International Journal of Grid and Utility Computing (83 papers in press)
Usage of DTNs for low cost IoT application in smart cities: performance evaluation of spray and wait routing protocol and its enhanced versions
by Evjola Spaho
Abstract: Delay Tolerant Networks (DTNs) can be used as a low cost solution to implement different applications of the Internet of Things (IoT) in a smart city. An issue that needs to be solved when this approach is used is the efficient transmission of data. In this paper, we create a DTN for a smart city IoT application and enhance the Binary Spray and Wait (B-S&W) routing protocol to improve delivery probability and average delay. We evaluate and compare the B-S&W routing protocol and our two enhanced versions of spray and wait (S&W-V1 and S&W-V2). The simulation results show that the proposed versions S&W-V1 and S&W-V2 improve the delivery probability and average latency.
Keywords: IoT; smart cities; delay tolerant networks; wireless Sensor networks; routing protocols; spray and wait protocol.
Public key encryption with equality test for vehicular system based on near-ring
by Muthukumaran Venkatesan, Ezhilmaran Devarasaran
Abstract: In recent years, vehicles have been increasingly integrated with an intelligent transport system (ITS). This has led to the development of Vehicular Ad hoc Networks(VANETs) through which the vehicles communicate with each other in an effective manner. Since VANET assists in both vehicle to vehicle and vehicle to infrastructure communication the matter of security and privacy has become a major concern. In this context, this work presents a public key Encryption with equality test based on DLP with decomposition problems over near-ring The proposed method is highly secure
and it solves the problem of quantum algorithm attacks in VANET systems. Further, the
proposed system prevents the chosen-ciphertext attack in type-I adversary and it is indistinguishable against the random oracle model for the type-II adversary. The proposed scheme is highly secure and the security analysis measures are stronger than existing techniques.
Keywords: near-ring; Diffie-Hellman; vehicular ad hoc networks.
FOGSYS: a system for the implementation of StaaS service in fog computing using embedded platforms
by José Dos Santos Machado, Danilo Souza Silva, Raphael Silva Fontes, Adauto Cavalcante Menezes, Edward David Moreno, Admilson De Ribamar Lima Ribeiro
Abstract: This work presents the concept of fog computing, its theoretical contextualisation, and related works, and develops the FogSys system with the main objective of simulating, receiving, validating and storing data from IoT devices to be transferred to cloud computing. Fog computing serves to provide the StaaS (Storage as a Service) service. The results showed that the implementation of this service in devices of embedded systems can be a good alternative to reduce one of these problems, in this case, the storage of data, which is faced currently by IoT devices.
Keywords: fog computing; cloud computing; IoT; embedded systems; StaaS.
Research on modelling analysis and maximum power point tracking strategies for distributed photovoltaic power generation systems based on adaptive control technology
by Yan Geng, Jianwei Ji, Bo Hu, Yingjun Ju
Abstract: As is well-known, the distributed photovoltaic power generation technology has been rapidly developed in recent years. The cost of distributed photovoltaic power generation is much higher than that of traditional power generation modes. Therefore, how to improve the effective use of photovoltaic cells has become a popular research direction. Based on the analysis of the characteristics of photovoltaic cells, this paper presents a mathematical model of photovoltaic cells and a maximum power point tracking algorithm based on hysteresis control and adaptive control technology variable step perturbation observation method. This algorithm can balance the control precision and control speed from the disturbance observation method and improve the tracking results significantly. Finally, the feasibility of the algorithm and the tracking effects are simulated by using Matlab/Simulink software.
Keywords: distributed photovoltaic; adaptive control technology; maximum power point tracking strategies.
Cloud infrastructure planning: models considering an optimisation method, cost and performance requirements
by Jamilson Dantas, Rubens Matos, Carlos Melo, Paulo Maciel
Abstract: Over the years, many companies have employed cloud computing systems as the best choice regarding the infrastructure to support their services, while keeping high availability and performance levels. The assurance of the availability of resources, considering the occurrence of failures and desired performance metrics, is a significant challenge for planning a cloud computing infrastructure. The dynamic behaviour of virtualised resources requires special attention to the effective amount of capacity that is available to users, so the system can be correctly sized. Therefore, planning computational infrastructure is an important activity for cloud infrastructure providers to analyse the cost-benefit trade-off among distinct architectures and deployment sizes. This paper proposes a methodology and models to support planning and the selection of a cloud infrastructure according to availability, COA, performance and cost requirements. An optimisation model based on GRASP meta-heuristic is used to generate a cloud infrastructure with a number of physical machines and Virtual Machines (VM) configurations. Such a system is represented using an SPN model and closed-form equations to estimate cost and dependability metrics. The proposed method is applied in a case study of a video transcoding service hosted in a cloud environment. The case study demonstrates the selection of cloud infrastructures with best performance and dependability metrics, considering the use of VP9, VP8 and H264 video codecs, as well as distinct VM setups. The results show the best configuration choice considering a six user profile. The results also show the computation of the probability of finalising a set of video transcoding jobs by a given time.
Keywords: cloud computing; performance; availability modelling; GRASP; COA; stochastic Petri nets; cost requirements.
Crowdsensing campaigns management in smart cities
by Carlos Roberto De Rolt, Julio Dias, Eliza Gomes, Marcelo Buosi
Abstract: The growth of cities is accompanied by a large number of different problems in the urban environment that makes effective management of public services a hard task. The use of information technology is one way to help in solving urban problems, aiming for the development of smart cities. Crowdsensing mechanism is an important tool in this process, exploring a collective intelligence and organising a collaboration of large groups of people. This work focuses mainly on the process of management of crowdsensing campaigns contributing to the theoretical framework regarding the theme. The learning in a use case contributed to improving the technical requirements of the computational platform used. Through a crowdsensing system, collaborative data collection and sensor monitoring campaigns were executed, which allowed learning about the management of crowdsensing campaigns, with results such as adjustments in the computational platform by the insertion of new types of campaign and the inclusion of feedback elements. This work reports the process of implementation and improvement of a crowdsensing system, which was initially developed from theoretical knowledge and deployed in the University of Bologna where students participated in campaigns managed through a computational platform entitled ParticipACT, resulting in several studies about the subject. In another context, based on this pioneering experience, the computer platform, ParticipACT, was transferred to LabGES, the Management Technologies Laboratory of UDESC (Santa Catarina State University), based on an international cooperation agreement. Collaborative data collection campaigns were carried out in a monitored way that enabled learning about crowdsensing campaigns management, resulting in significant contributions to the improvement of the system and propositions of adjustments in the theoretical framework of management campaign models.
Keywords: crowdsensing; smart cities; campaign management; ParticipACT.
A knowledge- and intelligence-based strategy for resource discovery on IaaS cloud systems
by Mohammad Samadi Gharajeh
Abstract: Resource discovery selects appropriate computing resources in cloud systems to accomplish the users jobs. This paper proposes a knowledge- and intelligence-based strategy for resource discovery in IaaS cloud systems, called KINRED. It uses a fuzzy system, a multi-criteria decision making (MCDM) controller, and an artificial neural node to discover suitable resources under various changes on network metrics. The suggested fuzzy system uses hardware specifications of the computing resources in which CPU speed, CPU core, memory, disk, the number of virtual machines, and usage rate are considered as inputs, and hardware type is considered as output of the system. The suggested MCDM controller makes proper decisions based on users requirements in which CPU speed, CPU core, memory, and disk are assumed as inputs, and job type is assumed as output of the controller. Furthermore, the artificial neural node selects the computing resource having the highest success rate based on both outputs of the fuzzy system and MCDM controller. Simulation results show that the proposed strategy surpasses some of the existing related works in terms of the number of successful jobs, system throughput, and service price.
Keywords: cloud computing; resource discovery; knowledge-based system; intelligent strategy; artificial neural node.
Performance impact of the MVMM algorithm for virtual machine migration in data centres
by Nawel Kortas, Habib Youssef
Abstract: Virtual machine (VM) migration mechanisms and the design of data centres for cloud computing have a significant impact on energy cost and negotiated Service Level Agreement (SLA). The recent work focuses on how to use VM migration to achieve stable physical machine (PM) usage with the objective of reducing energy consumption, under stated SLA constraints. This paper presents and evaluates a new scheduling algorithm called MVMM (Minimisation of Virtual Machine Migration) for VM migration within a data centre environment. MVMM makes use of a DBN (Dynamic Bayesian Network) to decide where and when a particular VM migrates. Indeed, the DBN takes as input the data centre parameters then computes a score for each VM candidate for migration in order to reduce the energy consumption by decreasing the number of future migrations according to the probabilistic dependencies between the data centre parameters. Furthermore, our performance study shows that the choice of a data centre scheduling algorithm and network architecture in cloud computing significantly impacts the energy cost and application performance under resource and service demand variations. To evaluate the proposed algorithm, we integrated the MVMM scheduler into the GreenCloud simulator while taking into consideration key data centre characteristics such as scheduling algorithm, DCN (Data re Network) architecture, link, load and communication between VMs. The performance results show that the use of the MVMM scheduler algorithm within a three-tier debug architecture can reduce energy consumption by over 35% when compared with five well-known schedulers, namely Round Robin, Random, Heros, Green, and Dens.
Keywords: MVMM algorithm; virtual machine; cloud computing; dynamic Bayesian networks; SLA; scheduler algorithm; data centre network architectures; VM migration.
SDSAM: a service-oriented approach for descriptive statistical analysis of multidimensional spatio-temporal big data
by Weilong Ding, Zhuofeng Zhao, Jie Zhou, Han Li
Abstract: With the expansion of the Internet of Things, spatio-temporal data has been widely used and generated. The rise of big data in space and time has led to a flood of new applications with statistical analysis characteristics. In addition, applications based on statistical analysis of these data must deal with the large capacity, diversity and frequent changes of data, as well as the query, integration and visualisation of data. Developing such applications is essentially a challenging and time-consuming task. In order to simplify the statistical analysis of spatio-temporal data, a service-oriented method is proposed in this paper. This method defines the model of spatio-temporal data service and functional service. It defines a process-based application of spatio-temporal big data statistics to invoke basic data services and functional services, and proposes an implementation method of spatio-temporal data service and functional service based on Hadoop environment. Taking the highway big data analysis as an example, the validity and applicability of this method are verified. The effectiveness of this method is verified by an example. The validity and applicability of the method are verified by a case study of Expressway large data analysis. An example is given to verify the validity of the method.
Keywords: spatio-temporal data; RESTful; web service.
Personality-aware recommendations: an Empirical study in education
by Yong Zheng, Archana Subramaniyan
Abstract: Recommender systems have been developed to deliver item recommendations to the users tailored to user preferences. The impact of the human personality has been realised in user decision making. There are several personality-aware recommendation models which incorporate the personality traits into the recommendations. They have been demonstrated to be effective in improving the quality of the recommendations in several domains, including movies, music and social networks. However, the impact on the area of education is still under investigation. In this paper, we discuss and summarise state-of-the-art personality-based collaborative filtering techniques for recommendations, and perform an empirical study on educational data. Particularly, we collect the personality traits in two ways: a user survey and a natural language processing system. We examine the effectiveness of the recommendation models by using these subjective and inferred personality traits, respectively. Our experimental results reveal that students with different personality traits may make different choices, and the inferred personality traits are more reliable and effective to be used in the process of recommendations.
Keywords: personality; recommender systems; education; empirical study.
Research on integrated energy system planning method considering wind power uncertainty
by Yong Wang, Yongqiang Mu, Jingbo Liu, Yongyi Tong, Hongbo Zhu, Mingfeng Chen, Peng Liang
Abstract: With the development of energy technology, the planning and operation of integrated energy systems coupled with electricity-gas-heat energy has become an important research topic in the future energy field. In order to solve the influence of wind power uncertainty on the unified planning of integrated energy systems, this paper constructs a wind energy uncertainty quantitative model based on intuitionistic fuzzy sets. Based on this, an integrated energy system planning model with optimal economic cost and environmental cost is established. The model is solved by the harmonic search algorithm. Finally, the proposed method is validated by simulation examples. The effectiveness of the integrated energy system planning method can improve the grid capacity of the wind power and reduce the CO2 of the system. And it has guiding significance for the long-term planning of integrated energy systems
Keywords: wind power uncertainty; planning method; electricity-gas-heat energy.
