This paper presents a novel framework for constructing channel knowledge maps with high accuracy and computational efficiency. The proposed radio propagation graph representation learning (RPGRL) framework integrates physicsbased modeling and data-driven learning through end-to-end optimization using only measured received power data. In RPGRL, both propagation probability and received power estimations are jointly optimized without requiring explicit path supervision, enabling adaptive graph-based representations that account for uncertainties in reflections caused by surface scattering and angular errors introduced by three-dimensional building map quantization. Experimental evaluations using real-world signal measurement data collected in an urban environment demonstrate that RPGRL significantly improves received power estimation accuracy while reducing computation time by more than 70% compared to conventional ray tracing. Furthermore, it outperforms existing deep learning-based methods, such as RadioUNet, by leveraging path-aware estimation grounded in radio propagation physics. These results highlight RPGRL as a promising and practical approach for environment-aware communication and intelligent wireless network design toward 6G.

Robust object detection in adverse weather conditions is critical for ensuring the safety and reliability of autonomous driving systems. In this work, we present a detailed study on the adversarial robustness of YOLO-based detectors using the RealDriveSim dataset, which includes foggy, rainy, and nighttime scenarios. We benchmark YOLOv9 and YOLOv10 under clean conditions and observe high performance, with YOLOv10 achieving a mean average precision (mAP) of 69.6%. To evaluate vulnerability, we introduce an adversarial patch optimized to suppress road object detections. After patch-based perturbation, mAP drops to 44.3%, highlighting the importance of a defense system. To counter this degradation, we propose a lightweight LiDAR-camera fusion framework that does not require model retraining or architectural changes. Our method projects 3D LiDAR point clouds into the 2D image plane using intrinsic and extrinsic calibration parameters and cross-validates each 2D detection by checking for supporting 3D LiDAR points within its bounding box with an inference time of only 7.2 ms. Our fusion strategy effectively filters adversarial false positives, leading to a recovery in mAP to 62.9%, without requiring model retraining or architectural changes. To the best of our knowledge, this is the first work to benchmark adversarial robustness and sensor-level fusion defense on the RealDriveSim dataset, setting a new standard for evaluating real-world physical attack resilience in autonomous perception.

Machine learning (ML) has become essential for tasks like detection and classification in autonomous vehicles (AVs). However, ML models are vulnerable to adversarial attacks, which can weaken passenger trust and raise safety concerns in autonomous driving systems. This is especially critical in systems like traffic sign recognition (TSR), where a misclassification caused by an adversarial attack could lead to serious safety risks. This research work explored the vulnerabilities of TSR models to adversarial attacks focusing on projected gradient descent (PGD) and the fast gradient sign method (FGSM). An adversarial attack pipeline is proposed that leverages ScoreCAM-based region-of-interest (ROI) localization to enhance the effectiveness of these attacks. Adversarial attacks manipulate the input data to mislead the models, achieving a high attack success rate (ASR) by exploiting their vulnerabilities. Experimental results on multiple models such as VGG19, convolutional neural network (CNN), ResNet50 and vision transformers (ViT) demonstrate significant increases in ASR. For instance, our method achieved a 97.67% ASR using PGD on VGG19 and a 95.89% ASR using FGSM on the same model, marking a considerable performance gain over traditional approaches. Moreover, these results are achieved with high computational efficiency, with average query times as low as 69.8 milliseconds.

The deployment of Internet of remote things (IoRT) devices in remote areas with insufficient communication infrastructure can employ the unmanned aerial vehicles (UAVs) for data collection. The previous works only considered the IoRT devices information coverage probability, but ignored the IoRT devices energy coverage probability in the UAV networks. This letter analyzes the joint information-energy coverage probability performance in UAV networks with hybrid energy harvesting (EH). Firstly, the closed-form expressions of the information coverage probability and the energy coverage probability are derived by utilizing the Laplace transform and the Campbell theorem, respectively. On this basis, the closed-form expression of the joint information-energy coverage probability is derived by the law of large numbers (LLN). Finally, the numerical results confirm the validity of the joint information-energy coverage probability performance.

In this paper, we investigate the performance analysis for intelligent reflecting surface (IRS)-assisted simultaneous wireless information and power transfer (SWIPT) with a geometry-based optimized phase shift under the spatially correlated Rician fading channels. In the geometry-based phase shift, the IRS can decide the optimal phase shift without channel information, thereby reducing the overhead of channel estimation. Our proposed closed-form expression can evaluate the harvested energy and data rate with consideration of the physical features of the transmitter, receiver, and IRS, i.e., antenna patterns and reflection patterns. We validate the proposed performance analysis by comparing it with the Monte Carlo simulations.

