arXiv subjects

Iordanis Koutsopoulos

Publications and source records attributed to Iordanis Koutsopoulos.

18 recordsLinked to original sources

arXiv2026

Learning to Fine-tune Foundation Models under Resource Limitations

We study the problem of optimal continual fine-tuning for a pre-trained Foundation Model deployed at a resource-limited device. At each time slot, a new batch of training data arrives, and the controller is faced with two options: either use the data to fine-tune the model and incur a compute cost, or do not fine-tune the model and discard the data. After the decision, the performance of the current model is measured in terms of an application-specific performance metric such as classification accuracy. Our objective is to learn an optimal policy that determines \emph{when to fine-tune the model} on a single task (e.g., sentiment analysis), under a finite compute budget. We formulate this online decision-making problem as a constrained Markov Decision Process, where the system state captures three essential aspects: (\textit{i}) model's performance, (\textit{ii}) computational budget, and (\textit{iii}) data distribution relevance to historic data encountered up to that point. The transition to the next state is stochastic and therefore, we propose a reinforcement learning-based method to solve this problem, namely the \emph{actor-critic} algorithm. We also consider the special case where the performance of fine-tuning for a given model can be predicted or estimated prior to decision; in this case the problem becomes a Dynamic Programming one. Experiments with a large pre-trained model on a widely-used text classification dataset demonstrate that our method consistently outperforms fine-tuning approaches with the same compute budget by more than $4\%$ in terms of accuracy and achieves $97\%$ of full-parameter fine-tuning accuracy while requiring only $25\%$ of the fine-tuning steps.

cs.LG↗

arXiv2026

Split Federated Learning Architectures for High-Accuracy and Low-Delay Model Training

Can we find a network architecture for ML model training so as to optimize training loss (and thus, accuracy) in Split Federated Learning (SFL)? And can this architecture also reduce training delay and communication overhead? While accuracy is not influenced by how we split the model in ordinary, state-of-the-art SFL, in this work we answer the questions above in the affirmative. Recent Hierarchical SFL (HSFL) architectures adopt a three-tier training structure consisting of clients, (local) aggregators, and a central server. In this architecture, the model is partitioned at two partitioning layers into three sub-models, which are executed across the three tiers. Despite their merits, HSFL architectures overlook the impact of the partitioning layers and client-to-aggregator assignments on accuracy, delay, and overhead. This work explicitly captures the impact of the partitioning layers and client-to-aggregator assignments on accuracy, delay and overhead by formulating a joint optimization problem. We prove that the problem is NP-hard and propose the first accuracy-aware heuristic algorithm that explicitly accounts for model accuracy, while remaining delay-efficient. Simulation results on public datasets show that our approach can improve accuracy by 3%, while reducing delay by 20% and overhead by 50%, compared to state-of-the-art SFL and HSFL schemes.

cs.LG↗

arXiv2025

Video Quality Monitoring for Remote Autonomous Vehicle Control

The delivery of high-quality, low-latency video streams is critical for remote autonomous vehicle control, where operators must intervene in real time. However, reliable video delivery over Fourth/Fifth-Generation (4G/5G) mobile networks is challenging due to signal variability, mobility-induced handovers, and transient congestion. In this paper, we present a comprehensive blueprint for an integrated video quality monitoring system, tailored to remote autonomous vehicle operation. Our proposed system includes subsystems for data collection onboard the vehicle, video capture and compression, data transmission to edge servers, real-time streaming data management, Artificial Intelligence (AI) model deployment and inference execution, and proactive decision-making based on predicted video quality. The AI models are trained on a hybrid dataset that combines field-trial measurements with synthetic stress segments and covers Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and encoder-only Transformer architectures. As a proof of concept, we benchmark 20 variants from these model classes together with feed-forward Deep Neural Network (DNN) and linear-regression baselines, reporting accuracy and inference latency. Finally, we study the trade-offs between onboard and edge-based inference. We further discuss the use of explainable AI techniques to enhance transparency and accountability during critical remote-control interventions. Our proactive approach to network adaptation and Quality of Experience (QoE) monitoring aims to enhance remote vehicle operation over next-generation wireless networks.