Research on design method of manoeuvring targets tracking generator based on LabVIEW programming
by Caiyun Gao, Shiqiang Wang, Huiyong Zeng, Juan Bai, Binfeng Zong, Jiliang Cai
Abstract: Aiming at the issue of poor visual display and non-real-time status output while describing maneuvering target track with data, a new design method of target track generator is proposed based on laboratory virtual instrument engineering workbench (LabVIEW). Firstly, the motion model of maneuvering target is builded. Secondly, the design requirement of track generator is discussed. Finally, target track of multiple targets and multiple maneuvering model is produced by visual panel and code design with LabVIEW. Simulation results indicate that the proposed method can output the target status in real time with different data rates while displaying the multiple targets maneuvering track directly and have good visibility. And also the generated track parameters are of high accuracy and effective data.
Keywords: LabVIEW; virtual instrument; target track simulation; situation display of radar.
Finite state transducer based light-weight cryptosystem for data confidentiality in cloud computing
by Basappa Kodada, Demian Antony D'Mello
Abstract: Cloud computing is derived from parallel, cluster, grid and distributed computing and is becoming one of the advanced and growing technologies. With the rapid growth of internet technology and its speed, the number of users for cloud computing is growing enormously, and huge amounts of data are being generated. With the growth of data in cloud, the security and safety of data, such as data confidentiality and privacy, are a paramount issue because data plays a vital role in the current trend. This paper proposes a new type of cryptosystem based on a finite state transducer to provide data confidentiality for cloud computing. The paper presents the protocol communication process and gives an insight into security analysis on the proposed scheme. The scheme proves that it is stronger and more secure than the existing schemes that can be derived from results as proof of concept.
Keywords: security; confidentiality; encryption; decryption; automata; finite state
machine; finite state transducer; cryptography; data safety.
FastGarble: an optimised garbled circuit construction framework
by A. Anasuya Innocent, G. Prakash, K. Sangeeta
Abstract: In the emerging field of cryptography, secure computation can be used to solve a number of distributed computing applications without loss of privacy of sensitive/ private data. The applications can be as simple as coin tossing or agreement between parties, or as complex as e-auctions, e-voting, or private data retrieval for the purpose of carrying out research on sensitive data, private editing on cloud, etc., without the help of a trusted third party. Confidentiality can be achieved by the use of conventional cryptographic techniques, but they require the data availability for working. For working on sensitive data some other technique is needed, and there comes the use of secure computation. Any protocol on secure computation starts with the construction of a garbled circuit of the underlying functionality, and the efficiency of protocol and circuit construction are directly proportional to each other. Hence, as the complexity of an application increases, the circuit size increases, resulting in poor efficiency of the protocol, which in turn restricts secure computation from finding its use in day-to-day applications. In this paper, an optimised garbled circuit construction framework, named FastGarble, is proposed, which has been shown to improve the time complexity of garbled circuit construction.
Keywords: secure computation; garbled circuit; performance; secure two-party computation; time complexity.
Fine-grained access control of files stored in cloud storage with traceable and revocable multi-authority CP-ABE scheme
by Bharati Mishra, Debasish Jena, Srikanta Patnaik
Abstract: Cloud computing is gaining increasing popularity among enterprises,
universities, government departments, and end-users. Geographically distributed users can collaborate by sharing files through the cloud. Ciphertext-policy attribute-based (CP-ABE) access control provides an efficient technique to enforce fine-grained access control by the data owner. Single authority CP-ABE schemes create a bottleneck for enterprise applications. Multi-authority CP-ABE systems deal with multiple attribute authorities performing the attribute registration or key distribution. Type I pairing is used in designing the existing multi-authority systems. They are vulnerable to some reported known attacks on them. This paper proposes a multi-authority CP-ABE scheme that supports attribute and policy revocation. Type III pairing is used in designing the scheme, which has higher security, faster group operations, and requires less memory to store the elements. The proposed scheme has been implemented using the Charm framework, which uses the PBC library. The OpenStack cloud platform is used for computing and storage services. It has been proved that the proposed scheme is collusion resistant, traceable, and revocable. AVISPA tool has been used to verify that the proposed scheme is secure against a replay attack and man-in-the-middle attack.
Keywords: cloud storage; access control; CP-ABE; attribute revocation; blockchain; multi-authority.
On generating Pareto optimal set in bi-objective reliable network topology design
by Basima Elshqeirat, Ahmad Aloqaily, Sieteng Soh, Kwan-Wu Chin, Amitava Datta
Abstract: This paper considers the following NP-hard network topology design (NTD) problem called NTD-CB/R: given (i) the location of network nodes, (ii) connecting links, and (iii) each links reliability, cost and bandwidth, design a topology with minimum cost (C) and maximum bandwidth (B) subject to a pre-defined reliability (R) constraint. A key challenge when solving the bi-objective optimisation problem is to simultaneously minimise C while maximising B. Existing solutions aim to obtain one topology with the largest bandwidth cost ratio. To this end, this paper aims to generate the best set of non-dominated feasible topologies, aka the Pareto Optimal Set (POS). It formally defines a dynamic programming (DP) formulation for NTD-CB/R. Then, it proposes two alternative Lagrange relaxations to compute a weight for each link from its reliability, bandwidth, and cost. The paper further proposes a DP approach, called DPCB/R-LP, to generate POS with maximum weight. It also describes a heuristic to enumerate only k?n paths to reduce the computational complexity for a network with n possible paths. Extensive simulations on hundreds of various sized networks that contain up to 299 paths show that DPCB/R-LP can generate 70.4% of the optimal POS while using only up to 984 paths and 27.06 CPU seconds. With respect to a widely used metric, called overall-Pareto-spread (OR), DPCB/R-LP produces 94.4% of POS with OS = 1, measured against the optimal POS. Finally, all generated POS each contains a topology that has the largest bandwidth cost ratio, significantly higher than 88% obtained by existing methods.
Keywords: bi-objective optimisation; dynamic programming; Lagrange relaxation; Pareto optimal set; network reliability; topology design.
Dynamic quality of service for different flow types in SDN networks
by Alessandro Lima, Eduardo Alchieri
Abstract: The structure of the internet makes it difficult to implement Quality of Service (QoS) in different flows generated by many different applications, ranging from an e-commerce application with a light demand to real-time applications such as VoIP or videoconferencing, which make heavy use of the internet. One of the challenges is the lack of technical knowledge and the difficulty of configuring network equipment with many proprietary technologies. Software Defined Networks (SDN) are a good alternative to mitigate these problems. By separating the control plane from the data plane, network administrators can efficiently use the network resources and, moreover, it is easier to provide new services and applications tailored to the needs of the network. However, SDN technology itself still suffers from the limitation of solid QoS mechanisms, especially considering flows classified as elephant (large data volume), cheetah (high throughput) and alpha (large bursts). Aiming to fill this gap, this work proposes a new SDN service, called QoS-Flux, that receives network information from the data plane, through the OpenFlow protocol, to apply different QoS algorithms and filters to dynamically deal with different flows. Experimental results show that QoS-Flux significantly improves the QoS metrics of delay, jitter, packet loss, and bandwidth in a SDN network.
Keywords: SDN; quality of service; elephant flow; alpha flow; cheetah flow.
Optimal controller design for an islanded microgrid during load change
by Bineeta Soreng, Raseswari Pradhan
Abstract: There is a high tendency of voltage and frequency variation in islanding mode compared with grid connected mode. This paper emphasises on developing a technique for optimal regulation of voltage and frequency for a Microgrid (MG). Here, the studied microgrid is a Photovoltaic (PV) based MG (PVMG). The proposed technique is a Sliding Mode Controller (SMC) optimised using the Whales Optimisation Algorithm (WOA) named as SMC-WOA. The effectiveness of the proposed technique is validated by the dynamic response of the studied PVMG during operation mode and load change. For controlling of the studied PVMG, two loops, namely voltage loop and current loop, are used. Again, droop controller is used for power sharing in the studied PVMG. For ensuring the efficacy of the proposed technique SMC-WOA, dynamic responses of the studied system with SMC-WOA are compared with that of the Grey Wolf Optimization (GWO) based SMC (SMC-GWO) and Sine Cosine Algorithm (SCA) based SMC (SMC-SCA). With proper analysis of the simulation results, it is found that the proposed SMC-WOA helps in yielding better results compared with SMC-GWO and SMC-SCA techniques in terms of faster solution with minimum voltage, frequency overshoot along with minimum output current and total harmonic distortion. The validation of the proposed technique is also tested by comparing with the PI controller optimised with the same WOA, GWO and SCA.
Keywords: microgrids; PI controller; sliding mode; whales optimisation algorithm; grey wolf optimisation; sine cosine algorithm; total harmonic distortion.
HyperGuard: on designing out-VM malware analysis approach to detect intrusions from hypervisor in cloud environment
by Prithviraj Singh Bisht, Preeti Mishra, Pushpanjali Chauhan, R.C. Joshi
Abstract: Cloud computing provides delivery of computing resources as a service on a pay-as-you-go basis. It represents a shift from products being purchased, to products being subscribed as a service, delivered to consumers over the internet from a large scale data centre. The main issue with cloud services is security from attackers who can easily compromise the Virtual Machines (VMs) and applications running over them. In this paper, we present a HyperGuard mechanism to detect malware that hide their presence by sensing the analysing environment or security tools installed in VMs. They may attach themselves with legitimate processes. Hence, HyperGuard is deployed at the hypervisor, outside the monitored VMs to detect such evasive attacks. It employs open source introspection libraries, such as DRAKVUF, LIbVMI etc., to capture the VM behaviour from hypervisor inform of syscall logs. It extracts the features in the form of n-grams. It makes use of Recursive Feature Elimination (RFE) and Support Vector Machine (SVM) to learn and detect the abnormal behaviour of evasive malware. The approach has been validated with a publicly available dataset (Trojan binaries) and a dataset obtained on request from University of new California (evasive malware binaries). The results seem to be promising.
Keywords: Cloud secuirty,Intrusion detection,virtual machine
introspection,system call traces; machine learning ; anaomaly
behviour detection; sypder.
Method for determining cloth simulation filtering threshold value based on curvature value of fitting curve
by Jialong Sun
Abstract: Cloth simulation filtering (CSF) is an algorithm that is effectively applied to mobile 3D laser point cloud filtering. The classification threshold in cloth simulation is a key parameter for separating ground points from non-ground points. The selection of the classification threshold directly affects the point cloud filtering effect. In this paper, based on the filtering of the CSF algorithm, a method for calculating the threshold-filtering total error fitting curve curvature value to determine the cloth simulation classification threshold is proposed. First, according to the relationship between the classification threshold and filtering error in CSF, the least squares optimal fitting function is established. Then the curvature value of the fitted curve is calculated to determine the optimal classification threshold; Finally, the final filtering errors of ground points and non-ground points are analysed by examples, and the effectiveness of this method in CSF filtering is verified, and a good filtering effect is obtained.
Keywords: cloth simulation filtering; laser point cloud; filtering; threshold.
Cloud workflow scheduling algorithm based on multi-objective hybrid particle swarm optimisation
by Baomin Xu
Abstract: Particle swarm optimisation has been widely used in solving scheduling problems. This paper proposes a hybrid algorithm namely Hill Climbing with Multi-objective Particle Swarm Optimization (HCMOPSO), which is based on heuristic local search and multi-objective particle swarm optimisation algorithm. HCMOPSO introduces hill climbing optimisation techniques into the particle swarm optimisation algorithm to improve the local search ability. Experimental results show that HCMOPSO is an effective cloud workflow scheduling algorithm, which has faster convergence velocity and better optimisation ability.