Deep joint source-channel coding is a promising solution for wireless image transmission systems because it can jointly optimize image compression and channel coding functionalities based on the actual image domain knowledge. However, the existing works ignore the time-selective fading effect which is a key characteristic of mobile image transmission systems. The paper, therefore, investigates a compression method to deal with the time-selective fading effect. Further, we analyze the optimal compression rate and show that the deep joint source-channel coding with optimal compression rate improves the image quality by up to 5.26 dB.

This paper proposes a novel model for radio propagation graph representation learning in multi-hop path representation. A concept of radio propagation graph representation learning has been proposed to express the radio propagation process using graph data structure, thereby improving the prediction accuracy with small computation time. In the existing work, a direct path representation model based on feedforward neural network (FFNN) has been proposed; however, it achieves poor prediction accuracy at points far from a transmitter due to the lack of consideration of reflection, i.e., multi-hop path. Therefore, this paper proposes a novel learning method to apply the existing concept to multi-hop path cases. Through the performance evaluation using Long Term Evolution signal data in a 2120 MHz band measured in an urban area, we demonstrate that the proposed model outperforms the existing representation model in terms of prediction accuracy and achieves shorter computation time compared with the ray tracing application.

Edge-AI offloading is a promising solution for real-time object detection with resource-constrained edge devices. However, current edge architecture, which uses conventional digital video compression and modulation/coding, cannot provide stable object detection due to the scarcity of wireless resources and the explosive increase of edge devices. We employ pseudo-analog video transmission for Edge-AI offloading systems to cope with the issue. Although pseudo-analog video transmission cannot provide clear images, it has a substantial advantage for Edge-AI offloading systems, i.e., it can transmit higher-resolution video with limited bandwidth in lower Signal-to-Noise Ratio (SNR) links and do not need to control resolution according to wireless channel quality. This paper explores the impact of pseudo-analog video transmission on object detection accuracy.

In this paper, we analyze performance of an intelligent reflecting surface (IRS)-assisted simultaneous wireless information and power transfer (SWIPT) system with the optimal phase shift. Specifically, we consider a transmitter sends power and information signals with the assistance of an IRS and spatially correlated fading channels. In practice, the channel between the transmitter and the IRS and between IRS and the receiver are spatially correlated, which constitutes a challenge for accurate performance analysis. In the system, we derive an optimal phase shift, in which the main lobe of the reflected signal at the IRS is directed to the receiver. Then, we develop a closed-form expression to evaluate the average harvested energy and information outage probability. We validate that the proposed model via Monte Carlo simulation.

This paper proposes a novel graph-learning-based radio propagation model, referred to as propagation graph representation learning. Recent advancements in deep learning have succeeded in developing site-specific path loss prediction; however, predicting shadowing without on-site measurement data is still a critical challenge. Propagation graph representation learning aims to express the site-specific propagation process. In the envisioned graph, nodes represent a transmitter, receivers, and obstructions, while edges represent propagation conditions between nodes. The graph structure enables us to recognize the reflection, diffraction, and shielding for accurate shadowing estimation. We also propose a feedforward neural network (FFNN) based representation model in a direct path scenario. Through the simulation using actual datasets in urban areas, we demonstrate that the proposal achieves twenty times faster computation than ray tracing while predicting shadowing well.

Wireless power transfer (WPT) is a promising solution for wireless sensor networks in IoT eras. However, due to the limitation of the spectrum, we need to address the research challenge of scheduling for both the power transfer and data transmission to maximize the spectrum efficiency. In this paper, we investigate harvest-then-transmit-based time division multiple access (TDMA) protocol with statistical channel state information. We show that the proposed protocol has the optimal time slot length to minimize the data transmission delay. Numerical results show that the proposed protocol achieves high spectrum efficiency and improve the battery charging ratio, compared with the existing harvest-then-transmit-based protocol.

Edge Cloud Computing is a key technology for enhancing mobile functionalities and real-time applications in devices with limited resources. This is done by sharing the resources of edge servers and offloading jobs to the edge cloud. In order to ensure a high-quality service and more efficient usage of resources, it is important not only to correctly configure the edge servers but also to carefully select where to deploy them. However, in Edge Cloud Computing there is a high amount of servers and, with the advent of 5G and Internet of Things, there will be a massive number of client devices as well. This would make the edge server deployment too complex to solve through convex techniques. In this situation, Machine Learning is the most appropriate approach. In this paper, we provide a deep analysis of the usage of k-Means Clustering and Particle Swarm Optimization in the edge cloud deployment problem. Our results show that the hyperparameters for these algorithms can significantly impact their running time as well as the efficiency of their results. Finally, we also provide how to best configure these algorithms for this specific problem.

