International Journal of Ad Hoc and Ubiquitous Computing (70 papers in press)
Enhancing Data Privacy through Decentralized Predictive Model with Blockchain-based Revenue
by Sandi Rahmadika, Kyung-Hyune Rhee
Abstract: Federated learning (FL) permits a vast number of connected to construct deep learning models while keeping their private training data on the device. Rather than uploading the training data and model to the server, FL only sends the local gradients gradually. Hence, FL preserves data privacy by design. FL leverages a decentralized approach where the training data is no longer concentrated. Similarly, blockchain uses the same approach by providing a digital ledger that can cover the flaws in the centralized system. Motivated by the merits of a decentralized approach, we construct a collaborative model of simultaneous distributed learning by employing multiple computing devices over shared memory with blockchain smart contracts as a secure incentive mechanism. The collaborative model preserves a value-driven of distributed learning in enhancing users' privacy. It is supported by blockchain with a secure decentralized incentive technique without having a single point of failure. Furthermore, potential vulnerabilities and plausible defenses are also outlined. The experimental results positively recommend that the collaborative model satisfies the design goals.
Keywords: Blockchain; Decentralized Revenue; Decentralized Training; Federated Learning; Predictive Model; User Privacy.
Optimal Solar Panel Placement in Mobile Edge Computing
by Seyed Yahya Vaezpour
Abstract: Mobile Edge Computing (MEC) enhances smart phone computational and storage capabilities by offloading tasks from mobile phones to the edge of the network where servers are placed near base stations. Since data centers consume electricity more and more each year, we consider locating renewable energy sources such as solar panels near each MEC. We tackle these challenges by strategically placing solar panels near MECs to minimize the energy costs. We formulate and solve the optimization problem and present two algorithms to solve the optimization problem. Moreover, we analyze the workload of each MEC during past years, and we present an algorithm to predict future workload of different MECs. We evaluate our solution using real Google traces collected over a 29-day period from a Google cluster. Simulation results show that with the help of our optimization algorithms, we can effectively reduce the electricity costs and increase the green energy consumption.
Keywords: Green Energy; Mobile Edge Computing; Solar Panel Placement.
A TDMA Protocol with Reinforcement Learning Slot Selection for MANETs
by Chih-Yu Lin, Chih-Hsiang Wang, Yu-Chee Tseng
Abstract: With the rapid development of wireless communication and the advantage of infrastructure-less technologies, mobile ad hoc networks (MANETs) have attracted great attention on military and rescue applications. Medium Access Control is an important issue in MANETs. Contention-based MAC protocols (e.g., CSMA) do not ensure a reliable transmission due to the possibility of collisions. On the contrary, schedule-based MAC protocols (e.g., TDMA) can solve the collision problem with a scheduled transmission plan. However, under an infrastructure-less environment, it is non-trivial for each node to determine its own transmission plan. This work investigates how to use Reinforcement Learning (RL) to help nodes determine their transmission plans in a TDMA protocol. More precisely, we design a cross-layer TDMA protocol with a RL-based slot selection algorithm. We have validated the proposed protocol by the ns-3 network simulator.
Keywords: medium access control; mobile ad hoc networks; ns-3; reinforcement learning; slot assignment; TDMA.
Time and Energy-Efficient Hybrid Job Scheduling Scheme for Mobile Cloud Computing Empowered Wireless Sensor Networks
by Mahfuzulhoq Chowdhury
Abstract: Today, the rapid growth of the emerging applications based on real-time ubiquitous sensory data collection, processing, and delivery poses a significant challenge to researchers due to the WSNs resource limitations. To minimize the WSNs job processing time, cloud computing technology has become a potential solution due to its high processing capability and storage resources. To minimize the WSNs battery power usage and speed up the overall job execution process, the development of a proper job scheduling scheme has emerged as a potential research challenge for cloud empowered WSNs by taking into account different job requirements, multiple job arrival, and resource characteristics. In this paper, we propose a time and energy-efficient hybrid job scheduling scheme for WSNs job execution that not only assigns sensor-cloud resources for each job processing but also allocates network timeslot resources for transferring sensory data over cloud empowered WSNs. We present an analytical model to evaluate the performance of our proposed hybrid job scheduling scheme and compare its performance with both traditional only contention and reservation-based scheme. The experimental results clearly show that the proposed hybrid job scheduling scheme outperforms the existing only contention and reservation-based scheduling schemes in terms of schedule length, mean job execution time, energy cost, financial cost, and communication cost-to computation cost ratio. Our experimental results demonstrate that the proposed hybrid job scheduling scheme supersedes the state of the art up to approximately 28.05% in the metric of schedule length.