Keywords: hill climbing algorithm; task scheduling; particle swarm optimisation; cloud workflow.
Dynamic Bayesian network based prediction of performance parameters in cloud computing
by Priyanka Bharti, Rajeev Ranjan
Abstract: Resource prediction is an important task in cloud computing environments. It can become more effective and practical for large Cloud Service Providers (CSPs) with a deeper understanding of their Virtual Machines (VM) workload's key characteristics. Resource prediction is also influenced by several factors including (but not constrained to) data centre resources, types of user application (workloads), network delay and bandwidth. Given the increasing number of users for cloud systems, if these factors can be accurately measured and predicted, improvements in resource prediction could be even greater. Existing prediction models have not explored how to capture the complex and uncertain (dynamic) relationships between these factors owing to the stochastic nature of cloud systems. Further, they are based on score-based Bayesian network (BN) algorithms having limited prediction accuracy when dependency exists between multiple variables. This work considers time-dependent factors in cloud performance prediction. It considers an application of Dynamic Bayesian Network (DBN) as an alternative model for dynamic prediction of cloud performance by extending the static capability of a BN. The developed model is trained using standard datasets from Microsoft Azure and Google Compute Engine. It is found to be effective in predicting the application workloads and its resource requirements with an enhanced accuracy compared with existing models. Further, it leads to better decision making processes with regard to response time and scalability in dynamic situations of the cloud environment.
Keywords: cloud computing; dynamic Bayesian network; resource prediction; response time; scalability.
A privacy-aware and fair self-exchanging self-trading scheme for IoT data based on smart contract
by Yuling Chen, Hongyan Yin, Yaocheng Zhang, Wei Ren, Yi Ren
Abstract: With the development of the era of big data, the demand for data sharing and usage is increasing, especially in the era of Internet of things, thus putting forward a keen demand for data exchanging and data trading. However, the existing data exchanging and trading platforms are usually centralized and usersrnhave to trust platforms. This paper proposes a secure and fair exchanging and trading protocol based on blockchain and smart contract, especially, self-governance without relying centralized trust. By using the protocol, it can guarantee fairness to defend against trade cheating, and security for data confidentiality. It can also guarantee efficiency by transferring data links instead of data between data owners and data buyers. The extensive analysisrnjustified that the proposed scheme can facilitate the self-exchanging and self-trading for big data in a secure, fair and efficient manner.
Keywords: big data; IoT; fair exchanging; blockchain; smart contract; oblivious protocol; fair trading.
An agent-based mechanism to form cloud federations and manage their requirements changes
by Nassima Bouchareb, Nacer Eddine Zarour
Abstract: Cloud computing is a business paradigm, where cloud providers offer computing resources (software and hardware) and cloud consumers use them. Forming cloud federations and managing them is a big problem. In this paper, we propose an agent-based mechanism to automatically manage cloud federations and their requirements changes, to accept the maximum of requests with minimum cost and energy consumption, by soliciting the best clouds. First, we present two strategies: offer strategy and acceptance strategy, which allow the formation of federations. Then, we describe how to manage the requirements changes of these strategies. Finally, this paper presents a case study to illustrate our federation management mechanism in cloud computing. We evaluate the proposed strategies by comparing them with other related works. Simulation results indicate that the proposed policies enhance the providers profit.
Keywords: cloud computing; federation; requirements change; multi-agent systems; trust; utility; green computing.
Clustering structure-based multiple measurement vectors model and its algorithm
by Tijian Cai, Xiaoyu Peng, Xin Xie, Wei Liu, Jia Mo
Abstract: Most current multi-measurement vector models are based on the ideal assumption of shared sparse structure. However, owing to time-varying and multiple focuses of complex data, it is often difficult to meet the assumption in reality. Therefore, people had been working hard to use various sparse structures to solve the problem. In this paper, we take the cluster sparsity of signals into account and propose a Cluster Sparsity-based MMV (CS-MMV) model, which not only uses shared sparse structure between coefficients but also considers the cluster characteristic within coefficients. Furthermore, we extend a classic algorithm to implement the new model. Experiments on simulation data and two face benchmarks show that the new model is more suitable for complex data with clustered structure, and the extended algorithm can effectively improve the performance of sparse recovery.
Keywords: compressed sensing; sparse recovery; multi-measurement vectors; structured sparsity.
Micro-PaaS fog: container based orchestration for IoT applications using SBC
by Walter D.O. Santo, Rubens De Souza Matos Júnior, Admilson De Ribamar Lima Ribeiro, Danilo Souza Silva, Reneilson Yves Carvalho Santos
Abstract: The Internet of Things (IoT) is an emerging technology paradigm in which ubiquitous sensors monitor physical infrastructures, environments, and people in real-time to help in decision making and improve the efficiency and reliability of the systems, adding comfort and life quality to society. In this sense, there are questions concerning the limitation of computational resources, high latency and different QoS requirements related to IoT that move cloud technologies to the fog computing direction, and the adoption of light virtualised solutions, as technologies based in containers to attend to many needs of different domains. This work, therefore, has as its goal to propose and implement a micro-Paas architecture for fog computing, in a cluster of single-board computers (SBC), for orchestration of applications using containers, applied to IoT and that attend to the QoS criteria, e.g. high availability, scalability, load balance, and latency. From this proposed model, the micro-Paas fog was implemented with virtualisation technology in containers using orchestration services in a cluster built with Raspberry Pi to monitor water and energy consumption at a total cost of property equivalent to 23% of a public platform as a service (PaaS).
Keywords: fog computing; cluster; orchestration; containers; single board computing.
A review on data replication strategies in cloud systems
by Riad Mokadem, Jorge Martinez-Gil, Abdelkader Hameurlain, Joseph Kueng
Abstract: Data replication constitutes an important issue in cloud data management. In this context, a significant number of replication strategies have been proposed for cloud systems. Most of the studies in the literature have classified these strategies into static vs. dynamic or centralised vs. decentralised strategies. In this paper, we propose a new classification of data replication strategies in cloud systems. It takes into account several other criteria, specific to cloud environments: (i) the orientation of the profit, (ii) the considered objective function, (iii) the number of tenant objectives, (iv) the nature of the cloud environment and (v) the consideration of economic costs. Dealing with the last criterion, we focus on the provider's economic profit and the consideration of energy consumption by the provider. Finally, the impact of some important factors is investigated in a simulation study.
Keywords: cloud systems; data replication; data replication strategies; classification; service level agreement; economic profit; performance.
Anomaly detection against mimicry attacks based on time decay modelling
by Akinori Muramatsu, Masayoshi Aritsugi
Abstract: Because cyberattackers attempt to cheat anomaly detection systems, it is required to make an anomaly detection system robust against such attempts. We focus on mimicry attacks and propose a system to detect such attacks in this paper. Mimicry attacks make use of ordinary operations in order not to be detected. We take account of time decay in modelling operations to give lower priorities to preceding operations, thereby enabling us to detect mimicry attacks. We empirically evaluate our proposal with varying time decay rates to demonstrate that our proposal can detect mimicry attacks that could not be detected by a state-of-the-art anomaly detection approach.
Keywords: anomaly detection; mimicry attacks; time decay modelling; stream processing.
A cloud-based spatiotemporal data warehouse approach
by Georgia Garani, Nunziato Cassavia, Ilias Savvas
Abstract: The arrival of the big data era introduces new necessities for accommodating data access and analysis by organisations. The evolution of data is three-fold, increase in volume, variety, and complexity. The majority of data nowadays is generated in the cloud. Cloud data warehouses enhance the benefits of the cloud by facilitating the integration of cloud data in the cloud. A data warehouse is developed in this paper, which supports both spatial and temporal dimensions. The research focuses on proposing a general design for spatiobitemporal objects implemented by nested dimension tables using the starnest schema approach. Experimental results reflect that the parallel processing of such data in the cloud can process OLAP queries efficiently. Furthermore, increasing the number of computational nodes significantly reduces the time of query execution. The feasibility, scalability, and utility of the proposed technique for querying spatiotemporal data is demonstrated.
Keywords: cloud computing; big data; hive; business intelligence; data warehouses; cloud based data warehouses; spatiotemporal data; spatiotemporal objects; starnest schema; OLAP; online analytical processing.
A truthful mechanism for crowdsourcing-based tourist spots detection in smart cities
by Anil Bikash Chowdhury, Vikash Kumar Singh, Sajal Mukhopadhyay, Abhishek Kumar, Meghana M. D
Abstract: With the advent of new technologies and the internet around the globe, many cities in different countries are involving the local residents (or city dwellers) for making decisions on various government policies and projects. In this paper, the problem of detecting tourist spots in a city with the help of city dwellers, in strategic setting, is addressed. The city dwellers vote against the different locations that may act as potential candidates for tourist spots. For the purpose of voting, the concept of single-peaked preferences is used, where each city dweller reports a privately held single-peaked value that signifies the location in a city. Given the above discussed scenario, the goal is to determine the location in the city as a tourist spot. For this purpose, we have designed the mechanisms (one of which is truthful). For measuring the efficacy of the proposed mechanisms the simulations are done.
Keywords: tourism; smart cities; crowdsourcing; city dwellers; voting; single-peaked preferences; truthful.
A proactive population dynamics load-balancing algorithm in cloud computing for QoS optimisation
by Shahbaz Afzal, G. Kavitha
Abstract: Load unbalancing is currently a concern among Cloud Service Providers (CSP) that has adverse consequences in terms of both deliverable Quality of Service (QoS) and profit turnout. Load balancing as a band-aid tries to overcome load imbalances by ensuring proper task deployment among cloud resources, yielding productive resource use, throughput, profit, estimated deadline and other QoS metrics. However, load unbalancing, being an NP-hard optimisation problem, restricts to achieve the desired QoS results and hence enforcing the implementation of an efficient scheduling mechanism and load-balancing policy in cloud computing systems. A priority-based scheduling technique is proposed with a proactive load-balancing feature performed by set partitioning, parallel queues, and population dynamic model. Tasks are scheduled on virtual machines (VM) using priority scheduling. The set partitioning process performs task partitioning and VM partitioning assisted by filtering and sorting. Tasks and VMs are partitioned into eight distinct classes. The filtered and sorted tasks are passed through parallel queues to avoid waiting times. Finally, the proactive based population dynamic load balancing (PPDLB) model is applied to limit the tasks within the maximum carrying capacity of VM. The experimental results showed that PPDLB approach avoids load unbalancing in advance with higher resource use. The PPDLB algorithm was compared with the existing Improved Weighted Round Robin (IWRR), Harris Hawk optimisation (HHO) and Spider Monkey Optimisation (SMO) algorithms. The proposed algorithm was found to be more efficient than existing algorithms in terms of makespan, resource use, degree of balance and number of task migrations. The PPDLBA improves the makespan time by 3.29%, 5.15%, and 10.85% with respect to IWRR, HHO, and SMO algorithms, respectively. The degree of balance of PPDLBA is improved to 4.68%, 5.74%, and 6.26% when compared with IWRR, HHO and SMO algorithms, respectively. Also, the percentage improvement of PPDLBA in terms of resource use is 2.17%, 3.55%, and 3.99%, respectively, with existing algorithms. The dominance of the proposed algorithm over existing load-balancing techniques lies in the fact that it eliminates the need to solve migration associated metrics such as migration time, migration cost, number of VMs required for migration, and number of task migrations. Moreover, it has greater convergence rate when search space becomes large which makes it feasible for large scale scheduling problems.
Keywords: cloud computing; scheduling; load unbalancing; proactive load balancing; set partitioning; parallel queues; population dynamics; Quality of Service.