cs.NI↗

arXiv2025

Large Language Model Partitioning for Low-Latency Inference at the Edge

Large Language Models (LLMs) based on autoregressive, decoder-only Transformers generate text one token at a time, where a token represents a discrete unit of text. As each newly produced token is appended to the partial output sequence, the length grows and so does the memory and compute load, due to the expanding key-value caches, which store intermediate representations of all previously generated tokens in the multi-head attention (MHA) layer. As this iterative process steadily increases memory and compute demands, layer-based partitioning in resource-constrained edge environments often results in memory overload or high inference latency. To address this and reduce inference latency, we propose a resource-aware Transformer architecture partitioning algorithm, where the partitioning decision is updated at regular intervals during token generation. The approach is myopic in that it is based on instantaneous information about device resource availability and network link bandwidths. When first executed, the algorithm places blocks on devices, and in later executions, it migrates these blocks among devices so that the sum of migration delay and inference delay remains low. Our approach partitions the decoder at the attention head level, co-locating each attention head with its key-value cache and allowing dynamic migrations whenever resources become tight. By allocating different attention heads to different devices, we exploit parallel execution of attention heads and thus achieve substantial reductions in inference delays. Our experiments show that in small-scale settings (3-5 devices), the proposed method achieves within 15 to 20 percent of an exact optimal solver's latency, while in larger-scale tests it achieves notable improvements in inference speed and memory usage compared to state-of-the-art layer-based partitioning approaches.

cs.DC↗

arXiv2025

Collaborative Split Federated Learning with Parallel Training and Aggregation

Federated learning (FL) operates based on model exchanges between the server and the clients, and it suffers from significant client-side computation and communication burden. Split federated learning (SFL) arises a promising solution by splitting the model into two parts, that are trained sequentially: the clients train the first part of the model (client-side model) and transmit it to the server that trains the second (server-side model). Existing SFL schemes though still exhibit long training delays and significant communication overhead, especially when clients of different computing capability participate. Thus, we propose Collaborative-Split Federated Learning~(C-SFL), a novel scheme that splits the model into three parts, namely the model parts trained at the computationally weak clients, the ones trained at the computationally strong clients, and the ones at the server. Unlike existing works, C-SFL enables parallel training and aggregation of model's parts at the clients and at the server, resulting in reduced training delays and commmunication overhead while improving the model's accuracy. Experiments verify the multiple gains of C-SFL against the existing schemes.

cs.DC↗

arXiv2025

Explainability and Continual Learning meet Federated Learning at the Network Edge

As edge devices become more capable and pervasive in wireless networks, there is growing interest in leveraging their collective compute power for distributed learning. However, optimizing learning at the network edge entails unique challenges, particularly when moving beyond conventional settings and objectives. While Federated Learning (FL) has emerged as a key paradigm for distributed model training, critical challenges persist. First, existing approaches often overlook the trade-off between predictive accuracy and interpretability. Second, they struggle to integrate inherently explainable models such as decision trees because their non-differentiable structure makes them not amenable to backpropagation-based training algorithms. Lastly, they lack meaningful mechanisms for continual Machine Learning (ML) model adaptation through Continual Learning (CL) in resource-limited environments. In this paper, we pave the way for a set of novel optimization problems that emerge in distributed learning at the network edge with wirelessly interconnected edge devices, and we identify key challenges and future directions. Specifically, we discuss how Multi-objective optimization (MOO) can be used to address the trade-off between predictive accuracy and explainability when using complex predictive models. Next, we discuss the implications of integrating inherently explainable tree-based models into distributed learning settings. Finally, we investigate how CL strategies can be effectively combined with FL to support adaptive, lifelong learning when limited-size buffers are used to store past data for retraining. Our approach offers a cohesive set of tools for designing privacy-preserving, adaptive, and trustworthy ML solutions tailored to the demands of edge computing and intelligent services.

cs.LG↗

arXiv2025

Joint Explainability-Performance Optimization With Surrogate Models for AI-Driven Edge Services