Keywords: Cloud Computing; Energy Consumption; Job Scheduling; Job ExecutionrnTime; Schedule Length; Wireless Sensor Networks.
Black Hole Attack Prevent Scheme using Blockchain-Block Approach in SDN-Enabled WSN
by Emre Karakoç, Celal ÇEKEN
Abstract: The success of Software-Defined Networking (SDN) technology in the wired environment has accelerated the research on the deployment of SDN in Wireless Sensor Networks (WSN). Since the wireless environment has a shared nature, it is more vulnerable to routing attacks. Most of the existing security protocols cannot be implemented for WSN, as the wireless environment has physical restrictions and the nodes in the network have limited energy, processing capacity, and memory. This paper investigates the possible threats in SDN-enabled WSN and analyzes the Black Hole attack in detail. In the paper, we have also proposed a novel lightweight security model that exploits the Blockchain-block approach to be able to protect the Flow Table in each node, which is the main target of possible routing attacks. We have generated an unchangeable fingerprint called the signature token for the flow entries with a secret key belonging to each node. Thanks to this token, not only have the flow entries been secured against tampering but also data packets have been transmitted securely along the path and secured against replay attacks and man-in-the-middle attacks. The results show that the proposed security model can be a promising solution for securing SDN-enabled WSN structures.
Keywords: IoT; SDN; WSN; Blockchain; Flow Table Security; Black Hole Attack.
Tree Reconstruction using Energy Aware Sink Mobility to Prolong Network Lifetime in WSN
by Hemanth Kumar, Nagarjun E, Dilip Kumar S M
Abstract: Activities sensed through out the sensor field are forwarded to the sink by the sensor nodes. This leads to faster dissipation of node energy that are positioned around the sink when compared to the energy of other nodes that are deployed away from the sink. This leads to energy hole problem. The main objective lies in collecting the sensed data from all over the sensor filed, aggregate and forward the same to the mobile sink to extend the network lifetime. The sensor network is logically divided into subgrids and each subgrid is uniquely identified and the sink is aware of the count of sensors under each subgrid. During data collection process, the sink node having mobility move towards the potential subgrid so that the energy of all the sensor nodes is equally utilised. The implementation is done through reconstructing the network and generating new routing paths in the network. The sink moves to a new subgrid where sensor nodes have more residual energy, gets associated with new neighbor nodes and continues to operate, which prolongs the network lifetime. This paper is an extended work of our proposed Energy-Aware Sink Mobility (EASM). Parameters like average life time of nodes, packet delivery ratio and throughput of the network are considered for performance analysis. Residual energy obtained from the simulation shows that the lifetime of the network is improved.
Keywords: Wireless Sensor Networks; Residual Energy; Lifetime; Tree Reconstruction;
Mobile Sink; subgrid.
Detecting LDoS in NB-IoTs by using Metaheuristic-based CNN
by Chi-Yuan Chen, Hsin-Hung Cho, Min-Yan Tsai, Augustine Ho Hann Sii, Han-Chieh Chao
Abstract: Narrow Band Internet of Things (NB-IoT) is a Low Power Wide Area Network (LPWAN) technology and has been included in the 13th edition of the 3rd Generation Partnership Project (3GPP) standard. It also proves that NB-IoT has great potential in the application of the Internet of Things. Therefore, the number of IoT devices will grow explosively, and the IoT network environment will become a hotbed of Botnets. Various IoT devices and platforms will be threatened by Denial-of-Service (DoS) attacks. However, NB-IoT has the characteristics of lower speed (60kpps uplink) so it is not an ideal environment for heavy traffic DoS attacks. Low-Rate Denial-of-Service (LDoS) is the main attack method in the NB-IoT environment. In this environment, the attacker can hide the attack packet for avoiding detection in a data stream with a sufficiently low rate has greatly increased the difficulty of detection. There are already many Convolutional Neural Networks that use artificial intelligence to identify the characteristics of this attack. However, these traditional methods will cause the amount of data to be insufficiently diverse. In order to improve the phenomenon of overfitting, this article uses Simulated Annealing to adjust the weight of the convolutional neural network to achieve better global search. The experimental results show that the method proposed in this article can detect LDoS attacks more effectively.
Keywords: Narrow Band Internet of Things; Intrusion Detection System; Deep Learning; Heuristic Algorithm.