Dont hurry, be green: scheduling server shutdowns in grid computing with deep reinforcement learning
by Mauricio Pillon, Lucas Casagrande, Guilherme Koslovski, Charles Miers, Nelson Gonzales
Abstract: Grid computing platforms dissipate massive amounts of energy. Energy efficiency, therefore, is an essential requirement that directly affects their sustainability. Resource management systems deploy rule-based approaches to mitigate this cost. However, these strategies do not consider the patterns of the workloads being executed. In this context, we demonstrate how a solution based on deep reinforcement learning is used to formulate an adaptive power-efficient policy. Specifically, we implement an off-reservation approach to overcome the disadvantages of an aggressive shutdown policy and minimise the frequency of shutdown events. Through simulation, we train the algorithm and evaluate it against commonly used shutdown policies using real traces from GRID5000. Based on the experiments, we observed a reduction of 46% on the averaged energy waste with an equivalent frequency of shutdown events compared with a soft shutdown policy.
Keywords: deep reinforcement learning; grid computing; energy-aware scheduling; shutdown strategy; Markov decision process; resource management.
Authentication and authorisation in service oriented grid architecture
by Arbër Beshiri, Anastas Mishev
Abstract: Applications (services) in nowadays request access to resources that are mostly distributed over the network (wide-area network). These applications usually rely on by mediums such as Grid Computing Infrastructure (GCI) that enable them to be executed. GCI has heterogeneous nature and supports security as an essential part in grid systems. Grid Security Infrastructure (GSI) is a technology standard for grid security. Authentication and even authorisation estimate is a security challenge for grids. This paper discusses the authentication and authorisation infrastructures in the grid, including technologies that cover these two major parts of this domain. Here are surveyed the challenges that security encounters, namely grid authentication mechanisms, grid authorisation mechanisms and models. The combination of the grid authorisation technologies and grid authentication technologies with authorisation infrastructures enables role-based and fine-grained authorisation. Such technologies provide promising solutions for service (resources) oriented grid architecture authentication and authorisation.
Keywords: grid; service oriented grid architecture; authorisation; authentication; security.
FastIoT: an efficient and very fast compression model for displaying a huge volume of IoT data in web environments
by Mateus Melchiades, Cesar Crovato, Rodrigo Da Rosa Righi, Everton Nedel, Lincoln Schreiber
Abstract: The widespread adoption of the Industry 4.0 concepts, including the Internet of Things and Big Data, in industries worldwide, leads to the generation of massive datasets that supervisors must appropriately analyse for an effective decision-making process. A dataset, however, can be excessively large, causing troubles when trying to visualise its content entirety. Furthermore, while efficient for local data analysis, traditional compression systems present slowdowns when dealing with more than a few thousand points. Analysing the state-of-the-art, we did not find initiatives that combine a real-time retrieval of data and a good user experience when sliding and analysing extensive datasets. The present work introduces FastIoT as a novel compression model that focuses on the visual representation of Industry 4.0 data through web environments. As a client-server proposal, FastIoT brings the idea of: (i) speed in data preparation at the server-side, since the proposed method is very simple, and (ii) efficiency, because we consider the target client plotting area so generating an optimised dataset fitted especially for such visual region. We are focusing on client-server web deployments where a high latency internet network usually takes place, mainly when addressing data access on multi-branch companies. FastIoT reduces the file sizes more than 6000 times, which is crucial for large queries through the web. Even adding a short time at the server-side for data preparation, with FastIoT we have data ready in the clients display up to 97% faster when compared with traditional plotting methods.
Keywords: Industry 4.0; Big Data; IoT; compression; data visualisation; web environment.
IoT service distributed management architecture and service discovery method for edge-cloud federation
by Dongju Yang, Weida Zhang
Abstract: With the continuously increasing number of IoT (Internet of Things) services, the distributed management of IoT services becomes an inevitable trend. Under IoT and edge-cloud federation framework, the primary issues to solve in IoT service management are how to design a suitable distributed management architecture for IoT services, reduce network bandwidth overhead, reduce system latency and support the dynamic awareness of service status and rapid discovery of services. In this paper, the distributed management of IoT services is implemented by constructing layer-ring collaboration architecture on the cloud and edge, the service addressing channel is established on the cloud and edge using master and slave node cluster. The slave node is close to service and is dynamically aware of the service status. At the same time, the master-chord is constructed based on chord protocol among masters to enable the collaborated addressing of multiple master nodes on the cloud. So a strong service routing network on the cloud and edge is established to enable the distributed management of IoT services. This paper focuses on the service registration and discovery methods under this framework, and finally verifies the effectiveness of the method through the highway emergency scenario.
Keywords: IoT service; service discovery; service registration; distributed management architecture; edge-cloud federation.
Resource consumption trade-off for reducing hotspot migration in modern data centres
by Mouhebeddine Berrima, Walid Louhichi, Narjes Ben Rajeb
Abstract: Virtual machine migration is a basic mechanism to implement dynamic resource management such as resource restraints management, physical machine consolidation, and load balancing. These are necessary to reduce the operational costs of cloud providers and thus increase their business profits. However, the migration operations interrupt the execution of user services leading to a degradation of service performance. In this paper, we propose a virtual machine placement algorithm significantly reducing the number of hotspot migrations performed to manage resource restraints. The reduction overhead is an extra use of resources depending on some cloud environment settings. The proposed algorithm is designed for dynamic modern data centres housed in truck-towed containers, and thus can be relocated at sites where utilities for electricity are cheaper. Through empirical evaluation, we analyse the trade-off between hotspot migrations and resource consumption according to different parameters. Simulation results affirm that our algorithm is more efficient in data centres with a greater degree of physical machine virtualisation. Moreover, compared with a proactive migration algorithm, our algorithm promotes better the reduction of hotspot migrations in high dynamic data centres.
Keywords: cloud computing; modern data centre; resource consumption; hotspot migration; virtual machine placement algorithm.
Maximising the availability of an internet of medical things system using surrogate models and nature-inspired approaches
by Guto Leoni Santos, Demis Gomes, Francisco Airton Silva, Patricia Takako Endo, Theo Lynn
Abstract: The emergence of new computing paradigms such as fog and edge computing provides the internet of things with needed connectivity and high availability. In the context of e-health systems, wearable sensors are being used to continuously collect information about our health, and forward it for processing by the Internet of Medical Things (IoMT). E-health systems are designed to assist subjects in real-time by providing them with a range of multimedia-based health services and personalised treatment with the promise of reducing the economic burden on health systems. Nonetheless, any service downtime, particularly in the case of emergency services, can lead to adverse outcomes and in the worst case, loss of life. In this paper, we use an interdisciplinary approach that combines stochastic models with surrogate-assisted optimisation algorithms to maximise e-health system availability considering the budget to acquire redundant components as a constraint, comparing three nature-inspired meta-heuristic optimisation algorithms.
Keywords: internet of medical things; availability; surrogate models; nature-inspired approaches.
Recommendation system based on space-time user similarity
by Wei Luo, Zhihao Peng, Ansheng Deng
Abstract: With the advent of 5G, the way people get information and the means of information transmission have become more and more important. As the main platform of information transmission, social media not only brings convenience to people's lives, but also generates huge amounts of redundant information because of the speed of information updating. In order to meet the personalised needs of users and enable users to find interesting information in a large volume of data, recommendation systems emerged as the times require. Recommendation systems, as an important tool to help users to filter internet information, play an extremely important role in both academia and industry. The traditional recommendation system assumes that all users are independent. In this paper, in order to improve the prediction accuracy, a recommendation system based on space-time user similarity is proposed. The experimental results on Sina Weibo dataset show that, compared with the traditional collaborative filtering recommendation system based on user similarity, the proposed method has better performance in precision, recall and F-measure evaluation value.
Keywords: time-based user similarity; space-based user similarity; recommendation system; user preference; collaborative filtering.
A graphical front-end interface for React.js considering state-transition diagrams
by Shotaro Naiki, Masaki Kohana, Michitoshi Niibori, Shusuke Okamoto, Yasuhiro Ohtaki, Masaru Kamada
An integrity control model for mass data transmission under big data cloud storage
by Zhengguo Zhu
Abstract: In order to improve the control level of massive data transmission integrity and improve the space utilisation efficiency and data transmission integrity under cloud storage, an integrity control model of mass data transmission under big data cloud storage is constructed. Firstly, the basic principle of big data cloud storage is studied. Secondly, the mass data feature extraction model is established. Thirdly, the massive data integrity control algorithm is designed based on ant algorithm. Finally, simulation analysis is carried out, and the effectiveness of the proposed method is verified.
Keywords: integrity control; mass data transmission; big data cloud storage.
Securing utility computing using enhanced elliptic curve cryptography and attribute-based access control policy
by Saira Varghese, S. Maria Celestin Vigila
Abstract: This paper proposes a data security model for utility computing that integrates securely mapped plain text using Elliptic Curve Cryptography (ECC) without certified keys, attribute-based access control using Reduced Ordered Binary Decision Diagram (ROBDD), cryptographically secure 256-bit pseudorandom numbers and fingerprint security. ECC uses standard elliptic curves of large prime and ROBDD is built with positive and negative attributes of data users and 256-bit pseudorandom secrets along its path. The secret key that is generated is based on the attributes along the valid path of the policy to enhance the security. The proposed model allows secure key exchange by making use of the property of the elliptic curve to convert numbers into points and secure data storage by associating authenticated secrets of end-users with the original secret key, which is beneficial in decentralised architecture. The result reveals that proposed model achieves high security and less space and time complexity for securing cloud data.
Keywords: secure utility computing; ROBDD; reduced ordered binary decision diagram; ECC; elliptic curve cryptography; attribute based access control; cryptographically secure pseudorandom numbers.
Design and analysis of novel hybrid load-balancing algorithm for cloud data centres
by Ajay Dubey, Vimal Mishra
Abstract: In recent the pandemic scenario there is a paradigm shift, from traditional computing to internet-based computing. Now is the time to store and compute the data in the cloud environment. The Cloud Service Providers (CSPs) establish and maintain a huge shared pool of computing resources that provide scalable and on-demand services around the clock without geographical restrictions. The cloud customers are able to access the services and pay according to the accession of resources. When millions of users across the globe connect to the cloud for their storage and computational needs, there might be issues such as delay in services. This problem is associated with load balancing in cloud computing. Hence, there is a need to develop effective load-balancing algorithms. The Novel Hybrid Load Balancing (NHLB) algorithm proposed in this paper manages the load of the virtual machine in the data centre. This paper is focused on certain problems such as optimisation of performance, maximum throughput, minimisation of makespan, and efficient resource use in load balancing. The NHLB algorithm is more efficient than conventional load-balancing algorithms with reduced completion time (makespan) and response time. This algorithm equally distributes the tasks among the virtual machines on the basis of the current state of the virtual machines and the task time required. The paper compares the result of proposed NHLB algorithm with dynamic load-balancing algorithm and honeybee algorithm. The result shows that the proposed algorithm is better than the dynamic and honeybee algorithms.
Keywords: cloud computing; data centre; load balancing; virtual machine; makespan; performance optimisation.
Virtual traditional craft simulation system in mixed reality environment
by Rihito Fuchigami, Tomoyuki Ishida
Abstract: In a previous study, we implemented a high presence virtual traditional craft system using a head-mounted display (HMD) and a mobile traditional craft presentation system using augmented reality (AR). The high presence virtual traditional craft system realised a simulation experience of traditional crafts in virtual space comprising different cultural styles. However, this system had to construct the different cultural architectures in advance, and the amount of work was enormous. The mobile traditional craft presentation system realised an AR application that allows users to easily arrange 3DCG (Three-Dimensional Computer Graphics) of traditional crafts in real space using mobile terminals. However, this system lacks a sense of presence because users experience traditional crafts on the mobile terminals flat display. Therefore, in this study, we focused on mixed reality (MR) technology and developed an MR virtual traditional craft simulation system using an HMD. With MR technology, we have overcome the work cost and low presence issues relative to the construction of a virtual reality space. As a result, our system provides a simulation experience that realises a high sense of presence and intuitive operation. We conducted a comparative evaluation experiment with 30 subjects to evaluate the constructed MR virtual traditional craft simulation system. We obtained high evaluations of the systems presence and applicability; however, several operability issues were identified.
