arXiv subjects

Cedomir Stefanovic

Publications and source records attributed to Cedomir Stefanovic.

At least 19 recordsLinked to original sources

arXiv2026

Semi-Autonomous Prosthesis Control Empowered by 5G and Mobile Edge Computing

Prosthetic hands equipped with cameras can use computer vision to plan grasps automatically, reducing cognitive effort. However, running modern vision models on wearable devices is impractical due to power and processing constraints. We present the first prototype of a 5G-connected mobile edge computing (MEC)-enabled semi-autonomous prosthetic hand, which streams RGB-D images to an edge server for real-time grasp planning. Thirteen able-bodied participants performed pick-and-place tasks under six conditions: manual EMG control, on-device inference, wired Ethernet connectivity, and three 5G connectivity configurations (private 20 MHz network, private 100 MHz network, and a commercial 5G link) to the server. All network-based conditions performed similarly, achieving task times around 8.6 s (34% faster than manual control), failure rates of 20-38%, and 62% lower overall workload. On-device processing performed the worst with 10.3 s task time and a 76% failure rate due to slow embedded inference (3 fps vs. 6-20 fps over the network). Network latencies remained below 180 ms for private 5G and 270 ms for commercial 5G. All 5G configurations, including bandwidth-constrained and commercially variable networks, matched wired Ethernet performance while significantly outperforming both manual control and local processing, establishing 5G edge-offloading as a practical path to deploying compute-intensive prosthesis control.

eess.SY↗

arXiv2026

JEPA for AI-Native 6G: Predictive Representations and Open Challenges

Sixth-generation (6G) networks are moving toward AI-native operation, where learning modules are embedded across the radio access network (RAN), edge, and core. This transition requires learning from limited labels, heterogeneous wireless and network data, partial observations, non-stationary propagation, and latency-constrained control loops. Joint-embedding predictive architecture (JEPA) is a promising self-supervised paradigm for this setting because it predicts missing or future representations in latent space instead of reconstructing raw measurements or using contrastive negative samples. This article presents a wireless-oriented tutorial on JEPA for 6G intelligence. We define the JEPA training mechanism, describe how CSI, beam measurements, KPIs, topology graphs, and sensing observations can be tokenized and masked, and position the learned encoder as a predictive representation layer for RAN, O-RAN, edge, and core functions, with task-specific heads or controllers producing final decisions. Then we present an illustrative, beam-management case study suggesting that a wireless-aware target, specifically an auxiliary future beam-energy target during self-supervised pretraining, can improve label efficiency and robustness across shifted deployment conditions relative to a supervised source domain. Finally, we outline open challenges in multi-timescale prediction, action-conditioned modeling, distributed training, trustworthiness, efficient deployment, benchmarking, and standardization.

cs.LG↗

arXiv2026

Reliability Analysis of Slotted Aloha with Capture for an OWC-based IoT system

In this article, we consider a random access scheme for an indoor Internet of Things (IoT) framework that uses optical wireless communication (OWC). We focus on a Slotted ALOHA (SA)-based solution where a number of OWC IoT users contend to send data to a central OWC receiver. In any given slot, and for a randomly selected active user, we consider the reliability of decoding the user's data packet at the receiver. This is done by deriving the signal-to-noise-and-interference-ratio (SINR) statistics from a randomly chosen user and evaluating the probability that the user's SINR is below a given threshold. By placing our analysis in the context of an indoor OWC IoT uplink setup, and employing the standard OWC channel model, we investigate the trade-offs between the reliability and the OWC system parameters such as the cell area or the transmitter's semi-angle. We obtain valuable insights into the design of an SA-based random access solution for a typical indoor OWC cell.

cs.IT↗

arXiv2026

Empowering Embodied AI in 6G Networks: Architecture, Enablers, and Open Challenges

Embodied artificial intelligence (AI) is emerging as a key driver of the sixth-generation (6G) wireless networks by enabling agents that continuously perceive, communicate, and act in dynamic physical environments. Unlike conventional AI systems that process disembodied data, embodied agents such as robots, autonomous vehicles, and extended reality (XR) devices operate through closed-loop perception-communication-action (PCA) interactions, where communication performance directly affects physical behavior, control stability, and task success. However, existing AI-native wireless architectures remain largely connectivity-centric and are not designed to support task-driven embodied intelligence at large scale. Therefore, we present a holistic framework for embodied AI-native 6G systems, in which communication, sensing, computation, and control are jointly designed as a unified closed-loop infrastructure. We introduce a system-level PCA architecture, discuss key enabling technologies and representative applications, and highlight major open challenges in multimodal intelligence, edge-aware deployment, evaluation, trustworthiness, and practical implementation. Our central argument is that future 6G systems must evolve from intelligent communication platforms into active enablers of embodied physical intelligence.