Building a Smart Rehabilitation Delivery Platform with Hybrid Matching Algorithm Based on Long-term Care Plan in Taiwan
by Lun-Ping Hung, Chien-Liang Chen, Ruay-Shiung Chang, Yu-Chen Hsu
Abstract: A smart city uses a new generation of information technology fully deployed in every business to optimize the functionality of the city with innovative ideas and services to improve the citizens quality of life. Smart medical care is a part of urban services. Long-term care is currently an important item urgently needed in the development of smart medical care and after-discharge convalescent medical care is an even more important research agenda. Most of the hospital resources are put into the care of acute patients. To solve the deficiency in medical care budgets and labor forces, the current medical care development trend for most smart cities is going toward dispersive care rather than the concentrated model. This research employs a hybrid matching algorithm to construct a community and home rehabilitation service mechanism so the therapist and the individual case will find the most suitable counterpart by setting the matching parameters. Combining recording of the rehabilitation activities, the feedback from the users, and the therapists recommendations, we make decision suggestions to related medical care agencies. It achieves the goal of optimizing medical care to the patients.
Keywords: Gale-Shapley Algorithm; Home Rehabilitation; Long Term Care; Matching Algorithm; Decentralized Health Services.
Optimal Allocation of renewable distributed generators and shunt capacitors in distribution system using hybrid intelligent approach
by Suresh Kumar Sudabattula, Velamuri Suresh, N. Prabaharan, R. Sitharthan, M. Rajesh
Abstract: Increased penetration of Renewable Energy Sources (RES) in a Distribution System (DS) is customary in recent years. These RES are intermittent in nature and impact the system performance significantly. In this paper, a methodology is utilized for bidding the lowest price to the consumer based on the availability of RES. A physical communication infrastructure between the utility and user is enabled to coordinate between the RES dominant utility and consumers. The main objectives of this communication enabled user utility interface system is to reduce the power loss and improve the voltage profile simultaneously. A multi-objective approach considering Active Power Loss Index (ALPI), Voltage Deviation Index (VDI), and Voltage stability index (VSI) are presented to solve the simultaneous allocation of Renewable Distributed Generation (RDG) and Shunt Capacitors (SCs) after receiving the communication signal from the user. A combination of sensitivity analysis and Random Walk Moth Flame Optimization (RWMFO) is utilized for solving the aforementioned multi-objective function. The developed method is implemented on IEEE 33 and IEEE 69 bus test systems and compared with recently available methods
Keywords: Secure communication; Privacy-preserving smart contracts; Renewable energy; Power Loss; Voltage profile; Sensitivity Analysis (SA); Random Walk Moth Flame Optimization (RWMFO).
Optimal Harvesting Duration for CDMA Systems
by Sami Touati, Rachid Sammouda, Musaed A. Alhussein
Abstract: This paper studies the throughput of Code Division Multiple Access (CDMA) systems with energy harvesting. A CDMA transmitter $T$ harvest energy using the received signals from node $N$. There are $P$ paths between $N$ and $T$. Then, the harvested energy is used to transmit data to a CDMA receiver $R$. There are $L$ paths between $T$ and $R$. We also suggest to optimize harvesting duration to maximize the throughput. CDMA systems with energy harvesting don\'t require to recharge or change the battery of transmitter $T$. Our results are valid for any number of paths between $N$ and $T$ and any number of paths between $T$ and $R$.
Keywords: Code Division Multiple Access (CDMA); energy harvesting; multipath channels; throughput analysis.
Design, Implementation, and Evaluation of a Shared LoRa Network Application Architecture
by I-Hsuan Peng, Yu-Chun Tai, Pei-Chun Lee
Abstract: This research proposes an Internet of Things application architecture based on the essence of sharing and reciprocity a Shared LoRa Network application architecture. To prove the feasibility of this architecture, we designed and implemented a Pet Tracking System which consists of the LoRa tracking Motes, the fixed and portable Shared LoRa Gateways, a service mobile app, and the back-end service, exploiting technologies such as LoRa, Global Positioning System, Bluetooth Low Energy, Arduino UNO microcontroller, mobile app, and RESTful API. Moreover, this research evaluates the performance of the Shared LoRa Network and the elec-tricity consumption of the LoRa tracking Mote as well as the portable Shared LoRa Gateway. We confirm that the portable Shared LoRa Gateways do improve the overall network performance, and the users do not need to worry about the battery life of the devices in everyday life situations.
Keywords: Internet of Things; IoT; LoRa; LoRaWAN; Low Power Wide Area Networks; LPWANs; performance evaluation; pets; shared; system design; tracking.