Keywords: mixed reality; augmented reality; interior simulation; Japanese traditional crafts; immersive system.
The role of smartphone-based social media capabilities in building social capital, trust, and credibility to engage consumers in eWOM: a social presence theory perspective
by Saqib Mahmood, Ahmad Jusoh, Khalil MD Nor
Abstract: Smartphone-based social media has become a well-established channel for users to develop and maintain intimate social relations that enable them to engage in brands-related information exchange, regardless of their time and location, such as eWOM. Nevertheless, little is known about the essential elements of smartphone-based social media that help consumers to develop intimate social relationships and engage them in eWOM. To this end, drawing on the theory of social presence, the current study develops a research model that proposes that interactivity and media richness enhance social presence, giving consumers a sense of psychological proximity. Subsequently, it leads to the development of trust and social capital bonding and bridging. As a result of the bridging and bonding of social capital, consumers' perceived credibility is expected to enable them to engage in eWOM. To empirically investigate the theoretical model, a survey of 407 smartphone-based social media users was conducted in Pakistan. Empirical results reveal that the interactivity and media richness enhance the social presence that proffers consumers' psychological proximity to developing trust and social capital, further enhancing their perceived credibility to engage in eWOM. Discussions, implications, and future directions on the results are described in the final section of the study.
Keywords: interactivity; media richness; social presence; psychological proximity; social capital; eWOM; trust; smartphone; social media.
A cloud-based system for distance learning supported by fog-cloud cooperation
by Lucas Larcher, Victor Ströele, Mário Dantas
Abstract: New IT approaches are enhancing the cloud computing paradigm when making this model more pervasive and on-demand to share computing resources on the web. These resources can be used readily by anyone anywhere. This fact fosters distance learning modality where students do not need to be physically present on the course place, and the teaching and learning process occurs by a Virtual Learning Environment (VLE). A centralised information unit (a university, for example) provides support to outside units with the necessary equipment and knowledge. In this ordinary structure, the cloud paradigm is mighty and primarily used, but it has some limits. The challenge is to turn this paradigm scalable, considering that the users' infrastructure is mainly characterised by poor internet connectivity. A fog-cloud model can be conceived in these cases to tackle this problem using a local pre-processing data stage before sending them to the cloud. The present research study reveals a situation that occurs in Brazilian universities that offer distance education courses. This paper proposes a low-cost solution using a fog-cloud computing paradigm as an environment to support a VLE. In the evaluation process, an experiment using real data was carried out, and the results obtained successfully point to the viability of the proposal.
Keywords: fog-cloud cooperation; distance learning; virtual learning environment; application performance; fog computing.
Novel crowd synthesis approach and GLCM-derived descriptor for crowd behaviour detection
by Yu Hao, Ying Liu, Jiulun Fan, Yuquan Gan
Abstract: Unlike an individuals behaviour, the dynamic nature of the crowd makes its behaviour difficult to define and classify. Therefore, techniques and approaches for crowd behaviour analysis usually encounter challenges different from an individuals behaviour. This paper aims to tackle several key issues in the procedure of crowd behaviour analysis. Firstly, a novel taxonomy is proposed to provide criteria for the explicit definition of different crowd behaviours. By adapting personal space, relative velocity, and group force into conventional social force, a crowd behaviour synthesis approach is devised to provide visually realistic data for model training. Secondly, this paper introduces an effective entropy-based motion texture extraction algorithm, in order to accurately obtain the spatio-temporal motion information from the crowd. Furthermore, this paper proposes a novel visual descriptor based on GLCM derived patterns to describe the visual essence of crowd anomalies. Experiment results indicate the proposed descriptor outperforms other mainstream patterns on the detection of panic dispersing and congestion.
Keywords: crowd behaviour; feature extraction; image classification; image texture analysis; information entropy.
Decentralised priority-based shortest job first queue model for IoT gateways in fog computing
by Jayashree Narayana Swamy, B. Satish Babu, Basavraj Talwar
Abstract: An increased growth in time-critical IoT applications has led to a rise in real time resource requirements. The stringent deadlines on latency have caused IoT applications to move out from far-away cloud servers to distributed fog computing devices infrastructure, which is available locally. In order to meet the touchstones of deadlines and processing times, there is a need to prioritise the job scheduling through the IoT gateways to appropriate fog devices. Studies showed that the queuing models exhibit uncertainties in choosing suitable computing devices, applying priorities to the jobs, deadline achievements, and minimum latency constraints. In this paper, we propose a decentralised priority-based shortest job first queuing model for the IoT gateways for a fog computing infrastructure, which uses the priority-based jobs sorting technique to achieve a better performance and also to overcome most of the uncertainties in queuing.
Keywords: queuing model; IoT; fog computing; M/M/c queues; M/M/c/1 queues.
Special Issue on: IoT Integration in Next-Generation Smart City Planning
Internet of things based architecture for additive manufacturing interface
by Swee King Phang, Norhijazi Ahmad, Chockalingam Vaithilingam Aravind, Xudong Chen
Abstract: This paper addresses the current challenges in managing multiple additive manufacturing units (i.e., 3D printers) without an online system. Managing multiple 3D printers is troublesome and difficult. The traditional process of selecting free printers and monitoring printing statuses manually has revealed a big flaw in the system as it requires physical interaction between the machine and human. As of today, there is little to none for a 3D printer online managing system. Most printing still requires human monitorisation and the project work to be printed must be fed physically to the printer via external drives. In this paper, a solution to zero physical interaction to additive manufacturing units is proposed. The objective is achieved by using the saturated IoT technologies. Webserver will be used to create a webpage to upload the file, for approval, and to check the printing status. A server will be used to store the files, slicing software, file queueing system and to store temporary information of the manufacturing unit's status. Cameras on multiple 3D printers will be used as sensors to monitor the project progress visually. In the end product of the IoT based 3D printing systems, the user will be able to upload the files, ask for superior approval (optional), queue to a specific manufacturing unit to print out by the algorithm set on the cloud server, receive important data from the server such as time estimation, progress percentage and the extruders temperature, and receive notification of error if any issues arise, and notification of completion. The proposed system is implemented and verified in the Additive Manufacturing Lab in Taylors University Malaysia.
Keywords: additive manufacturing units; 3D printing; online printing; printer management; cloud printing; printing networking; IoT printer; printing monitoring; heat monitor.
Enhanced authentication and access control in internet of things: a potential blockchain-based method
by Syeda Mariam Muzammal, Raja Kumar Murugesan
Abstract: With the rapid growth of Internet of Things (IoT), it can be foreseen that IoT services will be influencing several use-cases. IoT brings along the security and privacy issues that may hinder its widescale adoption. The scenarios in IoT applications are quite dynamic compared with the traditional internet. It is vital that only authenticated and authorised users get access to the services provided. Hence, there is a need for a novel authentication and access control technique that is compatible and practically applicable in diverse IoT scenarios to provide adequate security to devices and data communicated. This article aims to highlight the potential of blockchain for enhanced and secured authentication and access control in IoT. The proposed method relies on blockchain technology, which tends to eliminate the limitations of intermediaries for authentication and access control. Compared with existing systems, it has advantages of decentralisation, secured authentication, authorisation, adaptability, and scalability.
Keywords: internet of things; security; authentication; access control; blockchain.
Control and monitoring of air-conditioning units through cloud storage and control operations
by Chockalingam Aravind Vaithilingam, Mohsen Majrani
Abstract: Temperature control and monitoring of the air conditioning units is critically important towards energy savings. The purpose of this work is to design and develop an air conditioner monitoring system for monitoring and control using internet of things. The developed system uses an integrated mobile app using a cloud service that enables users to monitor and control its operations. The system consists of three subsystems, which are micro-controller, cloud storage and mobile app. The micro-controller can collect data from pressure transducer, differential pressure sensor, current transformer, accelerometer, and temperature and humidity sensor. The data collected by Arduino is sent to the cloud storage platform by using REST API. Cloud storage can store the data, display the data graphically, and send notification to specific users when a rule is activated. A hybrid mobile app is also developed with Ionic Framework. The mobile app can display the data stored in cloud storage. The data is fetched from the cloud storage by using the REST API. The system developed is able to monitor several critical parameters from the air conditioner, which are differential air pressure, refrigerant pressure, power, and angle of vibration on the x-axis, angle of vibration on the y-axis, temperature and humidity. With the data collected an algorithm to monitor and control the performance of such an air conditioning system through this embedded module is envisioned to be part of the energy-efficient systems.
Keywords: condition monitoring; internet of things; mobile app; cloud storage.
Forecasting of solar potential and investigation of voltage stability margin using FACTs device: a synopsis from Geography of Things perspective
by Masum Howlader, Khandaker Sultan Mahmood, Md.Golam Zakaria, Kazi Mahtab Kadir, Mirza Mursalin Iqbal
Abstract: The uncertain and erratic nature of renewable energy in solar form is quite distinctive from traditional and dispatchable fuels for generation and is laborious to integrate into conventional system operation. In the first part of this work, a machine-learning algorithm is used to train models on solar irradiance data and different meteorological weather information to predict solar irradiance for different cities to validate the forecasting model. The above-mentioned data for modelling purposes is taken from publicly available Geographical Information System (GIS) data. This can be realistically collected using Internet of Things (IoT) devices and sensors which, if based on a GIS approach, transforms the system into Geography of Things (GoT). Again, the intermittent and inertia-less nature of photovoltaic systems can produce significant power oscillations that can cause significant problems with the dynamic stability of the power system and also limit the penetration capacity of photovoltaics into the grid. In the second part, it is shown that the residue-based power oscillation damping (POD) controller significantly improves the inter-area oscillation damping. The validity and effectiveness of the proposed controller is demonstrated on a three-machine two-area test system that combines conventional synchronous generators and Flexible AC Transmission Systems (FACTs) devices using simulations. This report overall puts an in-depth analysis with regard to the challenges of solar resources with the integration, planning, operation and particularly the stability of the rest of the power grid, including existing generation resources, customer requirements and the transmission system itself that will lead to improved decision making in resource allocations and grid stability.
Keywords: solar forecasting; static var compensator; support vector machine; power oscillation damping; geography of things.
An analytical approach to real-time cloud services on IoT-based applications for smart city planning
by M.D. Shahrukh Adnan Khan, Kazi Mahtab Kadir, Md. Khairul Alam, Shoaib Mahmud, Shah Reza Mohammad Fahad Ul Hossain, Md. Pabel Sikder, Fiza Jefreen, Ainun Kamal
Abstract: This paper illustrates the cloud-based services on next generation smart living technology, first by providing the concept of smart technology of the next generation, followed by the wide area of application in this sector. Next, the current generation of technological enhancement, IoT (Internet of Things) is brought into the smart living. The entire IoT-based smart application sector has then been divided into five categories: power and energy sector, transport sector, healthcare sector, retail sector, and education sector. Each sector has been analysed in detail with respect to possible real-time cloud services, which can be incorporated into the respective area. Consequently, existing cloud services, current trends, limitations, and future scopes have been discussed, followed by recommendations in each section. For example, for IoT-based application in the power and energy sector, the limitations of cloud service have been found, such as unoptimised communication scheme, data complexity, interoperability, cyber security risk and data integrity issues. To address these limitations respectively, a policy to ensure backward and horizontal compatibility, a proposal to increase local processing, data compression and prediction, implementation of big data techniques, authentication and encryption, planning and redundancy have been included as recommendations. For the other four categories, an intense analysis has been carried out in a similar fashion. Finally, the recommendations have been added for each category for the next barrier scope of research.