Explainable AI is a crucial component for edge services, as it ensures reliable decision making based on complex AI models. Surrogate models are a prominent approach of XAI where human-interpretable models, such as a linear regression model, are trained to approximate a complex (black-box) model's predictions. This paper delves into the balance between the predictive accuracy of complex AI models and their approximation by surrogate ones, advocating that both these models benefit from being learned simultaneously. We derive a joint (bi-level) training scheme for both models and we introduce a new algorithm based on multi-objective optimization (MOO) to simultaneously minimize both the complex model's prediction error and the error between its outputs and those of the surrogate. Our approach leads to improvements that exceed 99% in the approximation of the black-box model through the surrogate one, as measured by the metric of Fidelity, for a compromise of less than 3% absolute reduction in the black-box model's predictive accuracy, compared to single-task and multi-task learning baselines. By improving Fidelity, we can derive more trustworthy explanations of the complex model's outcomes from the surrogate, enabling reliable AI applications for intelligent services at the network edge.

cs.LG↗

arXiv2024

Optimal Fidelity-Aware Entanglement Distribution in Linear Quantum Networks

We study the problem of entanglement distribution in terms of maximizing a utility function that captures the total fidelity of end-to-end entanglements in a two-link linear quantum network with a source, a repeater, and a destination. The nodes have several quantum memories, and the problem is how to coordinate entanglement purification in each of the links, and entanglement swapping across links, so as to achieve the goal above. We show that entanglement swapping (i.e, deciding on the pair of qubits from each link to perform swapping on) is equivalent to finding a max-weight matching on a bipartite graph. Further, entanglement purification (i.e, deciding which pairs of qubits in a link will undergo purification) is equivalent to finding a max-weight matching on a non-bipartite graph. We propose two polynomial algorithms, the Purify-then-Swap (PtS) and the Swap-then-Purify (StP) ones, where the decisions about purification and swapping are taken with different order. Numerical results show that PtS performs better than StP, and also that the omission of purification in StP gives substantial benefits.

quant-ph↗

arXiv2022

Personalized Federated Learning with Exact Stochastic Gradient Descent

We propose a Stochastic Gradient Descent (SGD)-type algorithm for Personalized Federated Learning which can be particularly attractive for mobile energy-limited regimes due to its low per-client computational cost. The model to be trained includes a set of common weights for all clients, and a set of personalized weights that are specific to each client. At each optimization round, randomly selected clients perform multiple full gradient-descent updates over their client-specific weights towards optimizing the loss function on their own datasets, without updating the common weights. This procedure is energy-efficient since it has low computational cost per client. At the final update of each round, each client computes the joint gradient over both the client-specific and the common weights and returns the gradient of common weights to the server, which allows to perform an exact SGD step over the full set of weights in a distributed manner. For the overall optimization scheme, we rigorously prove convergence, even in non-convex settings such as those encountered when training neural networks, with a rate of $\mathcal{O} \left (\frac{1}{\sqrt{T}} \right )$ with respect to communication rounds $T$. In practice, PFLEGO exhibits substantially lower per-round wall-clock time, used as a proxy for energy. Our theoretical guarantees translate to superior performance in practice against baselines such as FedAvg and FedPer, as evaluated in several multi-class classification datasets, in particular, Omniglot, CIFAR-10, MNIST, Fashion-MNIST, and EMNIST.

cs.LG↗

arXiv2021

Blind Optimal User Association in Small-Cell Networks

We learn optimal user association policies for traffic from different locations to Access Points(APs), in the presence of unknown dynamic traffic demand. We aim at minimizing a broad family of $α$-fair cost functions that express various objectives in load assignment in the wireless downlink, such as total load or total delay minimization. Finding an optimal user association policy in dynamic environments is challenging because traffic demand fluctuations over time are non-stationary and difficult to characterize statistically, which obstructs the computation of cost-efficient associations. Assuming arbitrary traffic patterns over time, we formulate the problem of online learning of optimal user association policies using the Online Convex Optimization (OCO) framework. We introduce a periodic benchmark for OCO problems that generalizes state-of-the-art benchmarks. We exploit inherent properties of the online user association problem and propose PerOnE, a simple online learning scheme that dynamically adapts the association policy to arbitrary traffic demand variations. We compare PerOnE against our periodic benchmark and prove that it enjoys the no-regret property, with additional sublinear dependence of the network size. To the best of our knowledge, this is the first work that introduces a periodic benchmark for OCO problems and a no-regret algorithm for the online user association problem. Our theoretical findings are validated through results on a real-trace dataset.

