Abstract Link prediction is a crucial task in complex network analysis, aiming to predict future connections between nodes in a graph. This problem is particularly challenging in sparse networks, where the low number of edges complicates traditional prediction methods. To address this, we propose singular value decomposition-graph attention network-gradient boosting (SVD-GAT-GB), a hybrid approach that densifies the sparse graph using truncated SVD, improves node representations through a GAT, and applies GB for accurate link prediction. Our method effectively handles the sparsity issue, significantly enhancing the prediction accuracy by leveraging structural information extracted from the densified graph. We validate our approach through extensive experiments on multiple datasets, demonstrating improvements in both F1-score and AUC by approximately 32% and 11%, respectively, over existing methods. The proposed model proves to be accurate and robust across diverse types of networks, making it an effective solution for link prediction in sparse graphs.

The Internet of Things (IoT), as a pervasive paradigm, is becoming an integral part of the tech industry and academic research in recent years. It forms a ubiquitous heterogeneous network connecting humans and things. The basic premise is acquiring data from the environment with sensors and remote intelligent management via actuators. For IoT service providers, time and place are functional parameters. Whereas most IoT scenarios are in indoor spaces and GPS cannot fully cover them, applying an indoor positioning system (IPS) is necessary. Besides, indoor enabling technologies can leverage the capability of IoT in context-aware services. In this article, we aim to provide a panoramic view of IPSs and localization services with the centrality of IoT. First, we explain the main concepts and review the latest positioning methods, techniques, and technologies with IoT remarks. Then, we discuss technical implementation challenges and open issues with feasible solutions. Finally, we mentioned location-based services (LBSs), real IoT applications, and active vendors in the realm of positioning services. This article provides a real insight into LBSs in IoT for future research.

Third-generation neural networks, or Spiking Neural Networks (SNNs), aim at harnessing the energy efficiency of spike-domain processing by building on computing elements that operate on, and exchange, spikes. In this paper, the problem of training a two-layer SNN is studied for the purpose of classification, under a Generalized Linear Model (GLM) probabilistic neural model that was previously considered within the computational neuroscience literature. Conventional classification rules for SNNs operate offline based on the number of output spikes at each output neuron. In contrast, a novel training method is proposed here for a first-to-spike decoding rule, whereby the SNN can perform an early classification decision once spike firing is detected at an output neuron. Numerical results bring insights into the optimal parameter selection for the GLM neuron and on the accuracy-complexity trade-off performance of conventional and first-to-spike decoding.

Energy detection has been for long constituting the most popular sensing method in RADAR and cognitive radio systems. The present paper investigates the sensing behaviour of an energy detector over Hoyt fading channels, which have been extensively shown to provide rather accurate characterization of enriched multipath fading conditions. To this end, a simple series representation and an exact closed-form expression are firstly derived for the corresponding average probability of detection for the conventional single-channel communication scenario. These expressions are subsequently employed in deriving novel analytic results for the case of both collaborative detection and square-law selection diversity reception. The derived expressions have a relatively tractable algebraic representation which renders them convenient to handle both analytically and numerically. As a result, they can be utilized in quantifying the effect of fading in energy detection based spectrum sensing and in the determination of the trade-offs between sensing performance and energy efficiency in cognitive radio communications. Based on this, it is shown that the performance of the energy detector depends highly on the severity of fading as even slight variations of the fading conditions affect the value of the average probability of detection. It is also clearly shown that the detection performance improves substantially as the number of branches or collaborating users increase. This improvement is substantial in both moderate and severe fading conditions and can practically provide full compensation for the latter cases.

Cognitive radio is regulated to permit efficient access to leftover licensed bands by unlicensed users in an opportunistic manner. Spectrum sensing is aimed to protect the primary users from interferences of the secondary users. Energy detection constitutes the most dominant method for spectrum sensing in cognitive radio and many other systems owing to its non-coherent structure as well as its simplicity and applicability. The sensing capability of wireless detectors in unreliable environments is challenged by adverse effects of multipath fading. Motivated by this, in the presented work the performance analysis of the energy detector is investigated over η-μ and α-μ fading channels. Novel closed-form expressions are derived for the detector using the area under the receiver operating characteristic curve (AUC) measure. The derived analytical results are verified by numerical computations and Monte Carlo simulations.

Energy detection constitutes a popular sensing approach thanks to its relatively satisfactory performance at low complexity requirements. Its efficiency can be practically enhanced by employing diversity schemes which are also capable of providing adequate mitigation of multipath fading effects. Based on this, the present work is devoted to the analysis of energy detection based spectrum sensing over generalized multipath fading channels using square law combining. Unlike the traditional evaluation based on the receiver operating characteristic (ROC) curves, the present analysis is based on the area under ROC curve (AUC), which is a particularly accurate performance measure that is used widely in natural sciences and engineering. To this end, novel closed-form expressions are firstly derived for the conventional AUC over the generalized κ − μ fading channels. These results are subsequently employed for deriving closed-form expressions for the case of square law combining. It is shown that the corresponding performance is, as expected, highly dependent upon the severity of fading and is improved substantially as the number of branches increase. In this context, it is also shown that using up to five branches ensures rather acceptable performance even at non-high signal-to-noise ratio values. Furthermore, the offered results have a relatively convenient algebraic representation and can be useful in analyses relating to cognitive radio and RADAR systems.

A fast and reliable detection scheme is essential in several wireless applications such as radar and cognitive radio systems. Energy detection is such a method as it does not require a priori information of the received signal while it exhibits low implementation complexity and costs. Since the detection capability of ED is largely affected by the effects of multipath fading, this paper is devoted to a thorough analysis of energy detection based spectrum sensing over generalized fading conditions. To this end, analytical expressions are firstly derived using the area under the receiver operating characteristic curve (AUC) under additive white Gaussian noise. This analysis is subsequently extended to the case of generalized fading conditions characterized by k — μ and η — μ fading distributions. The offered results are novel and are employed in analyzing the corresponding performance. It is shown that fading phenomena result to detrimental effects on the performance of spectrum sensing since the deviation between severe and non-severe conditions is rather substantial.