Keywords: cloud service; cloud storage; smart city; IoT application; real-time system.
Special Issue on: CONIITI 2019 Intelligent Software and Technological Convergence
A computer-based decision support system for knowledge management in the swine industry
by Johanna Trujillo-Díaz, Milton M. Herrera, Flor Nancy Díaz-Piraquive
Abstract: The swine industry contributes to food security around the world. However, the most vulnerable point in the industry occurs at the pig production cycle. This production cycle generates an imbalance between supply and demand, which affects profitability. This paper describes a computer-based decision support system for knowledge management (KM) which contributes to improving the profitability performance management into the swine industry. The computer-based system allows assessing decision alternatives on the dimensions of the KM capacity and profitability performance. This tool contributes to generating integration strategies for the swine industry four simulation scenarios was designed for representing a pig company in the Colombian case.
Keywords: decision support system; simulation; swine; knowledge management; system dynamics.
Computational intelligence system applied to plastic microparts manufacturing process
by Andrés Felipe Rojas Rojas, Miryam Liliana Chaves Acero, Antonio Vizan Idoipe
Abstract: In the search for knowledge and technological development, there has been an increase in new analysis and processing techniques closer to human reasoning. With the growth of computational systems, hardware production needs have also increased. Parts with millimetric to micrometric characteristics are required for optimal system performance, so the demand for injection moulding is also increasing. Injection moulding process in a complex manufacturing process because mathematical modelling is not yet established: therefore, to address the selection of correct values of injection variables, computational intelligence can be the solution. This article presents the development of a computational intelligence system integrating fuzzy logic and neural network techniques with CAE modelling system to support injection machine operators, in the selection of optimal machine process parameters to produce good quality microparts using fewer processes. The tests carried out with this computational intelligent system have shown a 30% improvement in the efficiency of the injection process cycles.
Keywords: computational intelligence; neural networks; fuzzy logic; micro-parts; plastic parts; computer vision; expert systems; injection processes; CAD; computer-aided design systems; CAE; computer-aided engineering.
Special Issue on: Novel Hybrid Artificial Intelligence for Intelligent Cloud Systems
QoS-driven hybrid task scheduling algorithm in a cloud computing environment
by Sirisha Potluri, Sachi Mohanty, Sarita Mohanty
Abstract: Cloud computing environment is a growing technology of distributed computing. Typically using cloud computing the services are deployed with individuals or organisations and to allow sharing of resources, services, and information based on the demand of users over the internet. CloudSim is a simulator tool used to simulate cloud scenarios. A QoS-driven hybrid task scheduling architecture and algorithm for dependent and independent tasks in a cloud computing environment is proposed in this paper. The results are compared against the Min-Min task scheduling algorithm, QoS-driven independent task scheduling algorithm, and QoS-driven hybrid task scheduling algorithm. QoS-driven hybrid task scheduling algorithm is compared with time and cost as QoS parameters and it gives a better result for these parameters.
Keywords: cloud computing; task scheduling; quality of service.
SD-6LN: improved existing internet of things framework by incorporating software defined network approach
by Rohit Das, Arnab Maji, Goutam Saha
Abstract: The Internet of Things (IoT) is a prominent technology in today's world, where its real-time applications are being used in various areas. In spite of the fact that the advances made by IoT are remarkable, the current IoT foundation experiences problems such as availability, reliability, scalability, resiliency, interoperability, security and privacy. Software-Defined Network (SDN) is an approach that can help in resolving many of these IoT limitations. SDN uses a controller that can improve upon the system performance. In this paper, a novel SD-6LN architecture is proposed for existing IPv6 over Low Power Wireless Personal Area Network (6LoWPAN) based IoT infrastructure. The SD-6LN architecture incorporates the network layer of IoT and SDN to address some of the challenges of traditional resource constraint IoT network systems, such as availability, reliability, scalability and resiliency. The experimentation was carried out in simulation. The experimental results indicated improved performance with respect to round trip time, jitter, packet drop, latency and throughput.
Keywords: Architecture; Internet of Things; OpenFlow; Software-Defined Network; 6LoWPAN.
A security analysis of lightweight consensus algorithm for wearable kidney
by Saurabh Jain, Adarsh Kumar
Abstract: Blockchain is a distributed ledger-based technology and provides a solution to many data-centric problems. In recent times, this area has encouraged innovations to handle challenges in many useful applications in which traditional approaches are not found to be successful. A smart healthcare system is one such application where it has been observed that blockchain can play a vital role in terms of combining technologies such as security, data storage, data retrieval, patient-centric approach, and data visualisation. This work proposes a game theory-based approach for consensus building in a distributed network. This approach builds consensus in a trustworthy environment where technologies are explored to provide a problem-centric solution. In this work, the wearable kidney model is analysed to understand the working of the game-theory-based consensus model. This example shows how blockchain technology can be used for consensus building in the healthcare system. The lightweight consensus algorithm consumes fewer resources (suitable for resource constraint devices) and provides an efficient solution to simulate the functionality of a wearable kidney model. The comparative analysis of the result shows that the proposed approach is efficient in fast bit-matching and quick consensus establishment. Results show that kidney blood and urine production are mapped to almost ideal conditions and variations in delay for bit-matching, and algorithm executions are evaluated thereafter. The comparative analysis of the algorithms shows that algorithm 1 outperforms (at least 2.1%) algorithm 2 in delay analysis because of less distributed functionality. Both algorithms are found to be efficient compared with state-of-the-art algorithms for trust establishment.
Keywords: game theory; blockchain; cryptocurrency; lightweightness; hash rate; bit-exchange; challenge-response; attacks.
Special Issue on: ICIMMI 2019 Emerging Trends in Multimedia Processing and Analytics
Handwritten Odia numeral recognition using combined CNN-RNN
by Abhishek Das, Mihir Narayan Mohanty
Abstract: Detection and recognition is a major task for current research. Almost all the parts of signal processing, including speech and images has the sub-content of it. Data compression mainly uses in multimedia communication, where the recognition is the major challenge. Keeping all these facts in view, the authors have taken an approach for handwritten numbers recognition. To meet the challenge of fake data, a generative adversarial network is used to generate some data and is considered along with original data. The database is collected from IIT, Bhubaneswar, and used in a GAN model to generate a huge amount of data. Further, a Convolutional Neural Network (CNN) and a Recurrent Neural Network (RNN) are considered for recognition purpose. Though Odia numerals are a little complex, the recognition task was found very interesting. A little work has been done in this direction. However, the application of a deep learning based approach is absent. Long Short Term Memory (LSTM) cells are used as recurrent units in this approach. We have added 1000 images generated through Deep Convolutional Generative Adversarial Network (DCGAN) to the IIT-BBSR dataset. In this method we have used the Adam optimisation algorithm for minimising the error, and to train the network we have used the supervised learning method. The result of this method gives 98.32% accuracy.
Keywords: character recognition; Odia numerals; deep learning; CNN; RNN; LSTM; DCGAN; Adam optimisation.
An optimal channel state information feedback design for improving the spectral efficiency of device-to-device communication
by Prabakar Dakshinamoorthy, Saminadan Vaitilingam
Abstract: This article introduces a regularised zero-forcing (RZF) based channel state information (CSI) feedback design for improving the spectral efficiency of device-to-device (D2D) communication. This proposed method exploits conventional feedback design along with the optimised CSI in regulating the communication flows in the communicating environment. The codebook-dependent precoder design improves the rate of feedback by streamlining time/frequency dependent scheduling. The incoming communication traffic is scheduled across the available channels by pre-estimating their adaptability and capacity across the underlying network. This helps to exchange partial channel information between the communicating devices without the help of base station services. These features reduce the transmission error rates to achieve better sum rate irrespective of the distance and transmit power of the devices.
Keywords: CSI; D2D; feedback design; precoding; zero-forcing.
Special Issue on: ITT 2019 Advances in Next-Generation Communications and Networked Applications
EPPR: blockchain for educational record sharing and recommendation using the Ethereum platform
by Akram Alkouz, Ahmed HajYasien, Abdulsalam Alarabeyyat, Khalid Samara, Mohammed Al-Saleh
Abstract: There has been a marked recent increase in the challenges of privacy, data interoperability and quality of Educational Professional Personal Record (EPPR). This calls into question the current model, in which different parties generate, exchange and monitor massive amounts of personal data related to EPPR. Ethereum blockchain has demonstrated that trusted, auditable transactions is visible using a decentralised network of nodes accompanied by a general ledger. Thus, due to the rapid development of educational and professional data generators such as online universities and distance learning requires learners to engage in detail into their EPPR as well as the educational and professional data generators. In this paper, we propose a novel decentralised approach to manage EPPR using Ethereum blockchain technology. The decentralised approach provides the owner of the EPPR a comprehensive immutable log and ease of access to their educational records across the educational record editors and consumers. In addition, it provides a recommender engine to endorse skills and competencies to education record owners and similar candidates for educational records editors and consumers. The use of Ethereum blockchain features can provide solutions in terms of exchanging of data among parties ensuring privacy, accountability and data interoperability. The aim of this approach is to also facilitate educational stakeholders (universities and employing agencies) to participate in the network as blockchain miners rewarded by pseudonymised data in compliance with General Data Protection Rules (GDPR) in United Arab Emirates (UAE).
Keywords: educational professional personal record; blockchain; Ethereum; GDPR; recommender engine; semantic graph.
Comparing the performance of supervised machine learning algorithms when used with a manual feature selection process to detect Zeus malware
by Mohamed Ali Kazi, Steve Woodhead, Diane Gan
Abstract: The Zeus banking malware is one of the most prolific banking malware variants ever to be discovered, and this paper compares and analyses the performance of several supervised machine learning (ML) algorithms when used to detect the Zeus banking malware (Zeus). The key to this paper is that the features that are used for the analysis and detection of Zeus are manually selected, providing the researcher better control over which features can and should be selected. This also helps the researcher to understand the features and the impact that the various feature combinations have on the accuracy of the algorithms when used to detect Zeus. The empirical analysis showed that the decision tree and random forest algorithms produced the best results as they detected all the Zeus samples. The empirical analysis also showed that selecting the feature combinations manually produces varying results, allowing the researchers to understand how the features impact the detection accuracy.
Keywords: Zeus banking malware; machine learning; binary classification algorithms; supervised machine learning; manual feature selection.
Collaborative ambient intelligence based demand variation prediction model
by Munir Naveed, Yasir Javed, Muhammed Adnan, Israr Ahmed
Abstract: Inventory control problem is faced by corporations on a daily basis to optimise the supply chain process and for predicting the optimal pricing for the item sales or for providing services. The problem is heavily dependent on a key factor, i.e. demand variations. Inventories must be aligned according to demand variations to avoid overheads or shortages. This work focuses on exploring various machine learning algorithms to solve demand variation problems in real time. Prediction of demand variations is a complex and non-trivial problem, particularly in the presence of open order. In this work, prediction of demand variation is addressed with the use-cases which are characterised with open orders. This work also presents a novel prediction model that is a hybrid of learning domains as well as domain-specific parameters. It exploits the use of Internet of Things (IoT) to extract domain-specific knowledge, while a reinforcement learning technique is used for predicting the variations in these domain-specific parameters, which depend on demand variations. The new model is explored and compared with state-of-the-art machine learning algorithms using Grupo Bimbo case study. The results show that the new model predicts the demand variations with significantly higher accuracy than other models.
Keywords: inventory management; reinforcement learning; IoT devices; Grupo Bimbo inventory demand variation.