cs.NI↗

arXiv2017

Community Networks and Sustainability: a Survey of Perceptions, Practices, and Proposed Solutions

Community network (CN) initiatives have been around for roughly two decades, evangelizing a distinctly different paradigm for building, maintaining, and sharing network infrastructure but also defending the basic human right to Internet access. Over this time they have evolved into a mosaic of systems that vary widely with respect to their network technologies, their offered services, their organizational structure, and the way they position themselves in the overall telecommunications' ecosystem. Common to all these highly differentiated initiatives is the sustainability challenge. We approach sustainability as a broad term with an economical, political, and cultural context. We first review the different perceptions of the term. These vary both across and within the different types of stakeholders involved in CNs and are reflected in their motivation to join such initiatives. Then, we study the diverse ways that CN operators pursue the sustainability goal. Depending on the actual context of the term, these range all the way from mechanisms to fund their activities and synergistic approaches with commercial service providers, to organizational structures and social activities that serve as incentives to maximize the engagement of their members. Finally, we iterate and discuss theoretical concepts of incentive mechanisms that have been proposed in the literature for these networks as well as implemented tools and processes designed to set the ground for CN participation. While, theoretical mechanisms leverage game theory, reputation frameworks, and social mechanisms, implemented mechanisms focus on organizational matters, education and services, all aiming to motivate the active and sustained participation of users and other actors in the CN.

cs.NI↗

arXiv2016

Streaming Big Data meets Backpressure in Distributed Network Computation

We study network response to queries that require computation of remotely located data and seek to characterize the performance limits in terms of maximum sustainable query rate that can be satisfied. The available resources include (i) a communication network graph with links over which data is routed, (ii) computation nodes, over which computation load is balanced, and (iii) network nodes that need to schedule raw and processed data transmissions. Our aim is to design a universal methodology and distributed algorithm to adaptively allocate resources in order to support maximum query rate. The proposed algorithms extend in a nontrivial way the backpressure (BP) algorithm to take into account computations operated over query streams. They contribute to the fundamental understanding of network computation performance limits when the query rate is limited by both the communication bandwidth and the computation capacity, a classical setting that arises in streaming big data applications in network clouds and fogs.

cs.NI↗

arXiv2013

Evolution of IEEE 802.11 compatible standards and impact on Energy Consumption

Over the last decade, the IEEE 802.11 has emerged as the most popular protocol in the wireless domain. Since the release of the first standard version, several amendments have been introduced in an effort to improve its throughput performance, with the most recent one being the IEEE 802.11n extension. In this work, we present detailed experimentally obtained measurements that evaluate the energy efficiency of the base standard in comparison with the latest 802.11n version. Moreover, we investigate the impact of various MAC layer enhancements, both vendor specific and standard compliant ones, on the energy consumption of wireless transceivers and total nodes as well. Results obtained under a wide range of settings, indicate that the latest standard enables reduction of energy expenditure, by more than 75%, when combined with innovative frame aggregation mechanisms.

cs.NI↗

arXiv2013

Optimal Primary-Secondary user Cooperation Policies in Cognitive Radio Networks

In cognitive radio networks, secondary users (SUs) may cooperate with the primary user (PU), so that the success probability of PU transmissions are improved, while SUs obtain more transmission opportunities. Thus, SUs have to take intelligent decisions on whether to cooperate or not and with what power level, in order to maximize their throughput subject to average power constraints. Cooperation policies in this framework require the solution of a constrained Markov decision problem with infinite state space. In our work, we restrict attention to the class of stationary policies that take randomized decisions in every time slot based only on spectrum sensing. The proposed class of policies is shown to achieve the same set of SU rates as the more general policies, and enlarge the stability region of PU queue. Moreover, algorithms for the distributed calculation of the set of probabilities used by the proposed class of policies are presented.

