Nowadays, coronavirus (COVID-19) is getting international attention due it considered as a life-threatened epidemic disease that hard to control the spread of infection around the world. Machine learning (ML) is one of intelligent technique that able to automatically predict the event with reasonable accuracy based on the experience and learning process. In the meantime, a rapid number of ML models have been proposed for predicate the cases of COVID-19. Thus, there is need for an evaluation and benchmarking of COVID-19 ML models which considered the main challenge of this study. Furthermore, there is no single study have addressed the problem of evaluation and benchmarking of COVID diagnosis models. However, this study proposed an intelligent methodology is to help the health organisations in the selection COVID-19 diagnosis system. The benchmarking and evaluation of diagnostic models for COVID-19 is not a trivial process. There are multiple criteria requires to evaluate and some of the criteria are conflicting with each other. Our study is formulated as a decision matrix (DM) that embedded mix of ten evaluation criteria and twelve diagnostic models for COVID-19. The multi-criteria decision-making (MCDM) method is employed to evaluate and benchmarking the different diagnostic models for COVID19 with respect to the evaluation criteria. An integrated MCDM method are proposed where TOPSIS applied for the benchmarking and ranking purpose while Entropy used to calculate the weights of criteria. The study results revealed that the benchmarking and selection problems associated with COVID19 diagnosis models can be effectively solved using the integration of Entropy and TOPSIS. The SVM (linear) classifier is selected as the best diagnosis model for COVID19 with the closeness coefficient value of 0.9899 for our case study data. Furthermore, the proposed methodology has solved the significant variance for each criterion in terms of ideal best and worst best value, beside issue when specific diagnosis models have same ideal best value.

Meeting the stringent requirements of real-time Internet of Things (IoT) is a bit challenging than expected. Indeed, IoT devices generate massive amounts of data through their sensing features and face some constraints in timely transmitting sensed data to collectors. To overcome this problem, Unmanned Aerial Vehicles (UAVs) are deployed to act as data collectors for IoT devices as they significantly improve the freshness of collected data in real-time applications. Nevertheless, IoT devices and UAVs have limited energy capacity. In this paper, we tackle the energy concern of IoT devices by utilizing UAVs as data collectors and energy transmitters and by promptly charging IoT devices whenever necessary. As for the energy concern of UAVs, we deploy a set of Unmanned Ground Vehicles (UGVs) to energy supply UAVs, allowing them to complete their tasks successfully. Our objective is to employ a multi-agent reinforcement learning method for optimally controlling the trajectories of both UGVs and UAVs so that it jointly decreases their energy consumption, reduces the Age of Information (AoI) of IoT devices, and timely charges UAVs and avoids their failures. We conducted a series of tests using a simulation tool to validate the effectiveness of the approach.

A drone is an unmanned aerial vehicle, which is deployed in a particular Fly Zone (FZ), and used to collect crucial information from its surrounding environment to be transmitted to the server for further processing. Generally, a Mobile User (MU) is required to access the real-time information collected by the drone stationed in a specific FZ securely. Therefore, to ensure secure and reliable communications an Authenticated Key Exchange (AKE) protocol is imperative to the Internet of Drone (IoD) environment. An AKE scheme ensures only authentic MU to access IoD network resources. Upon successful authentication, MU and drone can set up a secret session key for secure communication in the future. This paper presents a novel Lightweight AKE Protocol for IoD Environment (LAKE-IoD), which first ensures the authenticity of MU and also renders session key establishment mechanism between MU and drone with the help of a server. LAKE-IoD is an AKE protocol, which is based on an authenticated encryption scheme AEGIS, hash function, and bit-wise XOR operation. Meticulous formal security verification by employing a software tool known as Scyther and informal security analysis demonstrates that LAKE-IoD is protected against different well-known active and passive security attacks. Additionally, Burrows-Abadi-Needham logic is applied to verify the logical completeness of LAKE-IoD. Furthermore, a comparison of LAKE-IoD with the related schemes shows that LAKE-IoD incurs less communication, computational and storage overhead.

Due to the dynamic nature of connectivity in terrestrial vehicular networks, it is of great benefit to deploy unmanned aerial vehicles (UAVs) in these networks to act as relays. As a result, a remarkable number of studies have exploited UAVs to bridge the communication gaps between terrestrial vehicles, and sometimes despite their unoptimized mobility, their restricted communication coverage, and their limited energy resources. However, it was noted that for an intermittently connected vehicular network, UAVs could not cover all sparse areas all the time. Even worse, when deploying enough UAVs to cover all these areas, the probability of inter-UAV collisions increases, and it will be complex to control their movements efficiently. Consequently, it is required to dispatch an organized and intelligent group of UAVs to perform communication relays in the long term while keeping their connectivity, minimizing their average energy consumption, and providing an efficient coverage strategy. To meet these requirements, we propose a deep reinforcement learning (DRL) framework, called DISCOUNT (Dispatch of UAVs for Urban VANETs). Extensive simulations have been conducted to evaluate the performance of the proposed framework. It has been shown that the proposed framework significantly outperforms two commonly-used baseline techniques and some reinforcement learning methods in terms of energy consumption, coverage, and routing performances.

The edge computing infrastructure has enabled a massive amount of data in the smart grid environment by a large number of connected automated devices to be processed at the edge of the network in proximity to the data generation source. The demand response management is a fundamental requirement for an efficient and reliable smart grid environment, which can be accomplished by the transfer of data between smart devices and the utility center (UC) in a smart city, very frequently. However, this frequent data transfer is subject to multiple threats including the tempering. Several authentication schemes were proposed to secure smart grid environment. However, many such schemes are either insecure or lack the required efficiency. To counter the threats and to provide efficiency, a new authentication scheme for demand response management (DRMAS) is proposed in this paper. DRMAS provides all necessary security requirements and resists known attacks. The proposed DRMAS is provably secure under formal analysis supplemented by a brief discussion on attack resilience. Moreover, the DRMAS completes the authentication procedure in just 20.11 ms by exchanging only 2 messages.

Symmetric encryption has been considered as one of the essential means of ensuring security of end to end communication. The robustness and strength of modern day block encryption systems are based on the cryptographic features of substitution-boxes which are used to inject confusion ability during substitution-phases. In this paper, as an alternative to random, chaos or algebraic based construction methods, we propose to present an efficient method for the generation of cryptographic highly nonlinear substitution-boxes. The nature-inspired particle swarm optimization is reconnoitered to develop proposed method wherein the initial population is generated through a simple but with rich dynamics chaotic Renyi map. The anticipated method is analyzed for different scenarios such as change in population size, number of iterations, and linear increase in inertial weight. The performance assessment of generated S-boxes under standard criterions corroborate that the proposed method has excellent cryptographic features and found grander than many recent optimization based S-box methods. Moreover, an image encryption application of proposed S-boxes is also suggested to determine their suitability and applicability for image based security applications.