Pantelis A. Frangoudis’s research while affiliated with TU Wien and other places

Hierarchical federated learning (HFL) designs introduce intermediate aggregator nodes between clients and the global federated learning server in order to reduce communication costs and distribute server load. One side effect is that machine learning model replication at scale comes "for free" as part of the HFL process: model replicas are hosted at the client end, intermediate nodes, and the global server level and are readily available for serving inference requests. This creates opportunities for efficient model serving but simultaneously couples the training and serving processes and calls for their joint orchestration. This is particularly important for continual learning, where serving a model while (re)training it periodically, upon specific triggers, or continuously, takes place over shared infrastructure spanning the computing continuum. Consequently, training and inference workloads can interfere with detrimental effects on performance. To address this issue, we propose an inference load-aware HFL orchestration scheme, which makes informed decisions on HFL configuration, considering knowledge about inference workloads and the respective processing capacity. Applying our scheme to a continual learning use case in the transportation domain, we demonstrate that by optimizing aggregator node placement and device-aggregator association, significant inference latency savings can be achieved while communication costs are drastically reduced compared to flat centralized federated learning.

While various service orchestration aspects within Computing Continuum (CC) systems have been extensively addressed, including service placement, replication, and scheduling, an open challenge lies in ensuring uninterrupted data delivery from IoT devices to running service instances in this dynamic environment, while adhering to specific Quality of Service (QoS) requirements and balancing the load on service instances. To address this challenge, we introduce QEdgeProxy, an adaptive and QoS-aware load balancing framework specifically designed for routing client requests to appropriate IoT service instances in the CC. QEdgeProxy integrates naturally within Kubernetes, adapts to changes in dynamic environments, and manages to seamlessly deliver data to IoT service instances while consistently meeting QoS requirements and effectively distributing load across them. This is verified by extensive experiments over a realistic K3s cluster with instance failures and network variability, where QEdgeProxy outperforms both Kubernetes built-in mechanisms and a state-of-the-art solution, while introducing minimal computational overhead.

Sixth-generation (6G) networks anticipate intelligently supporting a wide range of smart services and innovative applications. Such a context urges a heavy usage of Machine Learning (ML) techniques, particularly Deep Learning (DL), to foster innovation and ease the deployment of intelligent network functions/operations, which are able to fulfill the various requirements of the envisioned 6G services. The revolution of 6G networks is driven by massive data availability, moving from centralized and big data towards small and distributed data. This trend has motivated the adoption of distributed and collaborative ML/DL techniques. Specifically, collaborative ML/DL consists of deploying a set of distributed agents that collaboratively train learning models without sharing their data, thus improving data privacy and reducing the time/communication overhead. This work provides a comprehensive study on how collaborative learning can be effectively deployed over 6G wireless networks. In particular, our study focuses on Split Federated Learning (SFL), a technique that recently emerged promising better performance compared with existing collaborative learning approaches. We first provide an overview of three emerging collaborative learning paradigms, including federated learning, split learning, and split federated learning, as well as of 6G networks along with their main vision and timeline of key developments.We then highlight the need for split federated learning towards the upcoming 6G networks in every aspect, including 6G technologies (e.g., intelligent physical layer, intelligent edge computing, zero-touch network management, intelligent resource management) and 6G use cases (e.g., smart grid 2.0, Industry 5.0, connected and autonomous systems). Furthermore, we review existing datasets along with frameworks that can help in implementing SFL for 6G networks. We finally identify key technical challenges, open issues, and future research directions related to SFL-enabled 6G networks.

Sixth-generation (6G) networks anticipate intelligently supporting a wide range of smart services and innovative applications. Such a context urges a heavy usage of Machine Learning (ML) techniques, particularly Deep Learning (DL), to foster innovation and ease the deployment of intelligent network functions/operations, which are able to fulfill the various requirements of the envisioned 6G services. The revolution of 6G networks is driven by massive data availability, moving from centralized and big data towards small and distributed data. This trend has motivated the adoption of distributed and collaborative ML/DL techniques. Specifically, collaborative ML/DL consists of deploying a set of distributed agents that collaboratively train learning models without sharing their data, thus improving data privacy and reducing the time/communication overhead. This work provides a comprehensive study on how collaborative learning can be effectively deployed over 6G wireless networks. In particular, our study focuses on Split Federated Learning (SFL), a technique recently emerged promising better performance compared with existing collaborative learning approaches. We first provide an overview of three emerging collaborative learning paradigms, including federated learning, split learning, and split federated learning, as well as of 6G networks along with their main vision and timeline of key developments. We then highlight the need for split federated learning towards the upcoming 6G networks in every aspect, including 6G technologies (e.g., intelligent physical layer, intelligent edge computing, zero-touch network management, intelligent resource management) and 6G use cases (e.g., smart grid 2.0, Industry 5.0, connected and autonomous systems). Furthermore, we review existing datasets along with frameworks that can help in implementing SFL for 6G networks. We finally identify key technical challenges, open issues, and future research directions related to SFL-enabled 6G networks.

Fog computing enables the execution of IoT applications on compute nodes which reside both in the cloud and at the edge of the network. To achieve this, most fog computing systems route the IoT data on a path which starts at the data source, and goes through various edge and cloud nodes. Each node on this path may accept the data if there are available resources to process this data locally. Otherwise, the data is forwarded to the next node on path. Notably, when the data is forwarded (rather than accepted), the communication latency increases by the delay to reach the next node. To avoid this, we propose a routing mechanism which maintains a history of all nodes that have accepted data of each context in the past. By processing this history, our mechanism sends the data directly to the closest node that tends to accept data of the same context. This lowers the forwarding by nodes on path, and can reduce the communication latency. We evaluate this approach using both prototype- and simulation-based experiments which show reduced communication latency (by up to 23%) and lower number of hops traveled (by up to 73%), compared to a state-of-the-art method.

We present an architecture for the provision of video Content Delivery Network (CDN) functionality as a service over a multi-domain cloud. We introduce the concept of a CDN slice, that is, a CDN service instance which is created upon a content provider's request, is autonomously managed, and spans multiple potentially heterogeneous edge cloud infrastructures. Our design is tailored to a 5G mobile network context, building on its inherent programmability, management flexibility, and the availability of cloud resources at the mobile edge level, thus close to end users. We exploit Network Functions Virtualization (NFV) and Multi-access Edge Computing (MEC) technologies, proposing a system which is aligned with the recent NFV and MEC standards. To deliver a Quality-of-Experience (QoE) optimized video service, we derive empirical models of video QoE as a function of service workload, which, coupled with multi-level service monitoring, drive our slice resource allocation and elastic management mechanisms. These management schemes feature autonomic compute resource scaling, and on-the-fly transcoding to adapt video bit-rate to the current network conditions. Their effectiveness is demonstrated via testbed experiments.