The increasing demand for mobile ad hoc networks (MANETs) calls for decentralized mechanisms that can allocate transmit power across nodes and channels under stringent resource constraints. Existing optimization-based approaches, however, do not account for expected settings where each link includes multiple channels (e.g., multi-band signaling). Motivated by recent advances in machine learning for distributed optimization, we propose MANET-GNN, a graph neural network (GNN)-based algorithm for decentralized power allocation in multi-channel MANETs. MANET-GNN explicitly exploits the network topology, scales efficiently with the number of nodes and frequency bands, generalizes across topologies and channel conditions, and enables near-instantaneous inference suitable for real-time deployment. Our design builds on a constrained optimization formulation and employs a dedicated GNN architecture inspired by message passing, trained via an unsupervised procedure that is robust to noisy channel state information. Numerical evaluations demonstrate that MANET-GNN achieves high-throughput multi-channel communication across diverse MANET scenarios.

In this paper we show how to find, for a given set of n wireless nodes in the plane representing wireless sensor network, an assignment of powers to the nodes, such that the obtained communication tree will be resilient having almost minimal routing cost and almost minimal total broadcast energy cost. We will also show that the obtained tree has additional useful properties and extend our problem for the case of multiple sources. We performed extensive set of simulations confirming our findings.

We study the k-means problem for a set 𝒮⊆ℝ^d of n segments, aiming to find k centers X ⊆ℝ^d that minimize D(𝒮,X) := ∑_S ∈𝒮min_x ∈ X D(S,x), where D(S,x) := ∫_p ∈ S |p - x| dp measures the total distance from each point along a segment to a center. Variants of this problem include handling outliers, employing alternative distance functions such as M-estimators, weighting distances to achieve balanced clustering, or enforcing unique cluster assignments. For any ε > 0, an ε-coreset is a weighted subset C ⊆ℝ^d that approximates D(𝒮,X) within a factor of 1 ±ε for any set of k centers, enabling efficient streaming, distributed, or parallel computation. We propose the first coreset construction that provably handles arbitrary input segments. For constant k and ε, it produces a coreset of size O(log^2 n) computable in O(nd) time. Experiments, including a real-time video tracking application, demonstrate substantial speedups with minimal loss in clustering accuracy, confirming both the practical efficiency and theoretical guarantees of our method.

We suggest efficient and provable methods to compute an approximation for imbalanced point clustering, that is, fitting k-centers to a set of points in ℝ^d , for any d,k≥ 1 , where we aim to minimize the variance over the clusters. To this end, we utilize coresets, which, in the context of the paper, are essentially weighted sets of points in ℝ^d that approximate the fitting loss for every model in a given set, up to a multiplicative factor of 1±ε . We provide experiments that show the empirical contribution of our suggested methods for real images (novel and reference), synthetic data, and real-world data. We also propose choice clustering, which, by combining clustering algorithms, yields better performance than each one separately.

Wireless rechargeable sensor networks (WRSNs) have overcome the energy limitation bottleneck through wireless power transfer (WPT) technology. Traditional research has primarily focused on enhancing charging efficiency, while the critical issue of location privacy security arising from wireless charging has received scant attention. Additionally, sensors are vulnerable to detection and harm by malicious attackers, posing a significant threat to network integrity. In this paper, we propose two attack schemes, termed Least Squares Method (LSM) attack model and Centroid Method (CM) attack model for compromising sensor location privacy by exploiting charging power information and mobile charger behaviors. To counter such threats, we develop a scheme aimed at maximizing the node location privacy protection capabilities of the network. We propose a theoretical analysis to exploit the features of the proposed scheme. Finally, extensive test-bed experiments and simulations have been conducted to validate the effectiveness of our algorithms. The results demonstrate that our algorithms can protect at least 78% of the nodes without significantly compromising their survival rate.

Network function virtualization (NFV) enables cost reduction and optimized service deployment. By means of virtualization, network functions which used to be executed on specialized hardware are being replaced with software called Virtual Network Functions (VNFs) that can run on commodity hardware. These VNFs are applied to data flows passing through network nodes with VNFs hosted on them. To fully realize the benefits of NFV, each flow must be fully processed on VNF nodes. Given the budget constraints, only a finite number of nodes can be selected to host VNFs, and these nodes also have limited capacity to process the flows passing through them. In this paper, we consider the problem of VNF node placement and capacity allocation in a network graph G=(V,E), i.e., selecting the best subset of VNF nodes and optimally distributing their bandwidth to maximize the total volume of fully processed traffic flows F. We propose a simpler algorithm for solving this problem than the previously proposed version, representing it as an integer linear programming problem with an approximation ratio of 12(1−1/e), and time complexity O(|V|2.5·|F|2.5·L), where L is the number of bits of input data.

The initialization of neural networks is of significant importance for their performance. Currently, the prevalent initialization method is a random sample based on the network’s structure. This work presents a general yet effective method to initialize neural networks. We provide repeated experiments of training variants of Mobile-Net over down-sampled variants of Image-Net, demonstrating accuracy gain and loss decrease across most of the test sets and validation sets. E.g., for Mobile-Net (v1) and Image-Net ($32 \times 32$), we had $2.5 \%$ accuracy improvement over both the test and validation sets.