Special Issue on: WETICE-2019 Novel Approaches to the Management and Protection of Emerging Distributed Computing Systems
Benchmarking management techniques for massive IIoT time series in a fog architecture
by Sergio Di Martino, Adriano Peron, Alberto Riccabone, Vincenzo Norman Vitale
Abstract: Within the Industrial Internet of Things (IIoT) scenario, the online availability of a growing number of assets in factories is enabling the collection of huge amounts of data. They can be used for big data analytics, with great possibilities for efficiency improvements and business growth. Each asset produces collections of time series, namely data streams, that must be properly handled with specific techniques providing, at the same time, effective ingestion and retrieval performance, in complex network architectures, maintaining compliance with company and infrastructure boundaries. In this paper, we describe an industrial experience in the management of massive time series from instrumented machinery, conducted in a plant of Avio Aero (part of General Electric Aviation). As a first step, we propose a fog-based architecture to ease the collection of these massive dataset, supporting local and remote data analytics tasks. Then, we present the results of an empirical comparison of four database management systems, namely PostgreSQL, Cassandra, MongoDB and InfluxDB, in the ingestion and retrieval of gigabytes of real IIoT data, collected from an instrumented dressing machine. More in detail, we tested different settings and indexing features offered by these DBMS, under different types of query. Results show that, in the investigated context, InfluxDB is able to provide very good performance, but PostgreSQL can still be a very interesting alternative, when more advanced settings are exploited. MongoDB and Cassandra, on the other hand, are not able to provide the performance of the two other DBMS.
Keywords: big data; time series; IIoT; fog architecture; TSMS; NoSQL Ddtabase; relational database; benchmarking.
DIOXIN: runtime security policy enforcement of Fog Computing applications
by Enrico Russo, Luca Verderame, Alessandro Armando, Alessio Merlo
Abstract: Fog Computing is an emerging distributed computational paradigm that moves the computation towards the edge (i.e., where data are produced). Although Fog operating systems provide basic security mechanisms, security controls over the behaviour of applications running on Fog nodes are limited. For this reason, applications are prone to a variety of attacks. We show how current Fog operating systems (with a specific focus on Cisco IOx) are actually unable to prevent these attacks. We propose a runtime policy enforcement mechanism that allows for the specification and enforcement of user-defined security policies on the communication channels adopted by interacting Fog applications. We prove that the proposed technique reduces the attack surface of Fog Computing w.r.t. malicious applications. We demonstrate the effectiveness of the proposed technique by carrying out an experimental evaluation against a realistic Fog-based IoT scenario for smart irrigation.
Keywords: Fog Computing; security assessment; Cisco IOx; runtime monitoring.
Black-box load testing to support autoscaling web applications in the cloud
by Marta Catillo, Luciano Ocone, Massimiliano Rak, Umberto Villano
Abstract: One of the most interesting features of cloud environments is the possibility to deploy scalable applications, which can automatically modulate the amount of leased resources so as to adapt to load variations and to guarantee the desired level of quality of service. As autoscaling has severe implications on execution costs, making optimal choices is of paramount importance. This paper presents a method based on off-line black-box load testing that allows to obtain performance indexes of a web application in multiple configurations under realistic load. These indexes, along with available resource cost information, can be exploited by autoscaler tools to implement the desired scaling policy, making a trade-off between cost and user-perceived performance.
Keywords: autoscaling; cloud computing; load testing.
LISA: a lean information service architecture for SLA management in multi-cloud environments
by Nicola Sfondrini, Gianmario Motta
Abstract: Cloud computing emerged as a disruptive technology for managing IT services over the Internet, evolving from grid computing, utility computing and Software-as-a-Service (SaaS). After an initial scepticism, international companies are widely migrating their IT workload on private and public clouds to optimise geographical coverage and launch new digital services. Currently, hybrid and multi-cloud environments introduce additional complexity in managing the Quality of Service (QoS), therefore requiring more sophisticated Service Level Agreements (SLAs). To deal with these issues, our research developed a SLA-aware Lean Information Service Architecture (LISA) for managing multi-cloud environments that support users in the whole service lifecycle. LISA's performance was tested in the Innovation Lab of a global telco operator, by deploying services on private cloud and public cloud providers. Experimental results prove not only LISAs effectiveness but also its efficiency in various aspects, such as preventing SLA violations and service performance degradations, optimising the QoS, and controlling the service components deployed across multiple public cloud providers.
Keywords: multi-cloud; SLA; service level management; QoS; cloud broker; cloud SLM framework; SLA-aware resource allocation.
Evaluation of innovative solutions for e-mobility
by Salvatore Venticinque, Rocco Aversa, Beniamino Di Martino, Shanshan Jiang, Marit Natvig, Regina Enrich Sard
Abstract: This paper presents the methodology designed to evaluate the effect of innovation and the stakeholder acceptance about the innovative solutions by the GreenCharge project to advance electric mobility. The presented methodology is based on the CIVITAS evaluation framework, which is introduced in the paper, and it is specialised according to the GreenCharge requirements. The measures to be evaluated have been put in place by the GreenCharge project in three different pilots. In particular, this paper provides an extensive list of Key Performance Indicators (KPIs) and which of them will be evaluated in each pilot to provide a quantitative estimation of impact of technology innovation. Some KPIs belong to the CIVITAS evaluation framework, new ones have been defined to evaluate e-mobility innovation. Three evaluation methodologies will be used to estimate KPIs: evaluation based on automatic computation from data collected in pilots, evaluation based on simulation, and evaluation based on analysis of surveys and interviews delivered to involved stakeholders and volunteers. The paper describes how to deliver and how to present evaluation results. The detailed schedule of data collection and evaluation activities has been planned in collaboration with implementation activities. Finally, a preliminary requirement analysis of simulation and a dashboard design for KPI presentation are discussed.
Keywords: evaluation; electric mobility; key performance indicators.
Special Issue on: 3PGCIC Cloud and Edge Systems and Applications
Quality of service prediction model in cloud computing using adaptive dynamic programming parameter tuner
by Monika Rd, Om Prakash Sangwan
Abstract: With the continuous proliferation of cloud services, the recommendation of optimal cloud service according to user requirement has become an important and critical issue and makes it highly infeasible for a single user, who wants to use the cloud services for some specific application with QoS requirements to try all the cloud services and thus depends on the information collected by other users about the QoS of various cloud services. These collected QoS values are highly nonlinear, complex and uncertain. To deal with the given scenario, there is a specific requirement to develop a recommender system for the prediction of unknown QoS values using some optimization techniques. In this paper, we have developed two models: i) optimised matrix factorisation prediction model ii) optimised fuzzy C-means prediction model. matrix factorisation and fuzzy C-means are some basic traditional techniques used with static model parameters for the prediction of missing values. But these techniques with static parameters are not able to handle the significant changes under the unpredictable internet conditions and sparsity of available historical QoS data. To overcome this problem, we have implied a novel backpropagation based ADP parameter tuning strategy to these two basic prediction techniques where backpropagation is an important mathematical tool of neural network. This is the first time it has been applied with ADP parameter tuner, to the best of our knowledge, for developing the self-adaptive intelligent system and this system provides an automatic parameter tuning capability to our proposed QoS prediction models. To evaluate the proposed approach, we have done a simulation of the approach on a real QOS dataset and experimental results show that our proposed approach yields better prediction accuracy when compared with other traditional approaches.
Keywords: cloud computing; QoS prediction; ADP parameter tuner; fuzzy C-means clustering; matrix factorisation; backpropagation neural network.
Towards a cloud model choice evaluation: comparison between cost/features and ontology-based analysis
by Pasquale Cantiello, Beniamino Di Martino, Michele Mastroianni, Luigi Colucci Cante, Mariangela Graziano
Abstract: In academic institutions, there is frequently the need to provide new services, in a cloud model, to be used in either teaching or research activities. One of the main decisions to be addressed is related to which cloud model to adopt (private, public or hybrid), and which mixing of functionalities to use for the hybrid one. In this paper, two different methodologies (cost/features and semantic-based) are tested in order to identify the best suited cloud model to adopt for a specific problem. The long-term perspective is to build a methodology to serve as a tool to be used as decision support for the ICT manager in order to assist in this decision. The comparison between the two different methodologies may show the strengths and weaknesses of both approaches.
Keywords: cloud model; decision support system; SWRL; OWL; cloud evaluation; cloud cost analysis.
An evaluation environment for high-performance computing combining supercomputing and cloud
by Yusuke Gotoh, Toshihiro Kotani
Abstract: Based on the characteristics of the supercomputer and the cloud system installed in the Hokkaido University Information Initiative Center, Japan, we aim to construct a high-performance computing environment by linking the two types of system. In this paper, we propose a high-performance computing environment for deep reinforcement learning that links the supercomputer and cloud systems. Our proposed system can construct a high-performance computing environment based on the scale of the computing process by the cooperation of the supercomputing and cloud systems with short physical distance and short network distance. In our evaluation of deep reinforcement learning using our proposed system, we confirmed that computer resources can be effectively used by allocating suitable processing for the supercomputer and the cloud according to the usage situations of the CPU, the GPU, and the memory.
Keywords: cloud service; high-performance computing; processing time; supercomputer.
Special Issue on: Intelligent Evaluations of Resource Management Techniques in Fog Computing
Web data mining algorithm based on cloud computing environment
by Yunpeng Liu, Xiaolong Gu, Jie Zhang
Abstract: With the rapid development of the internet, the daily growth of information has developed exponentially. To analyse useful information from it, there is already a bottleneck in the calculation and storage of a single node. In order to quickly extract valuable rules and patterns from massive and noise-containing data and make them easy to understand and apply directly, we used data mining technology. On the other hand, based on the characteristics of low cloud computing cost, large throughput, good fault tolerance and strong stability, the cloud computing method is selected for web data mining processing. This paper studies and analyses the K-means clustering algorithm, and the web data mining algorithm based on cloud computing environment improves the K-means algorithm, overcomes the shortcomings of the K-means algorithm itself, and builds a good cloud computing environment on the Hadoop platform, and parallelises and optimses the improved algorithm. We will focus on the K-means clustering algorithm. In order to solve the shortcomings of the K-means algorithm itself, we will consider improving the K-means algorithm and transplanting it to the Hadoop cloud computing platform. Finally, the experimental results in terms of effectiveness and acceleration ratio show that the improved and optimised algorithm solves the problem of insufficient speed and efficiency in the clustering process.
Keywords: cloud computing; data mining; clustering algorithm; K-means algorithm.
Intelligent manufacturing system based on data mining algorithm
by Xiaoya Liu, Qiongjie Zhou
Abstract: How to reasonably apply data mining methods to intelligent manufacturing systems is a major issue facing the current manufacturing industry. This article focuses on the evaluation model of an intelligent manufacturing system based on a data mining algorithm. Combining the data mining algorithm with the intelligent manufacturing system, the evaluation model of the intelligent manufacturing system is established successfully. A neural network is selected for the final evaluation. After training, perform error analysis, the problems that occur in optimisation algorithms, feature selection, or data collection are analysed. The highest accuracy rate of the training group was 69%, and the highest accuracy rate of the test group was 32.5%. The results show that using data mining algorithms for recognition can effectively cluster control chart patterns and improve recognition efficiency.
Keywords: data mining algorithm; intelligent manufacturing system; evaluation model; error analysis.
Visualisation technology in digital intelligent warehouse management System
by GuangHai Tang, Hui Zeng
Abstract: This study introduces visualisation technology into digital intelligent warehouse management and combines RFID technology and Web GIS technology. Through the pressure and performance test of the system, it is found that the user's waiting time is shorter, the system performance is stable, and the designed system can meet the needs of business operations, providing the warehouse management personnel with real-time information of goods location, inventory and making various reports and data. The results show that the system can reduce the cost of storage management by up to 40.3%, reduce the time of management by nearly half, and greatly improve the management efficiency. At the same time, owing to the use of intelligent information tools, it can also reduce the mistakes caused by manual operation and improve the competitiveness of enterprises.
Keywords: visualisation technology; intelligent warehouse management; RFID technology; web GIS technology.