UAV Deployment and Channel Allocation Considering Diverse QoS Constraints and Service Importance
by Cuntao Liu, Wendong Zhao
Abstract: The deployment of UAV (unmanned aerial vehicle) as well as the allocation of channels is important for satisfying users\' differentiated demand for information transmission in UAV-enabled communication. With diverse QoS (quality of service) requirements and service importance of users considered simultaneously, a joint optimization problem of UAV deployment and channel allocation is proposed. And the maximization of service-importance weighted system capacity, on the premise of maximizing the number of users served simultaneously, is taken as the optimization objective. A rapid hybrid algorithm combined greedy algorithm and swarm intelligence optimization algorithm is then proposed to solve this problem. Simulation results show that the weighted average capacity obtained based on our algorithm is about 10.4% higher, with slight influence on the unweighted capacity at the same time, than that without considering service importance, while maximizing the number of users accessing the channels.
Keywords: unmanned aerial vehicle (UAV); deployment; channel allocation; data integration; service importance.
Smart Grid Wireless Communications: A Cyber-Physical System Perspective
by Ruxiang Fu, Jiawei Zhang, Yu Zhang, Xudong Wang
Abstract: It is widely accepted that wireless communications play an important role in smart grid. However, wireless communication usually lacks sufficient quality, security, and reliability that are critically demanded by power grid. Thus, it is indispensable to study the applicability of wireless communications for smart grid. In this paper, interactions between a wireless communication network and smart grid are characterized from the perspective of a cyber-physical system (CPS). Based on this CPS framework, integration of smart grid and wireless networks is categorized into two scenarios: loosely-coupled CPS and tightly-coupled CPS. The features of both scenarios are discussed and compared, which leads to clear guidelines on the design of a smart grid wireless communication system. To concretely demonstrate these guidelines, two examples are presented to show different roles of wireless communications in loosely-coupled CPS and tightly-coupled CPS. Following these examples, remaining research challenges and potential solutions are discussed.
Keywords: wireless communications; smart grid; cyber-physical system; loosely-coupled; tightly-coupled; field trail.
A fault Tolerant Algorithm for Integrated Coverage and Connectivity in Wireless Sensor Networks
by Nishat Afshan Ansari, Umesh Deshpande, Sahista Mohammad
Abstract: There are two major requirements for any surveillance application of Wireless Sensor Networks (WSN) coverage and connectivity. Additionally we need the algorithm to be fault tolerant so that a high packet delivery ratio can be achieved for a large percentage of network lifetime. In the design of any algorithm for WSN, the major constraint is limited battery life time of the sensor nodes. Most of the existing work on combined problem of coverage and connectivity either require the clocks of the nodes to be synchronized or require exact location information of the nodes to be known. In this paper, we solve the combined problem of coverage, connectivity and fault tolerance and propose an algorithm called as a Fault Tolerant algorithm for Integrated Coverage and Connectivity (FTICC) in WSN. Our algorithm does not require clock synchronization of nodes at any stage neither does it require exact location information of the nodes. In the following, we discuss the approach used. First, we build a minimum hopcount graph for connectivity. Second, we propose a scheduling algorithm for providing desired coverage. Third, we define a fault tolerant connectivity maintenance algorithm (FTCM) to re-ensure connectivity as network may become disconnected due to scheduling. We give formal proof of the correctness of the algorithm and give complexity analysis at each stage of the algorithm. FTICC is energy efficient, fully distributed and scalable.
Keywords: Wireless Sensor Networks; Coverage; Connectivity; Fault tolerance.
Reward Based Service Provisioning Scheme for UAV-MEC assisted IoT Infrastructures
by Mahfuzulhoq Chowdhury
Abstract: To lower the IoT device task execution latency, Unmanned aerial
vehicles (UAVs) are gaining popularity among researchers that offers ubiquitous computing
and communication services to IoT devices. To maximize the system task execution time and energy saving in different real-time cases, collaboration among UAV and mobile-edge computing (MEC) for IoT device task execution seems to be a promising
solution. The existing task execution algorithm in the integrated UAV-MEC
system suffers from higher latency and energy wastage due to lack of a proper IoT service provisioning scheme. To accomplish both time and energy gain, this paper presents a reward-based service provisioning scheme for UAV-MEC assisted IoT infrastructures by
taking into account different types of UAV, IoT device, cloud,
and task properties. Simulation results indicate that our proposed reward-based service
provisioning algorithm outperforms the traditional air and ground offloading algorithm in terms of system timespan delay and energy cost latency.
