Search

Currently, with the widespread of the intelligent Internet of Things (IoT) in beyond 5G, wireless federated learning (WFL) has attracted a lot of attention to enable knowledge construction and sharing among a huge amount of distributed edge devices. However, under unstable wireless channel conditions, existing WFL schemes exist the following challenges: First, learning model parameters will be disturbed by bit errors because of interference and noise during wireless transmission, which will affect the training accuracy and the loss of the learning model. Second, traditional edge devices with CPU acceleration are inefficient due to the low throughout computation, especially in accelerating the encoding and decoding process during wireless transmission. Third, current hardware-level GPU acceleration methods cannot optimize complex operations, for instance, complex wireless coding in the WFL environment. To address the above challenges, we propose a software-defined GPU-CPU empowered efficient WFL architecture with embedding LDPC communication coding. Specifically, we embed wireless channel coding into the server weight aggregation and the client local training process respectively to resist interruptions in the learning process and design a GPU-CPU acceleration scheme for this architecture. The experimental results show its anti-interference ability and GPU-CPU acceleration ability during wireless transmission, which is 10 times the error control capability and 100 times faster than existing WFL schemes.

Information security policy (ISP) plays a crucial role in maintaining the availability, confidentiality, and integrity of sensitive data. However, it is of high complexity and heterogeneity due to the variety and redundancy of security policy practices and complexity of organisational systems. Various and duplicate ISP models and frameworks have been offered in the literature. The duplicate security policy practices, procedures, and processes in the existing models have made ISP disorganised, unstructured, and unclear to organisational users. As a result, there is still a need for a standardised and integrated model to make it simpler to share, manage, and reuse ISP practices amongst the organisations. The main objective of this study is to construct a metamodel to unify, organise, and structure ISP practices. By identifying, recognising, extracting, and combining the common information security policy practices from various ISP models in a built ISP metamodel called ISPM, we seek to make it simple for users and field specialists to derive/instantiate security policy models for their organisations. The development and validation process of the ISPM is based on the common security frameworks such as ISO 27001 frameworks. The developed ISPM consists of 19 common security practices: organisation, risk management, access control policy, edit, review, compliance, business management, backup and recovery, incident response, SETA program, security awareness, security training, security education, email security policy, cloud security policy, network security policy, website security policy, physical security policy, and privacy security policy. Each common security practice consists of several operations and attributes. The performance of the developed ISPM was compared to that of other models to evaluate its completeness and logicalness. Using ISO 27001 as a framework, the findings confirmed the comprehensiveness of ISPM. Therefore, it can contribute to organisations’ security by helping them to develop their own security policy models.

In the last years, due to the limited resources of consumer products, energy-saving is known as one of the design challenges of Wireless Sensor Networks (WSNs). Clustering is a practical technique to enhance the performance of the network including energy efficiency, network scalability, and network easier management. In cluster-based networks, the size of clusters has a key role in the network power consumption. Non-optimized clustering results in increasing the power consumption of the whole network. The small size clusters leads to appear coverage hole in the network, as well as this property is the opposite of being the scalability of the network. In addition, in non-optimized clusters, reducing the energy consumption of the nodes as the key objective of clustering, cannot be pursued, thus the clustering will result contrary. Consequently, the energy consumption reduction after clustering can be guaranteed by considering the power consumption of nodes before clustering in cluster size optimization. Hence, in this paper, an Energy-efficient and Coverage-guaranteed Unequal-sized Clustering (ECUC) scheme is proposed which considers both energy and coverage issues simultaneously in optimizing the cluster size. Based on the simulation results, the proposed scheme remarkably enhances the network lifetime by reducing the total dissipated energy while guarantying the coverage issue.

Unmanned aerial vehicles (UAVs) are adaptable and rapid mobile boards that can be applied to several purposes, especially in smart cities. These involve traffic observation, environmental monitoring, and public safety. The need to realize effective drone forensic processes has mainly been reinforced by drone-based evidence. Drone-based evidence collection and preservation entails accumulating and collecting digital evidence from the drone of the victim for subsequent analysis and presentation. Digital evidence must, however, be collected and analyzed in a forensically sound manner using the appropriate collection and analysis methodologies and tools to preserve the integrity of the evidence. For this purpose, various collection and analysis models have been proposed for drone forensics based on the existing literature; several models are inclined towards specific scenarios and drone systems. As a result, the literature lacks a suitable and standardized drone-based collection and analysis model devoid of commonalities, which can solve future problems that may arise in the drone forensics field. Therefore, this paper has three contributions: (a) studies the machine learning existing in the literature in the context of handling drone data to discover criminal actions, (b) highlights the existing forensic models proposed for drone forensics, and (c) proposes a novel comprehensive collection and analysis forensic model (CCAFM) applicable to the drone forensics field using the design science research approach. The proposed CCAFM consists of three main processes: (1) acquisition and preservation, (2) reconstruction and analysis, and (3) post-investigation process. CCAFM contextually leverages the initially proposed models herein incorporated in this study. CCAFM allows digital forensic investigators to collect, protect, rebuild, and examine volatile and nonvolatile items from the suspected drone based on scientific forensic techniques. Therefore, it enables sharing of knowledge on drone forensic investigation among practitioners working in the forensics domain.

After the September 11 attacks, security and surveillance measures have changed across the globe. Now, surveillance cameras are installed almost everywhere to monitor video footage. Though quite handy, these cameras produce videos in a massive size and volume. The major challenge faced by security agencies is the effort of analyzing the surveillance video data collected and generated daily. Problems related to these videos are twofold: (1) understanding the contents of video streams, and (2) conversion of the video contents to condensed formats, such as textual interpretations and summaries, to save storage space. In this paper, we have proposed a video description framework on a surveillance dataset. This framework is based on the multitask learning of high-level features (HLFs) using a convolutional neural network (CNN) and natural language generation (NLG) through bidirectional recurrent networks. For each specific task, a parallel pipeline is derived from the base visual geometry group (VGG)-16 model. Tasks include scene recognition, action recognition, object recognition and human face specific feature recognition. Experimental results on the TRECViD, UET Video Surveillance (UETVS) and AGRIINTRUSION datasets depict that the model outperforms state-of-the-art methods by a METEOR (Metric for Evaluation of Translation with Explicit ORdering) score of 33.9%, 34.3%, and 31.2%, respectively. Our results show that our framework has distinct advantages over traditional rule-based models for the recognition and generation of natural language descriptions.

Over the past years, carbon-based materials and especially graphene, have always been known as one of the most famous and popular materials for sensing applications. Graphene poses outstanding electrical and physical properties that make it favorable to be used as a transducer in the gas sensors structure. Graphene experiences remarkable changes in its physical and electrical properties when exposed to various gas molecules. Therefore, in this study, a set of new analytical models are developed to investigate energy band structure, the density of states (DOS), the velocity of charged carriers and I-V characteristics of the graphene after molecular (CO, NO2, H2O) adsorption. The results show that gas adsorption modulates the energy band structure of the graphene that leads to the variation of the energy bandgap, thus the DOS changes. Consequently, graphene converts to semiconducting material, which affects the graphene conductivity and together with the DOS variation, modulate velocity and I-V characteristics of the graphene. These parameters are important factors that can be implemented as sensing parameters and can be used to analyze and develop new sensors based on graphene material.