The advancement of wireless technology is affected by Spectrum scarcity and the overcrowding of free spectrum. Cognitive Radio Ad Hoc Networks (CRAHNs) have emerged as a possible solution to both the scarcity and overcrowding challenges of the spectrum. The CRAHNs ensure that the Secondary Users (SUs) do co-exist with Primary Users (PUs) in a non-interfering manner. The SUs access the licensed spectrum opportunistically when they are idle. CRAHNs have many use cases which include intermittent networks here referred to as intermittent CRAHNs (ICRAHNs). For example, the Military (MCRAHNs). MCRAHN is complex and characterized by a dynamic topology which is subject to frequent partitioning and route breakages due to attacks and destruction in combat. This study optimizes the routing protocols for intermittent networks such as the MCRAHNs. ICRAHN routing is a challenge due to the network’s intermittent attribute, which is subject to destruction in the case of MCRAHN which is characterized by frequent link breakages. The performance of the proposed routing scheme was evaluated through network simulations using the following metrics: throughput, and Routing Path delay, Node Relay delay, Spectrum Mobility delay. The simulation results show that the MAODV is the best-performing algorithm.
One-way ECS (Electric Car Sharing Service) is attracting attention as a new sustainable mobility option in urban areas. On the other hand, the vehicle uneven distribution problem occurs in one-way ECSs due to their usage patterns. In this paper, we propose a vehicle return prediction model for vehicle relocation to solve this problem. In the proposed method, two machine learning models are created to predict where and when a user will return a vehicle using static information such as departure time and location, and dynamic information such as the vehicle’s current location and direction of movement. The model is used to continually update the prediction results of vehicle returns during use, aiming for more accurate predictions. The proposed method has been evaluated using actual data from a one-way ECS and has achieved an accuracy of 0.93 for the prediction of stations to be returned. The method also achieved a MAE of 42.3 min and MAPE of 47
Utilizing edge and cloud computing to empower the profitability of manufacturing is drastically increasing in modern industries. As a result of that, several challenges have raised over the years that essentially require urgent attention. Among these, coping with different faults in edge and cloud computing and recovering from permanent and temporary faults became prominent issues to be solved. In this paper, we focus on the challenges of applying fault tolerance techniques on edge and cloud computing in the context of manufacturing and we investigate the current state of the proposed approaches by categorizing them into several groups. Moreover, we identify critical gaps in the research domain as open research directions.
In search by pattern in GPS trajectories, user draws a trajectory, the pattern query, and then receives a set of trajectories ranked by their similarity to the pattern query. We argue that when user draws a pattern query, an initial part of this query (prefix of chosen length) should have more weight than the rest of query. We assume that after receiving a set of similar trajectories, user can refine the pattern query in order to receive more relevant results. We give explanation of our approach by means of web search, where a user searches, for example, for “bratislava castle” and then adds a refinement to this query “opening hours”, where removing the initial part of query does not make sense, as search for “opening hour” alone would return irrelevant results. This idea has led us to considering pattern search that is weighted toward query prefix. We experimentally evaluate this approach, in our experimentation we apply the Geolife data set (Microsoft Research Asia).
We configure embedded devices with a smartphone via NFC using an open, platform independent protocol presented in this paper. A textual device specification defines the types of configuration values for a specific device and integrates the device into the configuration system. The specification needs to be provided by the embedded developer. It is translated into a C library that enables configuration value access, as well as blob that contains the compressed configuration metadata. A generic smartphone application interprets the metadata and configuration data read via NFC and allows the modification of the values according to the device specification encoded in the metadata. The modified configuration data can be stored, shared or transferred back to the embedded device. None of the configuration steps need an internet connection, which means data is kept private. Combined with the open protocol and the generic app, this ensures that embedded devices will not become obsolete through vendor decisions, as happens frequently with devices dependent on configuration via cloud services. Embedded developers only need to implement raw read and write binary access to an NFC storage device. The generated artifacts allow to transform that data into an easy-to-use data structure. A prototype system using a fully functional tool chain, a generic Android app and a single-board computer simulating an embedded device has been implemented and evaluated.
Multimodal problems are omnipresent in the real world: autonomous driving, robotic grasping, scene understanding, etc... Instead of proposing to improve an existing method or algorithm: we will use existing statistical methods to understand the features in already-existing neural networks. More precisely, we demonstrate that a fusion method relying on Canonical Correlation Analysis on features extracted from Deep Neural Networks using different sensors is equivalent to looking at the output of the networks themselves.
Performance analysis of cloud computing server provides the basis for ensuring Quality of Service (QoS), and the service strategy of server will directly affect the analysis of performance indicators. The performance indicators of QoS are usually defined in the form of Service Layer Agreement (SLA), such as the average response time, the average queue length, immediate service probability and so on. In this work, Service performance analysis models based on Geo/G/1 queuing system and queuing system with the vacation of the server are proposed. In these models, we analyze the main performance indicators of cloud computing server for the different parameters: the time between arrive of the task, the time of service, and the time of the provision of vacation. Furthermore, we discuss the optimizing concurrent number of the cloud computing.