Visual Encryption Schemes (VES) encode the pixels of a secret image into n subpixel maps (shares), each of size mxm, where m is a scheme parameter. The encoding relies on a visual property such as transparency, and the resulting shares appear as random patterns of black and white subpixels. Each individual share reveals no information about the original image. To reconstruct the secret image, at least k out of n shares must be stacked, where k is a predefined threshold. This structure poses a challenge in swarm-based applications-such as Unmanned Aerial Vehicle (UAV) swarms-where each UAV holds only a single share and collaborates to identify a target image. In this work, we propose an optical VES solution that leverages wave interference as a physical encoding mechanism. The reconstructed image contains only pure black and white pixels, thus overcoming the grayscale degradation associated with traditional VES methods. Furthermore, the proposed scheme retains the computational efficiency of classical approaches while offering an improved security model: it achieves statistical information-theoretic security against honest-but-curious adversaries and is also resilient to active Byzantine adversaries.

Covert channels, enabling concealed communication within seemingly innocuous data streams, pose significant challenges to traditional information security measures. This paper presents a novel method for constructing covert channels by leveraging Error-Correcting Codes (ECCs) and minimal-weight codewords. Adversaries embed their covert messages within the benign traffic of unsuspecting victims, carefully controlling the introduced errors to avoid detection. The proposed approach not only ensures the security and reliability covert communication but also provides protection against channel errors. By integrating ECCs and minimal weight codewords, our method offers an advanced level of stealth, data integrity, and resilience. Extensive simulations validate the feasibility and performance of the approach, highlighting its potential for practical implementation in real-world scenarios. This research opens new avenues for the construction of covert channels, enhancing information hiding capabilities and reinforcing communication security in the face ever-evolving threats.2

We examine well known facility location problems under the privacy challenges posed by big data environments. For a given set of n points U∈ℝ^d , previous works have introduced the “Topology Descriptor Grid” (TDG) [13, 14], a privacy-preserving framework under which some approximate solutions are possible for a variety of clustering problems. In this paper, we introduce the Equidistant “Location Estimation using Concentric Circles” (LECC) framework in ℝ^2 , which obfuscates exact point locations while preserving their relative distances to a predetermined point. We show, under this new framework, how to obtain 2+𝒪(1/n) -approximate solutions for the 1-center, 1-median, 1-mean, and k-centrum problems, and 𝒪(k),𝒪(k),𝒪(k^2) approximations for the k-center, k-median and k-means problems, respectively. For the TDG framework we provide a (√(d),k^d-1),(d,k^d-1) , and (d^2,k^d-1) approximations for the k-center, k-median, and k-means problems, respectively.

In this paper we introduce a new, minimum-cuts based spanner algorithm, when the goal is twofold: (a) to decrease the number of active links in the network and (b) to maintain the ability of the SDN (Software-Defined Networking) controller to perform load balancing. The proposed spanner concept also can be used in order to reduce the running time of the SDN centralized routing algorithm. In addition, we show how to maintain the spanner under dynamic link insertion, deletion and changed weight. The validation of our solution is made through the analysis and simulation that show the superiority of our approach in many cases.

In this paper we consider a special case of building a minimum cost data aggregation tree problem raised in the context of wireless sensor networks, when the sensors correspond to the nodes in the two-dimensional plane, and the distances between nodes are measured according to the (squared) Euclidean norm. We show that the problem is NP-hard for the general metric case and provide a number of approximation algorithms for this problem based on the concept of Dijkstra’s cheapest paths tree, Bounded-Hop Diameter Tree and Hamiltonian Cycle construction.

In this work, we suggest computer vision methods, specifically for video tracking and map creation from video. To this end, we utilize motion vectors and clusters, which are computed very efficiently in standard video encoders, usually via dedicated hardware. We suggest a provably good tracking algorithm for clustering these vectors, by considering them as segments. For this, we utilize a definition of a \emph{coreset} which is essentially a weighted set of points that approximates the fitting loss for every model, up to a multiplicative factor of $1\pm\varepsilon$. Our method supports $M$-estimators that are robust to outliers, convex shapes, lines, and hyper-planes. We demonstrate the empirical contribution of our clustering method for video tracking and map creation from video, by running it on micro-computers (Le-Potato and Raspberry Pi) on synthetic and real-world videos with real-time running time.

Covert channels, enabling concealed communication within seemingly innocuous data streams, pose significant chal-lenges to traditional information security measures. This paper presents a novel method for constructing covert channels by leveraging Error-Correcting Codes (ECCs) and minimal-weight codewords. Adversaries embed their covert messages within the benign traffic of unsuspecting victims, carefully controlling the introduced errors to avoid detection. The proposed approach not only ensures the security and reliability of covert communication but also provides protection against channel errors. By integrating ECCs and minimal-weight codewords, our method offers an advanced level of stealth, data integrity, and resilience. Extensive simulations validate the feasibility and performance of the approach, highlighting its potential for practical implementation in real-world scenarios. This research opens new avenues for the construction of covert channels, enhancing information hiding capabilities and reinforcing communication security in the face of ever-evolving threats.