cs.NI↗

arXiv2026

Statistical Analysis for Energy-Efficient Satellite Edge Computing with Latency Guarantees

Being able to provide latency guarantees for orbital edge computing applications through Low Earth Orbit (LEO) satellite constellations is a major milestone for their integration into 5G and 6G networks. However, achieving this is fundamentally challenged by the inherent randomness in both communication and computing latency, driven by complex network dynamics, satellite motion, and hardware variability. In this paper, we perform a statistical analysis of the latency of satellite edge computing using representative computing hardware and an object detection algorithm running on a satellite image dataset. The resulting model captures the trade-off between data availability and estimation uncertainty, enabling data-driven optimization methods to meet latency targets with statistical guarantees while minimizing energy consumption. Our results show that parametric estimation and quantile regression for the execution time of the image processing algorithms can be effectively combined with models for the communication latency to select an optimal GPU clock frequency. This achieves a 95% probability of meeting a $500$ ms end-to-end deadline while reducing energy consumption by more than 50% compared to a baseline that relies on a Chebyshev-Cantelli inequality to bound execution-time quantiles. The proposed framework is generalizable across satellite edge computing workloads and hardware platforms.

cs.NI↗

arXiv2026

Toward Efficient Deployment and Synchronization in Digital Twins-Empowered Networks

Digital twins (DTs) are envisioned as a key enabler of the cyber-physical continuum in future wireless networks. However, efficient deployment and synchronization of DTs in dynamic multi-access edge computing (MEC) environments remains challenging due to time-varying communication and computational resources. This paper investigates the joint optimization of DT deployment and synchronization in dynamic MEC environments. A deep reinforcement learning (DRL) framework is proposed for adaptive DT placement and association to minimize interaction latency between physical and digital entities. To ensure semantic freshness, an update scheduling policy is further designed to minimize the long-term weighted sum of the Age of Changed Information (AoCI) and the update cost. A relative policy iteration algorithm with a threshold-based structure is developed to derive the optimal policy. Simulation results show that the proposed methods achieve lower latency, enhanced information freshness, and reduced system cost compared with benchmark schemes

eess.SY↗

arXiv2026

Physical Layer Security in Finite Blocklength Massive IoT with Randomly Located Eavesdroppers

This paper analyzes the physical layer security performance of massive uplink Internet of Things (IoT) networks operating under the finite blocklength (FBL) regime. IoT devices and base stations (BS) are modeled using a stochastic geometry approach, while an eavesdropper is placed at a random location around the transmitting device. This system model captures security risks common in dense IoT deployments. Analytical expressions for the secure success probability, secrecy outage probability and secrecy throughput are derived to characterize how stochastic interference, fading and eavesdropper spatial uncertainty interact with FBL constraints in short packet uplink transmissions. Numerical results illustrate key system behavior under different network and channel conditions.

cs.IT↗

arXiv2025

Link-Aware Energy-Frugal Continual Learning for Fault Detection in IoT Networks

The use of lightweight machine learning (ML) models in internet of things (IoT) networks enables resource constrained IoT devices to perform on-device inference for several critical applications. However, the inference accuracy deteriorates due to the non-stationarity in the IoT environment and limited initial training data. To counteract this, the deployed models can be updated occasionally with new observed data samples. However, this approach consumes additional energy, which is undesirable for energy constrained IoT devices. This letter introduces an event-driven communication framework that strategically integrates continual learning (CL) in IoT networks for energy-efficient fault detection. Our framework enables the IoT device and the edge server (ES) to collaboratively update the lightweight ML model by adapting to the wireless link conditions for communication and the available energy budget. Evaluation on real-world datasets show that the proposed approach can outperform both periodic sampling and non-adaptive CL in terms of inference recall; our proposed approach achieves up to a 42.8% improvement, even under tight energy and bandwidth constraint.

cs.LG↗

arXiv2025

ISAC-Powered Distributed Matching and Resource Allocation in Multi-band NTN

Scalability is a major challenge in non-geostationary orbit (NGSO) satellite networks due to the massive number of ground users sharing the limited sub-6 GHz spectrum. Using K- and higher bands is a promising alternative to increase the accessible bandwidth, but these bands are subject to significant atmospheric attenuation, notably rainfall, which can lead to degraded performance and link outages. We present an integrated sensing and communications (ISAC)-powered framework for resilient and efficient operation of multi-band satellite networks. It is based on distributed mechanisms for atmospheric sensing, cell-to-satellite matching, and resource allocation (RA) in a 5G Non-Terrestrial Network (NTN) wide-area scenario with quasi-Earth fixed cells and a beam hopping mechanism. Results with a multi-layer multi-band constellation with satellites operating in the S- and K-bands demonstrate the benefits of our framework for ISAC-powered multi-band systems, which achieves 73% higher throughput per user when compared to single S- and K-band systems.

