In this paper, we present a novel approach to enhance data collection in Vehicular Ad-Hoc NETworks (VANETs). VANETs are a growing area of interest due to their unique characteristics and challenges, such as rapidly changing topology and frequent network disruptions. Efficient data collection is a critical issue in vehicular networks and has therefore become a focus of research. To address this challenge, we propose a stable clustering optimization solution based on adaptive multiple metrics. The cluster head selection is done based on both mobility metrics, such as position and relative speed, and Quality of Service (QoS) metrics, such as neighborhood degree and link quality. The proposed solution has been tested and evaluated through simulations using a vehicular mobility simulator in a realistic urban environment. The results show that the proposed approach provides more stable clusters with higher QoS, and allows for the selection of the appropriate cluster head to collect data from the vehicles and forward it to the destination.

Cellular deployment of new generations faces a coverage challenge due to the non-line-of-sight (NLOS) between clients' devices and base station (BS). Therefore, relaying on using the emerging technology; intelligent reflective surface (IRS) to reconfigure wireless signal propagation is considered the best solution that can address the mentioned challenge. Additionally, choosing the position of the IRS is not an easy task as the clients are mobile. Hence, there is a need for an efficient model to elect the best positions of the IRSs for a better network performance. In this work, two fold model is proposed to provide an automated solution to optimize IRS positions. The first one is the mixed integer linear programming (MILP) that solves the IRS positions problem in a classical way. Whereas the second one is based on the reinforcement learning optimization (RLO) with complex encoder and decoder network architecture to provide fast learning of the MILP results with a low mean square error. The proposed RLO model's validity is studied using 10 days of mobile dataset and actual cellular BSs' positions in the city of Rome (Italy). This study is based on the use of long short term memory (LSTM) and gated recurrent unit (GRU). The results show a significant performance of the proposed model based on LSTM compared to GRU.

Several reinforcement techniques are compared to take control of Unmanned Aerial Vehicles (UAVs) and optimize communication offloading in cellular networks. Navigation actions are calculated to send the drones to the required position and turn them back when not needed. First, a use case is expressed and solved in form of a linear programming problem (ILP). Then, a Q learning algorithm is designed and evaluated to solve the same problem. Finally, a deep neural network based on Long Short Term Memory recurrent networks is used. The results of the three approaches are obtained with a real dataset extracted from the CDRs (Call Detail Records) in Milan city, Italy. It is shown that Q learning needs long convergence times to succeed to approach the ILP optimal results. Also, we demonstrate that deep neural network techniques learn much faster and mimic the ILP with very high scores.

Migrating Virtual Machines (VM) is a useful tool for managing computing resources, load balancing, reducing energy consumption and facilitating fault management in cloud computing. In this paper, we propose a green energy-aware VM migration approach aiming to increase the use of green energy of the data center taking into account the characteristics of the VMs and the available resources of the data stores. In essence, we define an energy aware optimization problem in hybrid energy-powered data centers. The objective is to extend the existing Dynamic Resource Scheduling (DRS) of the virtualization solution in order to integrate other environmental variables. The obtained results based on an experimental study carried out in the pedagogic data center of the IT department of our university have shown that the energy efficiency can be managed by reassigning VMs to the data-center hosts.

An AI based energy management system (EMS) for microgrids is proposed. It is composed of three modules: a strategy based module, a deep learning (DL) and a reinforcement learning module (RL). This framework determines heuristically the optimal actions for the microgrid system under different time-dependent environmental conditions. In essence, a main innovation is applied to the EMS. Our deep learning algorithm uses recurrent neural networks (RNNs) instead of the habitual State Action Reward (SAR) approach (whether classical or deep). Learning is hence guided by successful actions rather than by blind exploration. A large improvement in learning rates is hence observed when compared to classical Q-learning on real datasets that present a large diversity in energy consumption profiles, acquired in French premises over a long period. It leads to question about the best appropriate reinforcement policies to adopt when solving large state environments.

The success of the Internet of Things (IoT) depends on the ability to provide reliable communication to the billions of devices that are used in many applications. In essence, estimating the quality of wireless links ensures the optimization of several protocols, reduces the end-to-end latency, and increases the reliability and the network lifetime. In this paper, we study the link quality in the Time Slotted Channel Hopping (TSCH) network by analyzing the received signal strength (RSSI) and error rates. The objective is to understand the temporal properties of these parameters which is important to select the appropriate channels for the critical applications and to enhance the Quality of Service (QoS) of the network. We apply machine learning techniques to a real dataset collected from a testbed IoT network deployed at Grenoble, France. We define five classes and present a classification of the 16 channels by comparing the performances of KNN (k-Nearest Neighbor) and LSTM (Long Short-Term Memory) algorithms.