Keywords: Air and ground offloading; Unmanned aerial vehicles (UAV); Internet of Things (IoT); Mobile-edge computing (MEC); Service provisioning; Task assignment; Throughput; Time reward; energy reward; Energy consumption.
RSSI-based Node Selection Using Neural Network Parameterized by Particle Swarm Optimization
by Cheng Wang, Kyung Tae Kim, Hee Yong Youn
Abstract: ZigBee is a popular wireless communication protocol developed for low-cost, short-range networking such as wireless sensor network (WSN) consisting of a large number of sensor nodes. Since the limited connectivity of each node, some nodes might become isolated from the network, especially during the stage of ZigBee commissioning. The performance of the existing node connection schemes for WSN is significantly influenced by the noise or operational condition of the environment. In this paper a machine learning based scheme for effective node connection of ZigBee-based WSN is proposed. It employs back propagation neural network (BPNN) to accurately estimate the received signal strength indicator (RSSI) samples used for the selection of the nearest node for connection, while the particle swarm optimization (PSO) algorithm is employed to properly initialize the BPNN. Computer simulation reveals that the proposed scheme consistently allows higher accuracy of node selection than the existing RSSI-based schemes in varying operational conditions, even with a small number of RSSI samples. It also requires smaller processing time.
Keywords: ZigBee-based WSN; Node connectivity; RSSI; Isolated node; Dixon’s test; BPNN-PSO.
Graph-Based Optimal Routing in Clustered WSNs
by Layla AZIZ, Said RAGHAY, Hanane AZNAOUI
Abstract: Wireless sensor networks (WSNs) represent a huge number of tiny sensors that
cooperate for easing wireless communication. They are widely used in various applications
such as transportation, environment control, and patient monitoring. However, these
applications still encounter difficulties because sensors are highly resource-constrained.
Hence, optimizing the whole network energy is a major goal of WSNs. Even though
great research works have been proposed for saving network energy and maximizing
network lifetime, WSNs are not sufficiently adapted to their applications requirements.
Graph theory represents an effective solution for modeling network communication. In
this paper, we aim to enhance the process of routing in WSN using an efficient graph
theory model that optimizes the routing cost. This work improved the inter-cluster
communication by the use of a strong network classification using a weighted function.
Moreover, it aims at routing data using an optimal routing path based on the graph theory
model. To enhance the generated solution, our work consists of activating only sensor
nodes that detect enough strong signals to minimize the cluster density. The use of this
technique eases minimizing the energy consumed during data aggregation. The graphbased solution is evaluated using the network density and the data size scenarios. The
numerical experiments prove that our new protocol improved significantly the network
stability compared to Low-Energy Adaptive Clustering Hierarchy (LEACH) and Modified
LEACH(MODLEACH) routing protocols.
Keywords: Clustering; Shortest path; Graph theory; Routing cost.
Design of Trust Identification System for Cloud Services
by Devi Rajendran, Shanmugalakshmi R
Abstract: Cloud computing is merging into every aspect of our information exchange including our personal information, business environment and other environments along with numerous security and trust challenges. With rapidly developing Cloud utilization, the range of security issues and trust challenges also increases exponentially. Hence an efficient Trust Identification System (TIS) is needed to identify the region of trusted Cloud Service Providers (CSPs) for a particular service. In this paper, a novel trust identification technique is proposed based on linear programming using objective functions which can smartly spot the region of trust from the CSP pool. In this system, the cluster of trusted CSP is defined and the remaining CSPs are not recommended to the Cloud Customer (CC). The Internet architecture is a wide area network with interoperability, scalability and durability. The analysis of the proposed system proves to be efficient in terms of accuracy and time complexity.
Keywords: Cloud computing; Cloud Service Provider; Linear programming; Trust Identification System;.
Machine learning soft computing fuzzy set to identify ubiquitous integrated network state at different AM operations
by Kadapa Harinadha Reddy
Abstract: Issues arising during the operation like interruption, disorder and uncertainty of network to be resolve. One of the situations is island state, and require to secure human beings and converter devices those are working in integrated energy sources environment. Fuzzy sets are to be identifying this state and these are updating with device control parameters of network like voltage, frequency. This paper presents a k-nearest neighbor Machine Learning (knn-ML) to obtain Upper Limit (UL) Lower Limit (LL) of parameters for continuous fuzzy based state estimation. Also proposed knn-ML helps to overcome existing challenges and motivate to better recognition of integrated network state. Outcomes of the paper, to update device control at every Abnormal Mode (AM) of operation and as when system parameter changes at different conditional operating modes. Effective outcome has been obtained with knn-ML soft computing algorithm.
