Searcharxiv⌕ Search

arXiv subjects

Sebastian Troia

Publications and source records attributed to Sebastian Troia.

5 recordsLinked to original sources
arXiv2026

Generative Explainability for Next-Generation Networks: LLM-Augmented XAI with Mutual Feature Interactions

As artificial intelligence and machine learning (AI/ML) models become integral to network operations, their lack of transparency poses a significant barrier to operator trust. Existing explainable artificial intelligence (XAI) techniques often fail to bridge this gap for non-specialists, producing technical outputs that are difficult to translate into actionable insights. This paper presents a framework specifically designed to address this shortcoming. It leverages a moderately sized large language model (LLM) and extends beyond the standard use of SHapley Additive exPlanations (SHAP) feature influence values. The framework employs a structured prompt enriched with mutual feature interaction data to generate human-understandable natural language explanations. To validate our framework, we performed an empirical evaluation on an optical quality of transmission (QoT) estimation use case with human evaluators. We collected independent performance evaluations from specialists, which showed a high inter-evaluator agreement. Compared to a state-of-the-art baseline that uses only SHAP feature influence values in a straightforward prompt, our approach improves the explanation usefulness and scope by 12.2% and 6.2%, while achieving 97.5% correctness.

cs.NI↗
arXiv2025

Exploiting the Alternatives: Coordinated Learning via Hierarchical RL for Dynamic VNEAP

Virtual Network Embedding (VNE) is a key enabler of network slicing, yet most formulations assume that each Virtual Network Request (VNR) has a fixed topology. Recently, VNE with Alternatives (VNEAP) was introduced to capture malleable VNRs, where each request can be instantiated using one of several functionally equivalent topologies that trade resources differently. This flexibility can improve embedding feasibility, but only if the orchestrator can jointly select suitable alternatives and embed them under dynamic arrivals. This paper proposes HRL-VNEAP, a hierarchical reinforcement learning approach for dynamic VNEAP. A high-level policy selects the most suitable alternative topology (or rejects the request), and a low-level policy embeds the chosen topology onto the substrate network. Experiments on realistic substrate topologies under varying arrival rates show that naive exploitation strategies provide only modest gains, whereas HRL-VNEAP outperforms state of the art approaches, improving acceptance ratio by up to 22%, and net profit by up to 20%. An offline MILP upper bound is also used on tractable instances to quantify the remaining optimality gap.

cs.NI↗
arXiv2025

On the Optimization of Model Aggregation for Federated Learning at the Network Edge

The rapid increase in connected devices has signifi- cantly intensified the computational and communication demands on modern telecommunication networks. To address these chal- lenges, integrating advanced Machine Learning (ML) techniques like Federated Learning (FL) with emerging paradigms such as Multi-access Edge Computing (MEC) and Software-Defined Wide Area Networks (SD-WANs) is crucial. This paper intro- duces online resource management strategies specifically designed for FL model aggregation, utilizing intermediate aggregation at edge nodes. Our analysis highlights the benefits of incorporating edge aggregators to reduce network link congestion and maximize the potential of edge computing nodes. However, the risk of network congestion persists. To mitigate this, we propose a novel aggregation approach that deploys an aggregator overlay network. We present an Integer Linear Programming (ILP) model and a heuristic algorithm to optimize the routing within this overlay network. Our solution demonstrates improved adapt- ability to network resource utilization, significantly reducing FL training round failure rates by up to 15% while also alleviating cloud link congestion.

cs.NI↗
arXiv2025

Pair-Bid Auction Model for Optimized Network Slicing in 5G RAN

Network slicing is a key 5G technology that enables multiple virtual networks to share physical infrastructure, optimizing flexibility and resource allocation. This involves Mobile Network Operators (MNO), Mobile Virtual Network Operators (MVNOs), and end users, where MNO leases network slices to MVNOs, and then provides customized services. This work considers end-to-end network slicing with a focus on fair sharing and financial-related power efficiency, modeled as a two level hierarchical combinatorial auction. At the upper level, an MNO auctions slices to competing MVNOs, while at the lower level, MVNOs allocate resources to end users through their own auctions. Dynamic user requests add complexity to the process. Our model optimizes resource allocation and revenue generation using a pair-bid mechanism and Vickrey-Clarke-Groves (VCG) pricing. The pair-bid approach enhances competition and efficiency, while VCG ensures truthful bidding based on marginal system impact. Simulations validate the model's effectiveness in resource distribution and financial performance, showing a 12.5% revenue improvement over the baseline.

cs.NI↗