Transfer learning (TL) is a useful technique that enables the wide spreading of neural networks after re-adaptation of their weights. In ths paper, two methods are introduced for transfer learning of recurrent neural networks: D-LSTM (Long Short Term Memory with deep layers) and CNN-1D (Convolutional Neural Network of One Dimension). The first is used to improve the prediction of time series when datasets are too small to obtain satisfactory results. The second enables personalizing and hence re-adaptation of an already-trained network to a new class of time series. In fact, the CNN-1D classification is applied to those real datasets to classify different behaviors in a large city. We show that our architecture drastically improves prediction when transfer learning is used in the same class of behavior but also on different classes of behaviors.

Unusual Spatio-Temporal fluctuations in cellular network traffic may lead to drastic network management misbehaviors and at least abnormal drops in quality of experience. It is also expected that the management of future cellular networks will mostly rely on machine learning and automation. In this article, we present a dynamic on-line data mining technique to detect these network anomalies allowing, network operators to pro-actively monitor and control a variety of real-world phenomena with less damage to the overall experience. To overcome the network performance degradation that can occur in real time, the network manager must imperatively and instantly identify abnormalities and hence provide a better continuous quality of service for the subscribers. Based on real cellular communication traces, we propose an automated framework, called STAD, ensuring spatio-temporal detection outliers using a combination of machine learning techniques including One-class SVM (OCSVM), Support Vector Regression (SVR) and recurrent neural networks, Long Short-Term Memory (LSTM). STAD is double checked with two real datasets of CDRs where results show high accuracy compared to the Isolation Forest and Auto-Regressive Integrated Moving Average (ARIMA) models.

This paper presents a framework based on the data analysis concept to automate the management of resources in cellular networks. Three processes are defined: identifying and detecting anomalies, analyzing the causes, and triggering adequate recovery actions. First, the proposed solution executes Deep Learning algorithms to forecast the normal behavior of the network and defines dynamic thresholds. Then, it identifies cells with peak demands and raises alarms if the measured real-time data exceeds the threshold values. Second, we define QoS optimization methods to proceed with suitable design for resource allocation as well as fault detection and avoidance. Hence, we distinguish three cases and define two classes of data: Real-time and non-real-time traffic. This solution is applied to a pre-analyzed semi-synthetic real dataset extracted from the CDRs (Call Detail Records) in Milan city, Italy. This dataset contains the Internet activity records of two months in three areas. The preliminary results elucidate the feasibility and preeminence of our proposed anomaly detection framework.

The Chadian electricity company (SNE) wants to modernize the electricity grid's infrastructure towards a smart grid by using intelligent sensors and considering the architectural aspects of the network and communication. This article presents an Advanced Metering Infrastructure (AMI) deployment in N'Djamena to replace traditional analog devices. The objective is to detect electricity theft by analyzing the network in near-real-time collected by the smart devices. Energy consumption equations at different points are also presented.

The Chadian electricity company (SNE) wants to deploy a good and a robust infrastructure in order to satisfy the increased demand needs, to have an effective management of the electricity grid and to detect the electricity theft. In fact, Fraudulent electricity consumption has many dangerous effects such as decreasing the supply quality, increasing generation load, affecting the overall economy, etc... The adaptation of smart grid and the use of the internet of things can significantly reduce this loss by analyzing the electricity consumption patterns of customers and identifying illegal consumers based on irregularities in consumption. In this paper, we present a sensor deployment infrastructure in N'Djamena city based on a self-stabilizing hierarchical algorithm. Then, we study the fault tolerance of the proposed algorithm and finally we detail the electricity theft detection approach. The objective of the SNE is not to completely replace the traditional electricity grid in one step, but to gradually evolve it into a smart grid by incorporating smart meters and sensors, taking into account the architectural aspects of the network and communication. The proposed method can effectively identify fraudulent users of the system. Extensive simulations using the OMNeT++ simulator reveal that the algorithm is convergent regardless of the parameter choice of the function choose.

Time Slotted Channel Hopping (TSCH) is one of the most used MAC mechanisms introduced by the new amendment IEEE 802.15.4e. It combines both slotted access with channel hopping technique to allow multiple communications while exploiting the 16 available channels of 2.4GHz band. The channel hopping mechanism of 802.15.4e considers an interference-free environment and does not specify how to build and manage a schedule for communication purpose. In this paper, we propose a new distributed channel hopping scheme that exploits Latin rectangles to avoid interference and collisions. In essence, the scheduling of links is performed by Latin rectangles where rows are channel offsets and columns are slot offsets. Thus, the frequency of communication is derived using Latin rectangles. Consequently, interference and multi-path fading are mitigated with more reliability and robustness. The efficiency of the proposed scheme has been validated by extensive simulation.