cs.NI↗

arXiv2025

Quantum Reinforcement Learning for 6G and Beyond Wireless Networks

While 5G is being deployed worldwide, 6G is receiving increasing attention from researchers to meet the growing demand for higher data rates, lower latency, higher density, and seamless communications worldwide. To meet the stringent requirements of 6G wireless communications networks, AI-integrated communications have become an indispensable part of supporting 6G systems with intelligence, automation, and big data training capabilities. However, traditional artificial intelligence (AI) systems are difficult to meet the stringent latency and high throughput requirements of 6G with limited resources. In this article, we summarize, analyze, discuss the potential, and benefits of Quantum Reinforcement Learning (QRL) in 6G. As an example, we show the superiority of QRL in dynamic spectrum access compared to the conventional Deep Reinforcement Learning (DRL) approach. In addition, we provide an overview of what DRL has accomplished in 6G and its challenges and limitations. From there, we introduce QRL and potential research directions that should continue to be of interest in 6G. To the best of our knowledge, this is the first review and vision article on QRL for 6G wireless communication networks.

cs.NI↗

arXiv2025

Using Visual Language Models to Control Bionic Hands: Assessment of Object Perception and Grasp Inference

This study examines the potential of utilizing Vision Language Models (VLMs) to improve the perceptual capabilities of semi-autonomous prosthetic hands. We introduce a unified benchmark for end-to-end perception and grasp inference, evaluating a single VLM to perform tasks that traditionally require complex pipelines with separate modules for object detection, pose estimation, and grasp planning. To establish the feasibility and current limitations of this approach, we benchmark eight contemporary VLMs on their ability to perform a unified task essential for bionic grasping. From a single static image, they should (1) identify common objects and their key properties (name, shape, orientation, and dimensions), and (2) infer appropriate grasp parameters (grasp type, wrist rotation, hand aperture, and number of fingers). A corresponding prompt requesting a structured JSON output was employed with a dataset of 34 snapshots of common objects. Key performance metrics, including accuracy for categorical attributes (e.g., object name, shape) and errors in numerical estimates (e.g., dimensions, hand aperture), along with latency and cost, were analyzed. The results demonstrated that most models exhibited high performance in object identification and shape recognition, while accuracy in estimating dimensions and inferring optimal grasp parameters, particularly hand rotation and aperture, varied more significantly. This work highlights the current capabilities and limitations of VLMs as advanced perceptual modules for semi-autonomous control of bionic limbs, demonstrating their potential for effective prosthetic applications.

cs.RO↗

arXiv2025

5G-Enabled Smart Prosthetic Hand: Connectivity Analysis and Assessment

In this paper, we demonstrate a proof-of-concept implementation of a framework for the development of edge-connected prosthetic systems. The framework is composed of a bionic hand equipped with a camera and connected to a Jetson device that establishes a wireless connection to the edge server, processing the received video stream and feeding back the inferred information about the environment. The hand-edge server connection is obtained either through a direct 5G link, where the edge server also functions as a 5G base station, or through a WiFi link. We evaluate the latency of closing the control loop in the system, showing that, in a realistic usage scenario, the connectivity and computation delays combined are well below 125 ms, which falls into the natural control range. To the best of our knowledge, this is the first analysis showcasing the feasibility of a 5G-enabled prosthetic system.

eess.SY↗

arXiv2025

Enabling Next-Generation Cloud-Connected Bionic Limbs Through 5G Connectivity

Despite the recent advancements in human-machine interfacing, contemporary assistive bionic limbs face critical challenges, including limited computational capabilities, high latency, and unintuitive control mechanisms, leading to suboptimal user experience and abandonment rates. Addressing these challenges requires a shift toward intelligent, interconnected solutions powered by advances in Internet of Things systems, particularly wireless connectivity and edge/cloud computing. This article presents a conceptual approach to transform bionic limbs by harnessing the pervasive connectivity of 5G and the significant computational power of cloud and edge servers, equipping them with capabilities not available hitherto. The system leverages a hierarchical distributed-computing architecture that integrates local, edge, and cloud computing layers. Time-critical tasks are handled by a local processing unit, while compute-intensive tasks are offloaded to edge and cloud servers, leveraging the high data rate, reliable and low latency capabilities of advanced cellular networks. We perform a proof-of-concept validation in a 5G testbed showing that such networks are capable of achieving data rates and fulfilling latency requirements for a natural prosthetic control, allowing for offloading of compute-intensive jobs to the edge/cloud servers. This is the first step towards the realization and real-world validation of cloud-connected bionic limb systems.

cs.NI↗

arXiv2025

On the Role of Early-Termination for Age of Information in Tree-Based Random Access Protocols