Keywords: Fuzzy Set; Fuzzy control; Island state; Integrated network; Machine learning; Abnormal mode.
A multi-group dragonfly algorithm for application in wireless sensor network deployment problem
by Xiaopeng Wang, Shu-Chuan Chu, Han-Chieh Chao, Jeng-Shyang Pan
Abstract: Metaheuristic algorithm is a popular research field. In recent years, a host of optimizers have been presented. Dragonfly algorithm (DA) mimicking the behavior of dragonfly performance markedly competitive to some optimization problems. However, the DA sometimes does not perform well when encountering some complicated problems, easily falls into the local optimum, and premature convergence. To overcome the deficiencies of the canonical DA, this paper presents an enhanced version of DA, namely the multi-group dragonfly algorithm (MDA). The proposed MDA with three communication strategies applies effectively the multi-group trick to improve the diversity of the population. To verify the performance of the proposed MDA, it is evaluated by different benchmarks include unimodal functions, multimodal functions, hybrid, and composed functions. The experimental data confirm that the MDA performs better than the DA. Besides, the MDA is also applied in the wireless sensor network deployment problem, the simulation results appear that the MDA can obtain a more ideal sensor node distribution.
Keywords: metaheuristic algorithm; multi-group dragonfly algorithm; dragonfly algorithm; wireless sensor network deployment problem.
A Trust-Based Secure AODV Routing Scheme for MANET
by Charu Wahi, Shampa Chakraverty, Vandana Bhattacherjee
Abstract: Secure routing in mobile ad hoc networks remains a challenging research problem due to their dynamic and distributed nature of the operation. Cryptographic approaches to establish secured routing are not quite suitable for ad hoc networks because of the inevitable presence of a third party and due to high computational overheads involved. A viable alternative approach stems from the trust-based communication protocol embedded in the network layer. We introduce a trust-based solution named Trust-based Secure AODV (TSAODV), to establish a secure and reliable inter-node communication with minimal memory and communication overhead. Each nodes trustworthiness is assessed based on its current performance as well as its past behavior. Simulation results demonstrate that the proposed trust driven model improves packet delivery ratio and throughput as compared with conventional ad-hoc routing schemes while minimizing the penalty in terms of end-to-end delay.
Keywords: Mobile ad-hoc networks; Trust-based routing; Adhoc On-demand Distance Vector (AODV) protocol; secured routing.
A General and Efficient Distance Bounding Protocol with multi-Objective Optimization for RFID Applications over a Noisy Channel
by Ebrahim Shafiee, Abolfazl Falahati
Abstract: It is common to employ the Distance Bounding (DB) protocols to prevent relay attacks over RFID communication systems. It is impossible to utilize only one specific version of a DB protocol as a general resolution for secure identification and optimal performance due to the diversity and development of the RFID platforms such as ad hoc and IoT. Therefore, a general DB (GDB) protocol employing a multi-objective optimization algorithm is proposed to find the appropriate GDB protocol's parameter values, increase security, and reduce costs. On the other hand, reducing the false-accept probability against possible attacks increases security, and decreasing the protocol false-reject probability, runtime, and memory reduces costs. The final desired optimal protocol is obtained from Pareto optimal solutions using the linear non-dimensionalization scaling method and the LINMAP decision-making process. Ultimately, thanks to the sensitivity analysis and the comparisons with other appropriate protocols, a novel general relation is extracted from the obtained optimal protocols. Introducing the Figure of Merit (FOM) criterion, analytical analysis, and simulation results indicate significant improvements in the Protocol efficiency.
Keywords: RFID; Distance Bounding Protocol; Relay attacks; Multi-objective ?optimization; Fraud attacks; Communication security; Noisy wireless channel.