Vehicular Ad hoc NETworks (VANETs) are characterized by the rapidly changing topology and then a frequent network disruption. Hence, connectivity of moving vehicles presents an important challenge that critically influences the data transmission. Furthermore, data delivery ratio depends on routing protocols, applications type as well as environment characteristics. As a matter of fact, real experimentation in vehicular networks are costly and hard to deploy especially on large scale. Consequently, a vehicular mobility simulator is a good compromise to study how efficient are the data transmission mechanisms. In this paper, we comprehensively study the impact of the radio connectivity on data communication in vehicular networks. The analysis were realized based on a vehicular mobility simulator which runs a realistic scenario of mobility traffic in a real urban environment. A simple scenario of a safety application was implemented to examine the behavior of three well-known topology-based routing protocols. For the purpose of the analysis, we varied the simulation setup such as the density and the data traffic rate to determine the impact of the connectivity. The simulation results show that a realistic modelling of radio propagation has an important role in data transmission.

Stimulated by the emerging Internet of Things (IoT) applications and their massive generated data, the cellular providers are introducing various IoT functionalities into their networks architecture. They should integrate intelligent and autonomous mechanisms that are able to detect sudden and anomalous behavior issues. In this paper, we present an adaptive anomaly detection approach in cellular networks consisting of two parts: the detection of overloaded base-stations using machine learning algorithm (LSTM - Long Short-Term Memory) and the deployment of drones as mobile base-stations that support and back up the overloaded cells. The proposed approach is validated using real dataset extracted from the CDR of Milan combined with semi-synthetic eHealth data. Initially, The LSTM algorithm analyzes the impact of eHealth applications on cellular networks and identifies cells with peak demands. Then, drones are deployed to collect the requested data from these cells. The obtained results show that the use of drones improves the quality of service and provides a better network performance.

Electrical vehicular (EV) energy management is a promising trend. Forecasting vehicular trajectories and delay is crucial for EV energy management. The presented work is devoted to the study and the application of deep learning techniques on specific road trajectories. First, exhaustive deep learning algorithms are considered. Second, road traces are converted to time series. Then, delays and road trajectories are analyzed. In fact, we consider two Recurrent Neural Networks (RNN): LSTM (Long Short Term Memory) and GRU (Gated Recurrent Units). Neural Networks are adapted and trained on 60 days of real urban traffic of Rome in Italy. We calculate the Loss function for both machine learning techniques which is defined by mean square error (MSE) and Root mean square error (RMSE). Experimental results demonstrate that both LSTM and GRU are adequate for the context of EV in terms of route trajectory and delay prediction.

In this paper, we present a hierarchical routing algorithm based on a bio-inspired heuristic named Cuckoo Search (CS) approach. In our proposed scheme, the network area is divided into a logical virtual grid. Nodes are deployed randomly and organized as static clusters where each cell represents one cluster. After the cluster heads are elected, data is collected, aggregated and forwarded to the base station using the cuckoo search approach. Extensive simulations showed the effectiveness of our data dissemination mechanism. They confirm that our proposed scheme outperforms other conventional existing techniques such as LEACH and M-GEAR in terms of energy saving.

Wireless Sensor Networks (WSN) consist of nodes with limited power deployed in the area of interest. Nodes cooperate to collect, transmit and forward data to a base station. In WSN, clustering and scheduling techniques ensure collecting data in an energy efficient manner. In this paper, we present a Power Aware Scheduling and Clustering algorithm based on Ant Colony Optimization (PASC-ACO). In the proposed approach, energy is saved by scheduling some nodes in the active state to generate data and keep network connectivity, while putting others in the sleep state. Then, ACO algorithm is used for routing data packets in the network in order to minimize the energy wasted in transferring the redundant data sent by sensors in a densely deployed network. ACO scheme can play a significant role in the enhancement of network lifetime by selecting the optimum path to reach the base station. The PASC-ACO algorithm was studied by simulation for various network scenarios. The results confirm that our proposed scheme outperforms other existing techniques such as LEACH, M-GEAR and our pervious work PASC. PASC-ACO achieves better performances in terms of lifetime by balancing the energy load among all the nodes.

This paper describes an energy efficient scheduling algorithm with adaptive fault tolerance for clustered mobile wireless sensor networks. This algorithm equalizes the cluster lifetime by making the total energy stored in the clusters proportional to their power consumption. It is proposed to relocate, in an optimized way, redundant sensors using cascaded movement to achieve optimal connectivity and coverage of the network. Simulation results confirm that our proposed scheme outperforms other existing techniques such as LEACH and M-GEAR and show that it achieves better performances in terms of lifetime by balancing the energy load among all the nodes.