This paper addresses the problem of cell zooming in green networks. We apply a state-of-the-art methodology, distributed optimization, to solve the privacy-preserved cell zooming issue in green networks. To this end, we approximate the cell zooming problem by a certain convex problem. We also develop a data masking technique, under which we can still obtain a correct solution of the convex problem. The proposed method outperforms a Q-learning-based approach in terms of energy efficiency and the charging rate of batteries.

This paper addresses the real-time control of transmission power for small cell base stations (SBSs) exploiting energy- harvesting sources. We employ model predictive control and optimize an objective function that contains the number of users with a given quality of experience and the average state of charge. We first determine the number of active SBSs in the viewpoint of energy efficiency and then approximate the objective function. Finally, we illustrate the proposed method through a numerical example, comparing it with a static method based on statistical information.

Mobile devices have several restrictions due to design choices that guarantee their mobility. A way of surpassing such limitations is to utilize cloud servers called cloudlets on the edge of the network through Mobile Edge Computing. However, as the number of clients and devices grows, the service must also increase its scalability in order to guarantee a latency limit and quality threshold. This can be achieved by deploying and activating more cloudlets, but this solution is expensive due to the cost of the physical servers. The best choice is to optimize the resources of the cloudlets through an intelligent choice of configuration that lowers delay and raises scalability. Thus, in this paper we propose an algorithm that utilizes Virtual Machine Migration and Transmission Power Control, together with a mathematical model of delay in Mobile Edge Computing and a heuristic algorithm called Particle Swarm Optimization, to balance the workload between cloudlets and consequently maximize cost-effectiveness. Our proposal is the first to consider simultaneously communication, computation, and migration in our assumed scale and, due to that, manages to outperform other conventional methods in terms of number of serviced users.

With files proactively stored at base stations (BSs), mobile edge caching enables direct content delivery without remote file fetching, which can reduce the end-to-end delay while relieving backhaul pressure. To effectively utilize the limited cache size in practice, cooperative caching can be leveraged to exploit caching diversity, by allowing users served by multiple base stations under the emerging user-centric network architecture. This paper explores delay-optimal cooperative edge caching in large-scale user-centric mobile networks, where the content placement and cluster size are optimized based on the stochastic information of network topology, traffic distribution, channel quality, and file popularity. Specifically, a greedy content placement algorithm is proposed based on the optimal bandwidth allocation, which can achieve (1 - 1/e)-optimality with linear computational complexity. In addition, the optimal user-centric cluster size is studied, and a condition constraining the maximal cluster size is presented in explicit form, which reflects the tradeoff between caching diversity and spectrum efficiency. Extensive simulations are conducted for analysis validation and performance evaluation. Numerical results demonstrate that the proposed greedy content placement algorithm can reduce the average file transmission delay up to 45 percent compared with the non-cooperative and hit-ratio-maximal schemes. Furthermore, the optimal clustering is also discussed considering the influences of different system parameters.

Mobile edge caching has the potential to reduce file transmission delay as well as core network load, by utilizing the cache of base stations to store content with high hit rates. However, in practice, the performance of mobile edge caching can be constrained by BS cache size. Traffic steering can enable end users to obtain requested file directly from the cache of a non-homing BS without remote file fetching, and thus enlarge the set of cached contents by exploiting the content diversity in space. On the other hand, traffic steering can also degrade spectrum efficiency, due to the higher path loss of steered users. In this paper, we investigate the performance of traffic steering on mobile edge caching, taking into account the tradeoff between content diversity and spectrum efficiency. The average file transmission delay is derived by applying stochastic geometry, under constraints of cache size and radio resources. Specifically, a greedy content placement algorithm is proposed, which can achieve near- optimal delay performance with low polynomial computational complexity. Simulation results demonstrate that the average file transmission delay can be reduced up to 55% when 10% contents can be stored in cache, by introducing traffic steering in mobile edge caching.