Special Issue on: Artificial Intelligence for Edge Computing in the Internet of Things
Application of Artificial Intelligence Technology in CNC System
by Chunhui Dong, Cheng Zhong
Abstract: Since the development of computer numerical control technology from hardware numerical control to software numerical control, computer numerical control technology is still in a period of continuous improvement of functions. Although some novel numerical control technologies have been proposed, on the whole, it has not yet broken through the traditional framework. And because of the reliability research process of CNC machine tools, it is difficult to collect reliability data, which makes the reliability distribution model not unique. Based on the above background, the purpose of this article is to study the application of artificial intelligence technology in numerical control systems. The main method is to use the ANN model to expand the small amount of reliability data collected, and then use the Since the development of computer numerical control technology from hardware numerical control to software numerical control, computer numerical control technology is still in a period of continuous improvement of functions. Although some novel numerical control technologies have been proposed, on the whole, it has not yet broken through the traditional framework. And because of the reliability research process of CNC machine tools, it is difficult to collect reliability data, which makes the reliability distribution model not unique. Based on the above background, the purpose of this article is to study the application of artificial intelligence technology in numerical control systems. The main method is to use the ANN model to expand the small amount of reliability data collected, and then use the KS test to analyze the expanded data to determine the reliability data model. At the same time, in the process of determining the parameters of the reliability distribution model, the mixed Particle Swarm Optimization (HPSO) algorithm is introduced into the maximum likelihood estimation to solve the problems of easy to fall into the local optimal solution and low efficiency when solving some complex distribution models with small sample data. The experimental results show that the reliability distribution model of CNC machine tools is not unique in the case of a small sample. The reliability model of CNC machine tools can be uniquely determined after analyzing the data using the ANN model and the KS test method. Achieve a good balance between solution efficiency and convergence performance. Comparing the results of all the solving models, the relative mean square error of the 2-fold 3-parameter Weibull distribution after the ANN model expansion is the smallest, with a value of 0.0428, which shows that using this method to solve the reliability distribution model of CNC machine tools is feasible and can obtain More accurate results.
Keywords: CNC Machine Tools; Artificial Intelligence; ANN Model; HPSO Algorithm.
Functional feature-aware APP recommendation with personalized PageRank
by Yueyue Xia, Xiangliang Zhong, Yiwen Zhang, Yuanting Yan
Abstract: With the explosive growth of mobile Apps and the widespread deployment of Internet of Things (IoT) services, how to recommend applications that users are interested in has become an urgent problem to be solved. And most of the current mobile APP recommendation methods are based on the user\'s behaviour data or context-aware information, to some extent, ignoring the user\'s preference for APP functional features. Therefore, a functional feature-aware mobile APP recommendation method, named S-AppRank, is proposed in this paper. S-AppRank first extracts the functional features of mobile applications and their internal association through weight calculations, then constructs a directed graph of user functional features with the association rules, and then adds user ratings to the traditional PageRank algorithm, incorporating explicit feedback into recommendation personally. Finally, the user\'s interests in the overall APP is predicted, and a recommendation list is generated. The experiments on the real dataset of Huawei application market show that the S-AppRank proposed in this paper is better than other comparison methods.
Keywords: functional feature-aware; APP recommendation; PageRank; user preference.
Development of Internet Finance Industry with the Core of E-commerce Platform Services Optimized by the Edge Computing of the Internet of Things Based on Artificial Intelligence
by Baojun Yu, Anni Zhao
Abstract: E-commerce is one of the important modes of modern commerce. This paper discusses how the logistics service of the Internet of things optimized by the edge computing of the Internet of things based on artificial intelligence affects the satisfaction of online shoppers. The purpose of this paper is to determine the importance of the Internet of things logistics service based on the edge computing optimization of the Internet of things based on artificial intelligence to affect the satisfaction of online shoppers. In this paper, a total of 178 respondents with online shopping experience were interviewed face-to-face using structured questionnaire. Pearson correlation and multiple regression were used to analyze the data. The results show that service recovery, delivery service and customer service are the positive factors that affect the satisfaction of e-commerce shoppers. The significant level of service recovery (P = 0.000) and delivery service (P = 0.001) was 1% of the importance. The significance level of customer service (P = 0.024) was 5%. Service recovery (? = 0.399) has the largest weight in influencing the satisfaction of e-commerce shoppers, followed by delivery service (? = 0.343) and customer service (? = 0.244). The results of this study show that the Internet of things based on the edge computing optimization of artificial intelligence is important to the service-oriented optimization of e-commerce platform. The Internet of things technology supported by edge computing can promote the service-oriented optimization of e-commerce platform.
Keywords: E-Commerce Platform; Internet of Things; Edge Computing; Logistics Services.
