Research Areas
Network Function Virtualization (NFV)
NFV decouples network processing from specialized telco-oriented network devices (e.g., firewalls, proxies, load balancers, etc.), empowering the deployment of virtualized network functions (VNFs) on commodity servers with unprecedented flexibility and low cost. The deployment, configuration, and run-time management of VNFs entails various challenges (e.g., network service embedding, service chaining, VNF scaling and state management), which have spurred interest in the design and implementation of NFV orchestration systems. NetCloud has made fundamental contributions to NFV resource orchestration, such as network service embedding within a single and across multiple domains, datacenter network fabrics for scalable service chaining, as well as service function chain transformations to aid VNF placement on resource-constrained NFV infrastructures. Lately, the focus of the group is on AI/ML-assisted resource orchestration and, more specifically, on zero-touch service management (ZSM) which paves the way for optimized run-time operations with minimum human intervention.
Software-Defined Networking (SDN)
SDN decouples the control logic from the underlying network devices, providing the means for fine-grained network control. The P4 programming language further lowers the barrier for introducing innovation in the network, by providing the ability to express how packets are processed by programming dataplanes of hardware/software switches and (Smart)NICs. This opens up opportunities for better observability in the network. In this respect, In-Band Network Telemetry (INT) has emerged as a more viable approach to network monitoring, empowering P4 switches to access and update pre-defined telemetry indicators, encapsulated into custom packets headers. As such, telemetry data can be exported directly from the dataplane. NetCloud has introduced deterministic and probabilistic approaches to INT in order to alleviate the transmission overhead of conventional INT techniques, such as P4-INT.
Time-Sensitive Networking (TSN)
TSN is expected to provide a solid underpinning that will support modern application connectivity in various use cases and operational environments, such as industrial networks, automotive communications, mobile and wireless networks, as well as service provider networks. In this respect, IEEE 802.1 TSN has introduced a wide range of techniques for time synchronization, traffic scheduling, frame preemption, asynchronous traffic shaping, and stream reservation, among others. NetCloud, in collaboration with industrial (e.g., Huawei) and academic partners, has employed segment routing for explicit path control of TSN flows. The group has also built a platform that integrates Time-Aware Shaper based scheduling with centralized TSN control. Currently, NetCloud seeks to reap the benefits of TSN for low-latency communication between IoT devices and their virtualized counterparts in converged IoT-cloud environments.
Internet-of-Things (IoT)
IoT technologies are evolving at a rapid pace over the last years, leading to the development of various solutions for the efficient, secure, and reliable deployment of IoT applications. Such solutions include the development of IoT platforms in the form of middlewares to tackle interaction with IoT devices and data fusion aspects, the development of data modeling solutions for the representation of the entities in the IoT world, and the collected data, as well as the development of virtual counterparts of IoT devices and systems in the form of Virtual Objects (VOs). In this context, NetCloud, in collaboration with NEPHELE project partners, is working on the development of a VOStack, augmenting VOs with the execution of generic or device-specific functions for data processing and fusion in order to alleviate the computational burden of IoT devices and reduce their energy consumption. Furthermore, VOStack can facilitate the representation and management of (clusters of) IoT devices through unified abstractions, whereas, at the same time, it can address convergence and interoperability aspects.
Edge Computing
5G verticals are often composed of various time- and mission-critical applications that are developed as end-to-end services over heterogeneous distributed computing and network infrastructures. Towards latency reduction and bandwidth savings, infrastructure providers offer computing, network, and storage resources at the edge of the network. To this end, edge computing has evolved as the emerging service delivery paradigm, raising various service management and orchestration challenges. This service ecosystem opens up opportunities for cross-service interactions, in the sense that one network service can consume another one, enhancing its functionality. For example, an augmented reality service offering overlaid metadata for current location could enrich its service delivery by providing personalized recommendations through a social network. In this respect, NetCloud, in collaboration with MESON project partners, has designed and implemented a platform that fosters cross-service communication over edge computing infrastructures in a secure and controlled manner.