Age of Information (AoI) has emerged as a key metric for assessing data freshness in IoT applications, where a large number of devices report time-stamped updates to a monitor. Such systems often rely on random access protocols based on variations of ALOHA at the link layer, where collision resolution algorithms play a fundamental role to enable reliable delivery of packets. In this context, we provide the first analytical characterization of average AoI for the classical Capetanakis tree-based algorithm with gated access under exogenous traffic, capturing the protocol's dynamics, driven by sporadic packet generation and variable collision resolution times. We also explore a variant with early termination, where contention is truncated after a maximum number of slots even if not all users are resolved. The approach introduces a fundamental trade-off between reliability and timeliness, allowing stale packets to be dropped to improve freshness.

cs.IT↗

arXiv2025

Proactive Radio Resource Allocation for 6G In-Factory Subnetworks

6G In-Factory Subnetworks (InF-S) have recently been introduced as short-range, low-power radio cells installed in robots and production modules to support the strict requirements of modern control systems. Information freshness, characterized by the Age of Information (AoI), is crucial to guarantee the stability and accuracy of the control loop in these systems. However, achieving strict AoI performance poses significant challenges considering the limited resources and the high dynamic environment of InF-S. In this work, we introduce a proactive radio resource allocation approach to minimize the AoI violation probability. The proposed approach adopts a decentralized learning framework using Bayesian Ridge Regression (BRR) to predict the future AoI by actively learning the system dynamics. Based on the predicted AoI value, radio resources are proactively allocated to minimize the probability of AoI exceeding a predefined threshold, hence enhancing the reliability and accuracy of the control loop. The conducted simulation results prove the effectiveness of our proposed approach to improve the AoI performance where a reduction of 98% is achieved in the AoI violation probability compared to relevant baseline methods.

eess.SY↗

arXiv2025

Joint Beamforming and Trajectory Optimization for Multi-UAV-Assisted Integrated Sensing and Communication Systems

In this paper, we investigate beamforming design and trajectory optimization for a multi-unmanned aerial vehicle (UAV)-assisted integrated sensing and communication (ISAC) system. The proposed system employs multiple UAVs equipped with dual-functional radar-communication capabilities to simultaneously perform target sensing and provide communication services to users. We formulate a joint optimization problem that aims to maximize the sum rate of users while maintaining target sensing performance through coordinated beamforming and UAV trajectory design. To address this challenging non-convex problem, we develop a block coordinated descent (BCD)-based iterative algorithm that decomposes the original problem into tractable subproblems. Then, the beamforming design problem is addressed using fractional programming, while the UAV trajectory is refined through the deep deterministic policy gradient (DDPG) algorithm. The simulation results demonstrate that the proposed joint optimization approach achieves significant performance improvements in both communication throughput and sensing accuracy compared to conventional, separated designs. We also show that proper coordination of multiple UAVs through optimized trajectories and beamforming patterns can effectively balance the tradeoff between sensing and communication objectives.

cs.NI↗

arXiv2023

Slotted Aloha for Optical Wireless Communications in Internet of Underwater Things

In this work, we design and analyse a Slotted ALOHA (SA) solution for Optical Wireless Communication (OWC)-based Internet of Underwater Things (IoUT). In the proposed system, user devices exchange data with an access point (AP) which exploits the capture effect. The space spanned by the IoUT nodes is three-dimensional, i.e., users are located in half-sphere centered at the AP placed at the bottom of a floating object at the water surface level. The analytical expressions for the system throughput and reliability expressed in terms of the outage probability are derived. Based on the simulated signal-to-noise-and-interference-ratio statistics and derived analytical expressions, we present numerical results that investigate the trade-off between the system performance and the IoUT system parameters, such as the number of users, activation probability and type of water medium. The presented conclusions provide valuable insights into the design of an SA-based solution for IoUT communications.

cs.NI↗

arXiv2023

A TETRA-based System for Remote-Health Monitoring of First Responders: Peak AoI Assessment in Direct and Trunked Mode

In this paper, we study peak age of information (PAoI) performance of a novel IoT solution for remote health-monitoring of first responders over TErrestrial Trunked RAdio (TETRA) links. The solution features a set of sensors embedded in a smart garment that periodically record and send physiological parameters of first responders to a remote agent. The received data is analyzed by the remote agent, which feeds back notifications and warnings to the first responders in the form of electrotactile stimuli. The communication in the system is performed over the TETRA Short Data Service (SDS), which is the default option for the development of third-party applications and which has rather limited capabilities. The choice of the PAoI as the parameter of interest is motivated by its suitable to measure data freshness in IoT applications with periodic monitoring. We derive closed-form expressions of PAoI for different packet-management schemes allowed by the TETRA standard, and verify the analytical results through extensive simulations under varying message generation rates. Our results provide important insights on the expected PAoI performance, which can be used for the system design guidelines. To the best of our knowledge, this is the first work that analyzes AoI performance of TETRA networks.

cs.NI↗