In hierarchical federated learning, client selection is typically performed at edge servers and the central server. Edge servers perform intermediate model aggregation before transmitting updates to the central server, which aggregates the local models’ parameters. One of the challenges in federated learning is coping with the heterogeneity of clients’ resources and datasets, which can lead to slow global model convergence and poor performance. This paper introduces a client selection algorithm to address heterogeneity and scalability challenges in hierarchical scenarios. Our algorithm ranks client clusters based on a scoring function that incorporates resource availability, communication quality, and data-related attributes. Then, clients are selected from the top-ranked cluster. The proposed method is formulated as a combinatorial optimization problem and evaluated via emulation using the MNIST and CIFAR-10 datasets under heterogeneous scenarios. Experimental results show that, compared to existing clustering-based baselines, our method reduces CPU utilization by up to 42%, memory usage by 58%, and energy consumption by 48%. These improvements are obtained with negligible loss in model accuracy on MNIST and a slightly higher reduction in accuracy on CIFAR-10, consistent with its higher task complexity.

Hierarchical federated learning (HFL) improves scalability and communication efficiency through intermediate edge servers, yet energy and network constraints remain critical in heterogeneous settings. We propose two energy-aware client-selection methods: HEPS-ML, a supervised approach enabling autonomous participation based on energy, resources, performance, and data features; and HEPS-SCA, which leverages the sine cosine algorithm to balance accuracy, energy, and latency. Edge-server selection is handled by a central MAB using historical and network information. Evaluations on CIFAR-10 and MNIST, including calibrated energy modeling and ablation of MAB-based server selection, show 65–80% energy savings over baselines while preserving accuracy under non-IID distributions.

Federated Learning is a promising technique for providing distributed learning without clients disclosing their private data. In Hierarchical Federated Learning, edge servers partially aggregate the parameters of their connected clients’ models, improving scalability and reducing computational overhead on the central server. To speed up the convergence of the global model, only those clients with potential contributions to the model performance will participate in model training. This paper introduces a two-step client selection approach for hierarchical federated learning and three novel algorithms, which consider a large set of features in this selection and the client’s contributions to the model performance. Compared to selected baseline algorithms, the proposed client selection algorithms reduce CPU utilization by more than 50%, memory usage by 80%, and energy consumption by 50%.

This paper introduces a batch grooming algorithm for establishing connections with different deadlines to be torn down for Elastic Optical Networks with Space-Division Multiplexing (EON-SDM). The proposed algorithm creates batches of requests for lightpath establishment and postpones their establishment to groom the highest possible number of lightpaths. Each batch comprises a set of requests that can be allocated as a single request without guard bands separating their lightpaths and using only one transponder per batch, which leads to higher spectrum efficiency and lower energy consumption. Results show that the algorithm produces lower blocking and higher energy efficiency than the existing algorithms.

In Elastic Optical Networks with Space-Division Multiplexing (EON-SDM), dynamic allocation and deallocation of the spectrum can cause spectrum fragmentation. A possible solution is the use of proactive techniques to reduce the occurrence of fragmentation. In this paper, we propose a proactive algorithm for EON-SDM networks using Multi-Core Fibers (MCFs), which employs a core prioritization and quadrant ordering to allocate requests. Results show that our algorithm can effectively reduce the blocking of requests for connection establishment.

In Elastic Optical Networks with Space-Division Multiplexing (EON-SDM), dynamic allocation and de-allocation can generate spectrum fragmentation, increasing the blocking probability. Proactive solutions attempt to minimize or prevent future fragmentation occurrence by trying to find paths and blocks of slots to allocate. These solutions increase the chances of future connection requests to be allocated. This paper presents two proactive algorithms to avoid spectrum fragmentation in EON-SDMs. The proposed algorithms take into consideration the fragmentation state of the spectrum as well as potential bottleneck formation. Results demonstrate that our algorithms can reduce significantly the blocking probability while respecting the inter-core cross-talk constraint in Multi-Core Fiber (MCF).