter and receiver subsystems by using different scripting languages (e.g., Matlab or Python). The generated scripts contain the necessary functions in order to implement the different blocks involved: serialization/parallelization, generation of the desired modulation format, up/down conversion to an intermediate RF (e.g., for variable guard band generation), synchronization, equalization, demapping, etc. Because of SDOT reconfigurability, optimization techniques as non-uniform bit/power loading can also be implemented for the transceivers.
- Optical OFDM systems (O-OFDM): Optical Orthogonal Frequency Division Multiplexing (O-OFDM) is a promising technology for future optical networks, thanks to its robustness against transmission impairments, its spectral efficiency and its unique flexibility and scalability to high-speed transmission. OFDM-based transceivers can be adapted to different modulation formats and bit/power loading schemes, while achieving subwavelength granularity. Among the different O-OFDM techniques, those based on intensity modulation and direct detection (IM/DD) constitute a cost-effective solution. Coherent (CO) OFDM schemes have been investigated in order to enhance the spectral efficiency, the attainable distance, and to support higher bit rate (up to 100 Gb/s). Preliminary tests of new devices acquired for CO schemes have also been performed, including different schemes with nested MZM for I-Q modulations and phase modulator for constant-envelope (CE) OFDM. In ADRENALINE, transceivers are implemented using either the fast Fourier transform (FFT) or the fast Hartley transform (FHT). In FHT-based O-OFDM systems, low-complex DSP is used, where only real algebra and M-PAM format are required.
- Bandwidth variable transponders for Elastic Optical Networks (BVT): The bandwidth variable transponder (BVT) is a key element of an Elastic Optical Network (EON). The ability of generating elastic optical paths is enabled by its flexibility in terms of variable parameters/attributes, such as the modulation format, the data rate, the number of carriers and the bandwidth occupancy. Software-defined BVTs allow reconfiguring the transmission scheme with a suitable selection of these flexible parameters, for an optimal resource usage in a flexgrid network. ADRENALINE roadmap includes future developments of elastic and sliceable transponders (S-BVT), now under investigation. Recently, several devices and equipment have been acquired in order to start the implementation of these transponders (e.g., programmable optical filters based on LCoS technology, with 1×1 and 1×4 configurations).
The right combination of the technologies above described, always within the framework of the ADRENALINE testbed, enables the flexible deployment of two main demonstration platforms. On the one hand, the SDN/NFV Cloud Computing Platform and Core Network for 5G Services encompasses multiple interrelated although independent components and prototypes, to offer end-to-end services, interconnecting users and applications across a wide range of heterogeneous networks technologies for the development and test of 5G services. Different components span IT and networking domains, and allow researchers, system vendors and operators to evaluate experimentally, in conditions close to production systems, all aspects related to cloud computing in distributed environments with multiple geographically split data centers, while jointly managing storage, computing and networking resources. ADRENALINE includes a multi-technology control plane for multilayer (packet over optical) networks, which manages the networking resources and covers the long-haul core transport and aggregation segments. In brief, a control plane is software that automates the processes involved in the provisioning of networking services, such as optical lightpaths, or Ethernet/MPLS-TP/IP connectivity services. The design of the ADRENALINE control plane follows broad SDN principles, such as stacking components in a hierarchical setting with different levels of abstraction. Network connectivity services are provisioned by an overarching control orchestration. In particular, at a given domain and layer, the control plane can based on the GMPLS technology and protocols (i.e., a distributed system in which a dedicated controller is responsible for each node autonomously) or follow SDN/OpenFlow principles, with a centralized controller that manages all the aspects of a network, dynamically configuring networks according to users an application needs. GMPLS control planes can be augmented with a PCE, which is an application or service that assumes specific tasks and responsibilities of the control plane such as computing optimal routes or acting as a central point for connection management (Active Stateful PCE). End to end Network Orchestration (to provide an overarching control regardless of the number of domains) is enabled with extensive usage of the ABNO architecture and framework, using the services of the ADRENALINE control plane. End-to-end network virtualization services are performed by a VNC, which is able to provide abstracted multi-layer network views to customers, ensuring security, isolation and independent SDN control (i.e., Customer SDN controllers). As mentioned, in the all-interconnected context in which end-to-end 5G services may span heterogeneous cloud-computing and networking technologies, ADRENALINE includes a SINO, which provides the platform to run user applications and virtualized network functions (VNF Manager). A NFV orchestrator is also provided in to deploy end-to-end VNF through VNF Forwarding Graphs. The Cloud Computing service manager is implemented in terms of a modified OpenStack software, one of the top open-source distributed cloud computing systems.
On the other hand, within the ADRENALINE testbed we have also implemented an Experimental Platform for Optical OFDM Systems (EOS). Specifically, the design of flexible, optical OFDM (O-OFDM) transmission schemes are studied and assessed. The experimental validation of the investigated modulation schemes based on multicarrier technology is enabled by our testbed equipment and it comprises two different parts. One is a photonic mesh network of 4 nodes with links of up to 150 km that is used to experimentally assess the different optical OFDM systems. The other one encompasses the different setups, including optical and optoelectronic systems and subsystems, for multicarrier modulation (either OFDM or DMT) transmission and reception based on offline processing. On one side, either an arbitrary waveform generator (up to 24 GSa/s and 9.6 GHz) or a high-speed digital to analog converter (up to 64 GSa/s and 13 GHz) provide an analog signal (from a digital one) to be modulated and transmitted on an optical link. On the other side, with a real-time digital phosphor oscilloscope (up to 100 GSa/s, 20 GHz bandwidth) placed on the reception side, the resulting electrical signals after optical reception are acquired and further processed. Different DSP modules are available and can be conveniently designed for software-defined optical transmission. Additionally, the EOS platform can be suitably configured by integrating several options of optical, electro-optical and electrical devices required to implement DSP-enabled sliceable bandwidth variable transponders (S-BVT). The multiple rate, multiple format, multiple reach, and multiple flow transmitters and receivers with sub and super-wavelength granularity can find application in different network segments, particularly for elastic optical networks and flexgrid technologies, possibly being SDN-enabled. Broadband interferometric modulators (Mach-Zehnder up to 40 GHz), based on different technologies (LiNbO3, GaAs) with the corresponding linear drivers are available to generate the optical signals from the analog electrical one. The platform is equipped with tunable laser sources with picometer resolution and several optical reception options, namely simple PIN photo-detection (capable to detect signals with bandwidth up to 50 GHz) or coherent reception front-end featuring phase diversity. In order to setup the desired transceiver/system configurations, other optical and electrical devices are available, including tunable optical filters with variable bandwidth and high power optical amplifiers.