Image Recognition Technology Based on Neural Network in Robot Vision System
by Yinggang He
Abstract: Robot vision system has great research value and broad application prospects in robot navigation and positioning, human-computer interaction, unmanned driving, disaster rescue and other fields, among which image recognition technology plays an important role. The purpose of this study is to analyse the application of image recognition technology based on neural network in robot vision system. This research uses CamVid training decoder to train the model, then fine tune the parameters on the collected data, label the manually collected data with LabetMe annotation tool, and cross verify the image and scene with neural network algorithm and image recognition principle technology. After five training cycles, the neural network in this study can achieve more than 90% recognition accuracy, and achieve convergence after storing about 10 cycles. Finally, the recognition accuracy in the test dataset can reach more than 95%. In the range of robot vision recognition, the maximum measurement deviation is only 2.54 cm and the error is less than 2%. It can be concluded that the method used in this study has fast convergence speed, high recognition accuracy, small error, and good practicability and effectiveness. It improves the recognition efficiency of the robot, the processing ability of the complex environment and the precise positioning of the object.
Keywords: neural network; image recognition; machine vision; recognition system.
Mechanical fault detection method of weighing device sensor faced on internet of things
by Yan Dong, Shiying Bi
Abstract: With the advancement of science and technology, electronic scales including a load sensor have been widely used in various industries to achieve fast and accurate material weighing. Especially with the advent of microprocessors and continuous improvement of automation degree in industrial production processes, load sensors have become a necessary device in weighing process control, but there is currently no method for mechanical fault diagnosis of load sensors. This experiment samples the zero point output signal of the weight sensor, and then having taken out n consecutive values by using a sliding window, it finds the standard deviation of n values. Finally, it takes the ratio of the standard deviation to the normal output standard deviation as the testing base. When the ratio is greater than the set threshold, the sensor is faulty, otherwise there is no fault. The experimental results show that 20 normal output data are randomly selected from the zero test data of the weighing sensor, and the standard deviations of one or more sequences are calculated based on these 20 data. The average of the 10 standard deviations is used as the weighing sensor, and there is no standard deviation at zero drift. This method can monitor the running status of multiple devices in real time, predict the time of equipment failure, and detect creep faults as early as possible. By setting the critical value, the system can indicate possible faults before reaching the absolute limit, and ensure the maintenance in advance to continue the normal operation of machinery and equipment.
Keywords: load sensor; fault diagnosis; signal sampling; creep fault.
Rapid analysis and detection algorithm and prevention countermeasures of urban traffic accidents under artificial intelligence
by Zhao Yang, Yingjie Qi
Abstract: This article is a study on the rapid analysis and detection algorithm and preventive countermeasures of urban traffic accidents under the artificial intelligence threshold. It analyses the characteristics of artificial intelligence technology and uses its flexibility, comprehensiveness, and practicality to simulate a set of rapid analysis and detection models for accidents. The experimental data show that: the accident probability of the accident warning is highest when the vehicle density is 50 vehicles/km, and the detection accuracy is highest when the vehicle density is maintained at 20 vehicles/km. The experimental conclusions show that the artificial intelligence-based urban traffic accident risk prediction model constructed in this paper can effectively predict the possible and potential accidents.
Keywords: artificial intelligence visual threshold; urban traffic accidents; particle filter lane detection; traffic accident black spots.
Deep learning-based comprehensive monitor for smart power station
by Yerong Zhong
Abstract: With the wider distribution of power substations, monitoring and control of substations at large scale become more difficult by solely relying on human inspection. Smart monitoring systems are increasingly important to realise fast response, low-cost maintenance, and autonomous control. In this paper, we develop a novel inspection system based on deep learning and edge computing techniques. Firstly, the on-site video acquisition is completed by drones only when abnormal situations are detected, realising flexible and low-cost inspection. Using deep Q-learning, we design an efficient and reliable navigation algorithm that guides drones to the target location with minimum human intervention. To reduce the response latency and support large-scale data processing, we take the advantages of edge computing and build a high-performance edge system. Moreover, several strategies from algorithm to hardware are proposed to optimise the processing pipeline of constructed edge computing system. The experiment and simulation results demonstrate the reliability and efficiency of our proposed system in the case of autonomous substation monitoring.
Keywords: UAV; deep reinforcement learning; power substation control; edge computing.
Power transmission line anomaly detection scheme based on CNN-transformer model
by Min Gao, Wenfei Zhang
Abstract: The anomaly of power transmission lines has resulted in the failure of power delivery system, which brings about tremendous loss for the economy and industry. The wider distribution of power delivery system has imposed huge challenges on monitoring and making response to the anomaly cases in a short time. In this work, we introduce an autonomous anomaly detection system by exploiting computer vision (CV) and Internet of Thing (IoT) techniques. At the first step, we design and develop an IoT sensor that can detect and feedback physical conditions around the power tower. Once the anomaly situation occurs, the on-site image acquisition is carried out by drones. To simplify the construction of image analysis pipelines while maintaining high accuracy, we adopt the state-of-the-art (SOTA) cascaded convolutional neural network (CNN)-transformer model. According to our experiment results, the CNN-transformer model is able to provide promising performance for anomaly detection of power lines, achieving higher average precision while consuming almost the same computing resources. The proposed anomaly detection scheme is of importance for realising large-scale and autonomous anomaly detection for power lines.
Keywords: IoT sensor; power automation; anomaly detection; computer vision; neural network.
High-performance polar decoder for wireless sensor networks
by Sufang Wang
Abstract: The Internet of Things (IoT) has promoted lots of advanced applications and become the hot topic for the development of next-generation networks. As an important part of the IoT, low power and high reliability are two crucial metrics when designing wireless sensor networks (WSN). In this paper, we propose a high-performance decoder for polar codes to improve the link reliability and transmission efficiency. We modify and optimise the original polar belief propagation decoding algorithms through investigating approximation and several types of factor. Moreover, the systematic coding of polar codes is used to further improve the error-correction performance by increasing acceptable encoding complexity. We conduct detailed experiment results to demonstrate the low-complexity and high-performance advantages of the proposed decoder. The improved robustness and low complexity of communication reduce the energy consumption and improving the information transmission reliability, which is suitable for the battery-operated low-complexity WSN applications.
Keywords: wireless sensor network; forward error correction; polar codes; internet of things.
K-means clustering algorithm for data distribution in cloud computing environment
by Hailan Pan, Yongmei Lei, Shi Yin
Abstract: Clustering analysis is a very active field of data mining technology. It can find things in the potential data of interest to people. The purpose of this study is to make a clustering algorithm can be used in customer relationship management, realise the development of related marketing activities and decision-making, and let the company maintain customer resources to improve competitiveness. A good customer management model can greatly help companies to predict the development of customers and set customer level. In this paper, the data structure of clustering analysis is analysed. The common clustering algorithms will use data matrix, dissimilarity matrix and similarity matrix. K-means clustering algorithm is an algorithm that does not need to be supervised and depends on the number of clusters. It is a clustering method that randomly selects the known number of points and then continues to expand. Through the comparative experiments of clustering accuracy of different similarity matrices, the experimental analysis of model effectiveness, the distribution of e-commerce data under cloud computing and the computing time of different clustering algorithms, we can better understand the K-means clustering algorithm and the status of e-commerce in cloud computing environment. The experimental results show that if the appropriate similarity function is selected, the result of spectral clustering is not lower than that of K-means clustering. When the number of users reaches 4000, the list reading time of K-means clustering algorithm is 3.15 s, and the times of the three other algorithms are more than that. The efficiency of K-means clustering algorithm in small data processing effect is not obvious, but when it reaches a certain level of data volume, the efficiency will be significantly improved, and the distributed computing ability is fast.
Keywords: cloud computing; data distribution oriented; K-means clustering algorithm; e-commerce platform.
A medical specialty outpatient clinic recommendation system based on text mining
by Qing-Chang Li, Xiao-Qi Ling, Hsiu-Sen Chiang, Kai-Jui Yang
Abstract: Many prospective medical patients have difficulty determining which type of outpatient specialist to consult for their complaint, and their resulting enquiries impose an additional administration cost for hospitals. This research collects illness control data from various hospitals to establish a database fronted by a chatbot-based interface for the development of a medical specialty outpatient clinic recommendation system using speech recognition and text mining. The proposed system integrates speech recognition, the Jieba word segmentation algorithm and the conditional random field algorithm to retrieve keywords during the dialogue process. Based on C4.5 decision tree, analysis results are used to provide clinical department referrals for the patients reported symptoms. Results are tracked and feedback is sent to the cloud database to gradually correct errors and improve decision performance. Tree nodes reduce the error rate of outpatient recommendations, freeing medical staff from having to make such referrals or redirect patients to the correct department, thus reducing medical staff workload and the amount of time patients spend in the hospital.
Keywords: chatbot; text-mining; medical department; recommendation system; decision tree.
Neural network classifier based on genetic algorithm image segmentation of subject robot optimisation system
by Hongbo Ji, Mingyue Wang, Mingwei Sun, Qiang Liu
Abstract: A robot optimisation system is a kind of complex, nonlinear, strong coupling system with serious uncertainty. The effect of image segmentation has become an important index to judge the merits of many algorithms. The purpose of this study is to explore the effect of neural network based on genetic algorithm on image segmentation in the optimisation system of a classifier subject robot. The method used is to calculate the pretrained VGGl6 NET model as the pretraining model through the framework of the genetic algorithm. The resolution of the training picture used is 640 * 480, the learning rate is 10-5, the value of batch size is l, the number of iterations is set to 12,000, and then the trained model is used to detect the image. The results show that the average error of group B of SNN trained by BP algorithm is 11.62%, the SNN trained by SGA has reduced the result to 9.75%, and the error is reduced to 7.75% by the genetic algorithm in this study. Moreover, the genetic algorithm is better in feature point extraction, and the detection rate reaches 94.62%, which is higher than 77.53% and 88.74% of other methods. The missing rate of this study is only 3.04%, far lower than 12.49% and 7.36%. The conclusion is that our genetic algorithm has obvious advantages, small error, high efficiency and good applicability. The neural network based on genetic algorithm in this study has a certain value in image segmentation technology.
Keywords: genetic algorithm; neural network classifier; robot optimisation system; image segmentation; feature point extraction.
Special Issue on: Smart Living Technology and AIoT Innovations
The use of intelligent remote monitoring system in ship energy efficiency management based on internet of things
by Jianhua Deng, Songyan Mai, Zeng Ji, He Zhang, Bowen Jin, Longyu Bu, Chaochun Huang, Hui Jiang
Abstract: The aim of this work is to carry out remote monitoring of ships and realise intelligent recording of ship energy efficiency. Big data technology, 6G communication technology and embedded technology are used to build a remote monitoring system of ship operation energy efficiency from the level of hardware and software. The specific research work can be divided into three aspects. First, the existing ship operation energy efficiency indicators are analysed, and the ship energy efficiency operation indicator (EEOI) with the most balanced performance is found as the evaluation indicator of ship operation energy efficiency. Then, the requirement analysis is carried out, and the overall framework design is completed according to the analysed requirements. Moreover, the hardware module selection and peripheral circuit design are completed based on the framework parameters. Finally, the software part of the system is built. The test results prove that the functionality and stability of the platform meet the actual needs. Then, the system is tested on board for 20 days. The test results show that, in 20 days of data collection, the experimental ships are only in the range of level 1 operation energy efficiency 15 times, in the range of level 2 operation energy efficiency 40 times, and in the range of level 3 operation energy efficiency 5 times. Then, according to the classification results, the ship handling is continuously optimised, and the ship energy efficiency level is continuously improved. The system proposed has a certain reference value to help ships in China to achieve energy saving and emission reduction. Meanwhile, to a certain extent, it solves the problem of the lack of effective monitoring means and management methods for inland ship operation energy efficiency in China.
Keywords: big data; ship energy efficiency management; ship remote monitoring; internet of things.