Unmanned aerial vehicles (UAV) or drones play many roles in a modern smart city such as the delivery of goods, mapping real-time road traffic and monitoring pollution. The ability of drones to perform these functions often requires the support of machine learning technology. However, traditional machine learning models for drones encounter data privacy problems, communication costs and energy limitations. Federated Learning, an emerging distributed machine learning approach, is an excellent solution to address these issues. Federated learning (FL) allows drones to train local models without transmitting raw data. However, existing FL requires a central server to aggregate the trained model parameters of the UAV. A failure of the central server can significantly impact the overall training. In this paper, we propose two aggregation methods: Commutative FL and Alternate FL, based on the existing architecture of decentralised Federated Learning for UAV Networks (DFL-UN) by adding a unique aggregation method of decentralised FL. Those two methods can effectively control energy consumption and communication cost by controlling the number of local training epochs, local communication, and global communication. The simulation results of the proposed training methods are also presented to verify the feasibility and efficiency of the architecture compared with two benchmark methods (e.g. standard machine learning training and standard single aggregation server training). The simulation results show that the proposed methods outperform the benchmark methods in terms of operational stability, energy consumption and communication cost.
With the rapidly evolving mobile technology, governments are delivering services to the citizen through a mobile platform. These services include administrative services, health services, and awareness campaigns. To effectively provide mobile services to citizens, it is necessary to understand user perceptions of these services thoroughly. Therefore, the acceptance rate is influenced by a variety of factors. These elements are categorized as social, technological, cultural, personal, or facilitating. This paper aims to present a study on the acceptance of the Mobile-Government (M-Government) system in Saudi Arabia. One of the primary goals of this research is to promote M-Government adoption in developing countries such as Saudi Arabia. As a result, a study is being carried out to determine ‘How citizens’ cultures and attitudes affect the acceptability of M-Government?’ By identifying and analyzing cultural influences on M-Government, it is possible to understand people’s needs better. The primary aim of this research is to identify the limitations and research gaps in previous studies and broaden the scope of technology acceptance models to determine the acceptance rate of M-Government services. The Technology Acceptance Model (TAM) and the Unified Theory of Acceptance and Use of Technology (UTAUT) were used to investigate the impact of various factors on M-Government system acceptance. Previous studies’ limitations, which are addressed in this paper, include more appropriate constructs added to the models for hypothesis building. These hypotheses are based on Saudi Arabia’s demographic profiles, sociological and technological foundations. The findings will help policymakers, and government officials better understand the factors that influence service’s user acceptance.
A disconnect is frequent regarding the length of time a person claims to have brushed their teeth and the actual duration; the recommended brushing duration is 2 min. This paper seeks to bridge this particular disconnect. We introduce YouBrush,—a low-latency, low-friction, and responsive mobile application—to improve oral care regimens in users. YouBrush is an IOS mobile application that democratizes features previously available only to intelligent toothbrush users by incorporating a highly accurate deep learning brushing detection model—developed by Apple’s createML—on the device. The machine learning model, running on the edge, allows for a low-latency, highly responsive scripted-coaching brushing experience for the user. Moreover, we craft in-app gamification techniques to further user interaction, stickiness, and oral care adherence.
The Attribute-Based Access Control (ABAC) model is widely used for IoT due to its capacity to express access policies through attributes, making this method granular and flexible. However, if we assume that attributes are essentially mutable, the irreducible network latency and the architectures proposed to acquire a better communication performance of the IoT expose the point where those policies are evaluated as outdated attributes. Therefore, access policies can be wrongly evaluated, resulting in consistency and security problems. In this paper, we propose a method to reduce this exposure through a bi-directional attribute synchronization capable of mapping all attributes and evaluating their current consistency after a change. If the modified attribute does not affect the access, it will remain valid. Otherwise, a revocation occurs, reducing the risks of unintended accesses. Our modeling allows demonstrating the correctness of our method and its capability to revoke every unintended access that may occur after an attribute change.
To personalize modern mobile services (e.g., advertisement, navigation, healthcare) for individual users, mobile apps continuously collect and analyze sensor data. By sharing their sensor data collections, app providers can improve the quality of mobile services. However, the data privacy of both app providers and users must be protected against data leakage attacks. To address this problem, we present differentially privatized on-device sharing of sensor data, a framework through which app providers can safely collaborate with each other to personalize their mobile services. As a trusted intermediary, the framework aggregates the sensor data contributed by individual apps, accepting statistical queries against the combined datasets. A novel adaptive privacy-preserving scheme: 1) balances utility and privacy by computing and adding the required amount of noise to the query results; 2) incentivizes app providers to keep contributing data; 3) secures all data processing by integrating a Trusted Execution Environment. Our evaluation demonstrates the framework’s efficiency, utility, and safety: all queries complete in <10 ms; the data sharing collaborations satisfy participants’ dissimilar privacy/utility requirements; mobile services are effectively personalized, while preserving the data privacy of both app providers and users.