cs.NI↗

arXiv2011

Competition and Regulation in Wireless Services Markets

We consider a wireless services market where a set of operators compete for a large common pool of users. The latter have a reservation utility of U0 units or, equivalently, an alternative option to satisfy their communication needs. The operators must satisfy these minimum requirements in order to attract the users. We model the users decisions and interaction as an evolutionary game and the competition among the operators as a non cooperative price game which is proved to be a potential game. For each set of prices selected by the operators, the evolutionary game attains a different stationary point. We show that the outcome of both games depend on the reservation utility of the users and the amount of spectrum W the operators have at their disposal. We express the market welfare and the revenue of the operators as functions of these two parameters. Accordingly, we consider the scenario where a regulating agency is able to intervene and change the outcome of the market by tuning W and/or U0. Different regulators may have different objectives and criteria according to which they intervene. We analyze the various possible regulation methods and discuss their requirements, implications and impact on the market.

cs.NI↗

arXiv2011

Incentive Mechanisms for Hierarchical Spectrum Markets

In this paper, we study spectrum allocation mechanisms in hierarchical multi-layer markets which are expected to proliferate in the near future based on the current spectrum policy reform proposals. We consider a setting where a state agency sells spectrum channels to Primary Operators (POs) who subsequently resell them to Secondary Operators (SOs) through auctions. We show that these hierarchical markets do not result in a socially efficient spectrum allocation which is aimed by the agency, due to lack of coordination among the entities in different layers and the inherently selfish revenue-maximizing strategy of POs. In order to reconcile these opposing objectives, we propose an incentive mechanism which aligns the strategy and the actions of the POs with the objective of the agency, and thus leads to system performance improvement in terms of social welfare. This pricing-based scheme constitutes a method for hierarchical market regulation. A basic component of the proposed incentive mechanism is a novel auction scheme which enables POs to allocate their spectrum by balancing their derived revenue and the welfare of the SOs.

cs.NI↗

arXiv2011

The impact of energy constraints on the medium access

Contemporary mobile devices are battery powered and due to their shrinking size and increasing complexity operate on a tight energy budget. Thus, energy consumption is becoming one of the major concerns regarding the current and upcoming wireless communication systems. On the other hand, the available bandwidth resources are limited and modern applications are throughput demanding, leading thus to strong competition for the medium. In this direction, we consider a stochastic contention based medium access scheme, where the devices may choose to turn off for some time in order to save energy. We perform an analysis for a slotted ALOHA scenario and we show that the energy constraints, if properly exploited, may reduce contention for the medium. Our results give valuable insights on the energy--throughput tradeoff for any contention based system.

cs.OH↗

arXiv2010

Control and Optimization Meet the Smart Power Grid - Scheduling of Power Demands for Optimal Energy Management

The smart power grid aims at harnessing information and communication technologies to enhance reliability and enforce sensible use of energy. Its realization is geared by the fundamental goal of effective management of demand load. In this work, we envision a scenario with real-time communication between the operator and consumers. The grid operator controller receives requests for power demands from consumers, with different power requirement, duration, and a deadline by which it is to be completed. The objective is to devise a power demand task scheduling policy that minimizes the grid operational cost over a time horizon. The operational cost is a convex function of instantaneous power consumption and reflects the fact that each additional unit of power needed to serve demands is more expensive as demand load increases.First, we study the off-line demand scheduling problem, where parameters are fixed and known. Next, we devise a stochastic model for the case when demands are generated continually and scheduling decisions are taken online and focus on long-term average cost. We present two instances of power consumption control based on observing current consumption. First, the controller may choose to serve a new demand request upon arrival or to postpone it to the end of its deadline. Second, the additional option exists to activate one of the postponed demands when an active demand terminates. For both instances, the optimal policies are threshold based. We derive a lower performance bound over all policies, which is asymptotically tight as deadlines increase. We propose the Controlled Release threshold policy and prove it is asymptotically optimal. The policy activates a new demand request if the current power consumption is less than a threshold, otherwise it is queued. Queued demands are scheduled when their deadline expires or when the consumption drops below the threshold.

cs.NI↗