Mobile devices are naturally limited due to their portable sizes and will therefore never be equal to their desktop counterparts. To overcome this, Edge Cloud Computing can be utilized to execute tasks on behalf of the devices, allowing them to run applications that would normally be too demanding. In this service model, it is important to maintain a low Service Delay to keep the service transparent to the user. This can be achieved by focusing on lowering the Transmission Delay and Processing Delay. While existing approaches in the literature focus on one of those two, we postulate that only when considering both delays you can efficiently lower Service Delay and provide quality to all applications. In order to do that while being feasible, we propose a method based on Particle Swarm Optimization for lowering Service Delay in Edge Cloud Computing. Our proposal is shown to be close to optimality while still maintaining a low execution time for multiple cloudlets scenarios. Moreover, our proposal outperforms existing approaches from the literature with single focus on computation or communication, even in situations with high processing and transmission burdens, proving the superiority of a dual focus approach.

Due to physical limitations, mobile devices are restricted in memory, battery, processing, among other characteristics. This results in many applications that cannot be run in such devices. This problem is fixed by Edge Cloud Computing, where the users offload tasks they cannot run to cloudlet servers in the edge of the network. The main requirement of such a system is having a low Service Delay, which would correspond to a high Quality of Service. This paper presents a method for minimizing Service Delay in a scenario with two cloudlet servers. The method has a dual focus on computation and communication elements, controlling Processing Delay through virtual machine migration and improving Transmission Delay with Transmission Power Control. The foundation of the proposal is a mathematical model of the scenario, whose analysis is used on a comparison between the proposed approach and two other conventional methods; these methods have single focus and only make an effort to improve either Transmission Delay or Processing Delay, but not both. As expected, the proposal presents the lowest Service Delay in all study cases, corroborating our conclusion that a dual focus approach is the best way to tackle the Service Delay problem in Edge Cloud Computing.

A concept of Cloud Radio Access Networks (C-RANs) with Power over Fiber (PoF) is expected to work as a post- disaster access network architecture. In this architecture, since external power is supplied to Remote Radio Heads (RRHs) through the optical-fiber cable for data communication, RRHs can operate even when power cables are disrupted due to disasters. This network, however, needs to reduce the power consumption for RRH operation and consider Quality of Experience (QoE) to provide sustainable service with enough user satisfaction. To this end, we propose user position guidance approaches which give advices of the best position to users. Our proposed approaches are able to achieve high sustainability while satisfying the QoE constraint. Furthermore, the effectiveness of our proposed approaches is evaluated by numerical calculation.

Wireless communication devices have spread widely in our society. However, they usually depend heavily on communication infrastructure, leaving them vulnerable to disasters or congestion of base stations. In these situations, a method to send out data without the support of infrastructure is required. Data transmission by D2D communication is a reliable method that does not rely on infrastructure. In this paper, we aim to improve the data dissemination using D2D transmission by applying the concept of assigning “modes” to devices according to their own mobility. In our study, we assume a multi-channel environment, where devices will be allocated different amounts of frequency channels according to their modes. We propose a mode selection function that uses velocity information of the devices to assign modes. By using this function, it is possible to allocate more frequency channels to devices of high mobility, so that they can transmit their data to more devices as they move through a wide area. By mathematical analysis, we evaluate the fairness of disseminated data density among devices of various velocities and the obtained results indicate the effectiveness of the proposed method for improving the efficiency of data dissemination.

There are many applications which cannot be executed by mobile devices due to their limitations in memory, processing, battery, among others. One solution to this would be offloading heavy tasks to cloud servers in the edge of the network, in a service model called Edge Cloud Computing. The main Quality of Service requirement of this model is a low Service Delay, which can be achieved by lowering Transmission Delay and Processing Delay. Works in literature focus on either one of those two types of delay. This paper, however, argues that an approach which combines transmission and processing technologies to lower Service Delay would be more efficient. This idea is defended by an analysis of the service model and existing stochastic modeling of the Edge Cloud Computing system. We conclude that a dual focus approach would be the only way of truly minimizing the Service Delay, therefore being the desired method to improve Quality of Service. We conclude by laying the foundation for a future model that follows such concept.

Wireless Data Center Network (Wi-DCN) is considered one of the most promising future data center architectures due to its low installation and management cost and high flexibility of network design. However, the existing Wi-DCN is, still, not capable of providing an efficient big data mining service such as MapReduce because its topology (i.e., Cayley graph with same degree) cannot achieve enough connectivity on the breakdown of servers and spectrum efficiency, which are important factors to improve the performance of big data mining. Therefore, in order to modify the existing Wi-DCN for big data mining, this paper proposes a spherical rack architecture based on a bimodal degree distribution that improves both failure tolerance and spectrum efficiency. Extensive computer simulations demonstrate the effectiveness of our proposed rack architecture in terms of data transmission time required for MapReduce under a failure-prone environment.