Community detection has been widely studied from many different perspectives, which include heuristic approaches in the past and graph neural network in recent years. With increasing security and privacy concerns, community detectors have been demonstrated to be vulnerable. A slight perturbation to the graph data can greatly change the detection results. In this paper, we focus on dealing with a kind of attack on one of the communities by manipulating the graph structure. We formulate this case as target community problem. The big challenge to solve this problem is the universality on different detectors. For this, we define structural information gain (SIG) to guide the manipulation and design an attack algorithm named SIGM. We compare SIGM with some recent attacks on five graph datasets. Results show that our attack is effective on misleading community detector.
With the development of intelligent transportation system, the detection method of traffic signs plays an important role in unmanned driving. However, due to the real-time and reliability characteristics of the automatic driving system, each traffic sign needs to be processed in a specific time interval to ensure the precision of the test results. Automatic driving is developing rapidly and has made great progress. Various traffic sign detection algorithms are proposed. Especially, convolutional neural network algorithm is concerned because of its fast execution and high recognition rate. But in the real world of complex traffic conditions, those algorithms still have problems such as poor real-time detection, low precision, false detection and high missed detection rate. To overcome those problems, this paper proposed an improved algorithm named as YOLO-RFB based on YOLO V4 network. Based on YOLO V4 network, the main feature extraction network is pruned, and convolution layer is replaced by RFB structure in two output feature layers. In the detection results of GTSDB data sets, the mAP of improved algorithm achieves 85.59
Modern embedded systems—autonomous vehicle-to-vehicle communication, smart cities, and military Joint All-Domain Operations— feature increasingly heterogeneous distributed components. As a result, existing communication methods, tightly coupled with specific networking layers and individual applications, can no longer balance the flexibility of modern data distribution with the traditional constraints of embedded systems. To address this problem, this paper presents a domainspecific language, designed around the Representational State Transfer (REST) architecture, most famously used on the web. Our language, called the Communication Language for Embedded Systems (CLES), supports both traditional point-to-point data communication and allocation of decentralized distributed tasks. To meet the traditional constraints of embedded execution, CLES’s novel runtime allocates decentralized distributed tasks across a heterogeneous network of embedded devices, overcoming limitations of centralized management and limited operating system integration. We evaluated CLES with performance micro-benchmarks, implementation of distributed stochastic gradient descent, and by applying it to design versatile stateless services for vehicleto-vehicle communication and military Joint All-Domain Command and Control, thus meeting the data distribution needs of realistic cyberphysical embedded systems.
A major theme in the study of social dynamics is the formation of a community structure on a social network, i.e., the network contains several densely connected region that are sparsely linked between each other. In this paper, we investigate the network integration process in which edges are added to dissolve the communities into a single unified network. In particular, we study the following problem which we refer to as togetherness improvement: given two communities in a network, iteratively establish new edges between the communities so that they appear as a single community in the network. Towards an effective strategy for this process, we employ tools from structural information theory. The aim here is to capture the inherent amount of structural information that is encoded in a community, thereby identifying the edge to establish which will maximize the information of the combined community. Based on this principle, we design an efficient algorithm that iteratively establish edges. Experimental results validate the effectiveness of our algorithm for network integration compared to existing benchmarks.
Blockchain can provide trusted ledgers on distributed architecture without the help of any central authority. Since all the transactions are saved in the ledgers, they can be obtained in public. By using the transactions, this paper first proposes a blockchain-based Diffie-Hellman key agreement (BDKA) protocol. Then, a blockchain-based group key agreement (BGKA) protocol is further proposed. In addition, both BDKA and BGKA protocols are implemented in the Bitcoin system. The performance of protocol execution and transaction fee are analyzed in the experiments.
Human activity recognition (HAR) has been adopting deep learning to substitute well-established analysis techniques that rely on hand-crafted feature extraction and classication techniques. However, the architecture of convolutional neural network (CNN) models used in HAR tasks still mostly uses VGG-like models while more and more novel architectures keep emerging. In this work, we present a novel approach to HAR by incorporating elements of residual learning in our ResNet-like CNN model to improve existing approaches by reducing the computational complexity of the recognition task without sacrificing accuracy. Specifically, we design our ResNet-like CNN based on residual learning and achieve nearly 1% better accuracy than the state-of-the-art, with over 10 times parameter reduction. At the same time, we adopt the Saliency Map method to visualize the importance of every input channel. This enables us to conduct further work such as dimension reduction to improve computational efficiency or finding the optimal sensor node(s) position(s).
With the development of 5G, network attacking becomes more and more easy. Many system vulnerability are utilized to be attacked via 5G technology. It leads that the network attack frequency turn to high, and the network attack strength turns to strong. Among all network attack identification methods, outlier detection is one of the most important one. It aims to find data which is much different from most of the others. In this paper, we propose an outlier detection based framework to support network-attack identification. It first uses a novel algorithm to construct core point set so as support efficiently outlier detection. Next, it uses a novel index named ZB-Tree to manage these core points. Thirdly, we propose a predictive IP-table to handle and predict suspicious IP addresses. In this way, we could identify most suspicious IP addresses based on the position relationships among different base stations. Theoretical analysis and extensive experimental results demonstrate the effectiveness of the proposed algorithms.