Known Visual Encryption Scheme (VES) schemes encode the secret image pixels into n subpixel maps (shares) of size m x m, where m is a parameter of the scheme. The pixel encoding is based on some pixel visual property, for example transparency. The resulting pixel maps contain black and white pixels and look like random collection of black and white pixels, such that it is impossible to reconstruct the original pixel. To reconstruct the original secret image, at least k out of n shares must be stacked together, where k is the scheme parameter. The reconstructed image appears grey, with varying shades of darker and lighter pixels. In this work, we introduce an optical VES solution that utilizes a physical model of wave interference. The image reconstructed using the proposed VES consists of pure black and white pixels, while maintaining the computational efficiency of traditional VES methods. An additional advantage of the proposed VES scheme is its enhanced security model. Besides being perfectly information-theoretic secure against honest and curious adversaries, it is also resilient against active, Byzantine adversaries. The proposed VES can be utilized in Flying Adhoc Networks, where a swarm of Unmanned Aerial Vehicles collaborates to search for a target based on a pre-assigned secret image.

Collaborative secure image matching is a problem that is applicable in various domains, for both – data in rest and data in motion. The problem is defined as follows. There is a secret image, and a set of n mobile agents. The set of mobile agents should match (compare) an observed image to the original secret image. In this paper we discuss some of the existing approaches, and present an alternative solution applied and analyzed for different applications. The first application is a swarm of Unmanned Aerial Vehicles (UAV) that search for a target specified by an image. The second application is a social network that serves as a smart storage device capable of performing distributed, secret image matching operations. Our solution is based on the well-known Visual Encryption Scheme (VES) and projections of visual bit maps rather than (quadratic complexity) messages exchange in implementing Secure Multi Party Computation (MPC) scheme. We present a perfect-information-theoretic secure solution for this problem. To keep the original image secrecy, at least k out of n mobile agents are required to retrieve any information about the original image.

In this work, we study the connected version of the covering problem motivated by the coverage of ad-hoc drones’ swarm. We focus on the situation where the number of drones is given, and this number is not necessarily enough to cover all users. That is, we deal with a budget optimization problem, where the budget is the number of given drones. We assume that each ground user has different QoS requirements. Additionally, each ground user has a weight that corresponds to the importance (rank) of the user. Moreover, we consider the case when there is no third-party entity that provides connectivity to the drones. In this paper, we propose a 3D deployment scheme with the given number of drones such that the sum of the weights (ranks) of the ground users covered by drones is maximized (when the covering radii satisfy QoS of these users), and the drones form a connected graph. We present a number of approximate solutions with provable guaranteed performance evaluation that have been validated also through the simulation platform.

In this paper, we consider the problem of prediction of Radio Link Failures (RLF) in flying ad hoc networks (FANETs).Many environmental factors that influence the quality of radio wave propagation are dynamic, and thus, drones must continually learn and update their radio link quality prediction model while they operate online.Online machine learning algorithms can be used to build adaptive RLF predictors without requiring a pre-deployment effort.To predict the RLF, we use an online machine learning algorithm and information gathering by message-passing from the neighbors.We propose an algorithm called ML-Net (Machine Learning and Network algorithm) to predict RLF.To the best of our knowledge, the combination of online machine learning algorithms together with the message-passing algorithm has not been used before.The proposed methodology outperforms the state-of-the-art online machine learning algorithms.

Vehicular ad hoc network (VANET) is an emerging technology with a promising future and great challenges. It aims to promote safe driving, improve traffic flow and also enables a variety of entertainment applications. A fundamental need in such a network is the ability to assess vehicular speed. This enables the collection of statistics for the purpose of traffic engineering and long-term planning, and is also critical information for law enforcement groups. Many existing speed assessment technologies suffer from high physical visibility, and relatively expensive hardware. Even those that avoid detection, are inflexible due to being location specific. Therefore, reducing the ability to track and enforce traffic speed and limiting the collection of statistics for traffic engineering. In this paper, we propose a method for vehicle speed assessment, by extracting an induced Communication Connectivity Graph (CCG) from VANET optimized link state routing (OLSR) protocol, and composing an optimization problem for assessing the speed boundaries, using graph hop distance based constraints. We performed evaluation experiments in different traffic scenarios using traffic simulation tool. Our method can provide a cost-effective, easy to implement and hard to uncover solution for vehicles speed assessment on highways.

We examine the problem of discovering the set P of points in a given topology that constitutes a k-median set for that topology, while maintaining location privacy. That is, there exists a set U of points in a d-dimensional topology for which a k-median set must be found by some algorithm A, without disclosing the location of points in U to the executor of A. We define a privacy preserving data model for a coordinate system we call a "Topology Descriptor Grid", and show how it can be used to find the rectilinear 1-median of the system and a constant factor approximation for the Euclidean 1-median. We achieve a constant factor approximation for the rectilinear 2-median of a grid topology. Additionally we show upper and lower bounds for the k-center problem.