A concept of cloud radio access networks (C-RANs) is becoming a popular solution to support the required communication quality for new emerging service in the future network environment, i.e., more than 10 Gbps capacity, less than 1 ms latency, and connectivity for numerous devices. In this paper, we envision a C-RAN based on passive optical network (PON) exploiting power over fiber (PoF), which achieves low installation and operation costs since it is capable of providing communication services without external power supply for large amount of remote radio heads (RRHs). This network, however, needs to reduce the optical transmission power of PoF due to the fiber fuse issue. Additionally, the diversification of services, devices, and personality indicates the need to improve user satisfaction, i.e., quality of experience (QoE), based on the user's perspective, which is different from previous approaches that aim to guarantee quality of services (QoS). Therefore, we propose a QoE-guaranteed and power-efficient network operation strategy. Our proposed operation is able to reduce the transmission power while satisfying the QoE constraint by controlling both the schedule of RRH's sleep and optical transmission power of PoF. Furthermore, the effectiveness of our proposed operation scheme is evaluated through extensive computer simulations.

The recent increase of wireless devices (i.e., STAtions (STAs)) brings forward an increase of energy consumption. We address the energy consumption issues for STAs in Wireless Mesh Networks (WMNs). One of the available techniques for power-saving is the sleep technique. However, since the common transmission mode from the STA to the server is performed based on end-to-end transmission, this results in increase of energy consumption of the STAs since they cannot enter sleep mode until end-to-end communications are completed. To cope with the issue, we focus on an Access Point equipped with External Storage (APES), which provide end-to-end communication instead of STAs. Using this, STAs can shorten the transmission time and decrease the energy consumption since they communicate with neighboring APES. In this paper, we propose a novel method to select the adequate APES as a proxy server based on the number of STAs and the amount of traffic from STAs. The proposed method effectively transmits data and reduces the energy consumption of the STAs. Moreover, we validate the efficiency of the proposed APES selection method through numerical analysis.

MapReduce architecture has been considered as one of the most promising candidates for efficient and reliable big data mining. While current MapReduce is basically designed for data center and enterprise networks, in which a number of servers are interconnected with optical fiber cables, prospective MapReduce would be applied in optical-wireless environment such as optical-wireless data center network, fiber-wireless (FiWi) access network, and so forth. To modify MapReduce for opticalwireless hybrid network, we need to answer the fundamental research problem, “How does MapReduce architecture use optical and wireless resources for task allocation?” To answer this question, this paper reveals some challenging issues and proposes a context-aware task allocation scheme that is designed by considering characteristics of both optical and wireless communications. Our proposed task allocation scheme can minimize the completion time of big data processing. Numerical results are presented to demonstrate the effectiveness of our proposed method compared with existing task allocation schemes.

We study a novel distributed storage system integrated Data Centers (DCs) and satellite networks. This integrated system is expected as distributed storage system that can keep the storage service even if disasters strike because satellite is tolerant to link disruption caused by disasters. In this paper, we focus on data distribution method in the integrated system, and assume an erasure coding and a simple replication as data distribution method. We evaluate the storage volume and transmission time on each method which are required to restore lost data when some DCs are damaged by disasters. The storage volume of the erasure coding becomes lower than that of the replication while the transmission time becomes higher. A data transfer method is proposed in this paper to shorten the transmission time of the erasure coding. The proposed method can reduce the transmission volume on downlink communication by using network coding technologies. The numerical results show that the proposed method can restore the lost data in less time.

Hierarchical unstructured peer-to-peer (P2P) networks for file sharing systems such as Gnutella and Kazaa have made a tremendous achievement in the last decade. However, while these P2P networks can be tolerant to churn, i.e., the dynamics of peer participation and departure (or fault), there still remains the issue of vulnerability to Denial of Service (DoS) attacks, i.e., when the highest degree peers are removed. In order to overcome this shortcoming, we focus on a bimodal degree distribution, which is tolerant to both churn and DoS attacks. However, the network topology affects the network stability that was not taken into considered in the previous works. Therefore, we analyze the optimal network topology for DoS attack tolerance, and accordingly develop the peer joining procedure to construct and maintain the proposed network topology. Our proposed scheme is dubbed THUP (churn/DoS Tolerant, Hierarchical, Unstructured, P2P network). Performance evaluation conducted through computer simulations shows that THUP substantially improves the stability and communication efficiency compared with other existing P2P networking structures.