Special Issue on: Distributed Secure Computing for Smart Mobile IoT Networks
An optimized machine learning algorithm for classification of Epileptic Seizures using EMD based dynamic features of EEG
by Sateesh Kumar Reddy Ch, Suchetha M
Abstract: The electrical activity in the brain will establish the seizure due to unexpected change in neurons, which leads to the second most common disease of the brain called epilepsy. An automatic seizure detection technique is essential for primary diagnosis and treatment because the traditional methods of seizure detection are time-consuming and inaccurate. In this regards, this proposal shows a novel seizure detection technique centred on two unique features of time-frequency analysis. The proposed two novel features based on empirical mode decomposition (EMD) such as relaxation time (RT) and dynamic bandwidth (DB) to distinguish seizure movements in EEG. The study was carried out on two different data sets. Then the two novel features are used to classify healthy and seizure subjects by Support Vector Machine (SVM) with Marginal Sampling approach. The efficiency of the proposed method has compared with different classification methods such as K-Nearest Neighbours (KNN), Decision Tree (DT), andrnNaive Bayes (NB) respectively. The MATLAB platform is used to carry the process of the proposed method, and the performance results are calculated in terms of accuracy, sensitivity, specificity, precision, and F-measure. We observed that the proposed two novel features withrnSVM marginal sampling combined with k-fold cross validation achieve the best average performance with an accuracy of 99.23% with less computational time. The result analysis shows that the proposed method is an efficient and suitable method for the classification of seizure data than existing techniques in electroencephalogram signal processing
Keywords: Empirical mode decomposition;Seizure detection; Electroencephalogram (EEG) Support vector machine (SVM); Epilepsy.
Secured Personal Health Records using Pattern Based Verification and 2-Way Polynomial Protocol in Cloud Infrastructure
by DNVSLS Indira, R. Abinaya, Ch Suresh Babu, Ramesh Vatambeti
Abstract: This present research proposes the digitalized healthcare systems enable patients to generate, aggregate and store in the form of Personal Health Records (PHR). This requires more attention on cost effectiveness and less response time on public cloud platform. The emerging need of PHR monitoring and data collection on dynamic data sets, the companies need to adapt the open framework analyzing and effective storage tuples. Unfortunately, the third-party companies are failed to implement the systemic approach for immediate verification and correction models on increasing data sets. The storage and computation are two prime factors. Moreover, cloud systems need more attention on security and privacy breaches. In this proposed model the publisher-observer pattern-based healthcare systems allow the patients to verify and correct the PHR before any type of computations. The cloud system act as backend framework that offers openness and easy accessibility. The experimental segment ensures the computational cost and response time for multiple polynomial PHR variations. The details evaluation also ensures the security and privacy preservation on sensitive healthcare data sets.
Keywords: Privacy; Security; Patient Health Records; Correctness; Healthcare; 2-way polynomial.
An Energy and Delay Aware Routing Protocol for Wireless Sensor Network assisted IoT to maximize network lifetime
by Vinmathi M S, Josephine M.S, Jeyabalaraja V.
Abstract: IoT is a technology where different devices are connected and integrated to provide solutions. This is otherwise in digital world called the Internet of everything which comprises of components from web which is the ultimate source of gathering and processing a large data obtained from the respective ecosystems by making use of the sensors and various other communication devices. In this research article, A protocol for routing is proposed which is aware of the delay and energy is a WSN (EDAR- WIoT) for the maximization of the lifetime of a network. The proposed EDAR-WIoT protocol enhances and compromises the energy and delay without affecting network lifetime. The IoT network is composed by serving nodes and end users. Here, the optimal clustering is performed by the chaotic spiral dynamic (CSD) algorithm, which reduces the chaotic in nature of energy utilization. Then, a queue based swarm optimization (QSO) algorithm is utilized to next optimal node for inter cluster routing. The proposed EDAR-WIoT protocol preserves the energy efficiency and network lifetime in high density sensor networks. The proposed EDAR-WIoT protocol is experimented using the NS-2 simulator tool. It is observed from the experiments that the proposed protocol outperformed the other existing protocols in terms of Energy Consumption, lifetime of the network , throughput , rotational frequency of the head and end-end delay.
Keywords: Routing Protocol; internet of things; dynamic algorithm; optimization algorithm; network security.
SLIDING WINDOWING ASSISTED MUTUAL REDUNDANCY BASED FEATURE SELECTION FOR INTRUSION DETECTION SYSTEM
by THOTAKURA VEERANNA, Kiran Kumar R
Abstract: Due to the widespread utilization of intent, the computer systems are more prone to several kinds of security threats that have led to the invention of Intrusion Detection Systems (IDSs). However, the major issue in IDSs is the presence of huge redundant and duplicate features which cause a larger processing time. These features not only slow down the process and also make the classifier to take inaccurate decision and consequences to an insertion of serious attacks into the system. To solve these problems, we propose a Sliding Windowing Assisted Mutual Redundancy Based Feature Selection (MRFS) algorithm that finds the duplicate features analytically and selects optimal feature for detection. This MRFS evaluates the mutual redundancy between as well as within network traffic connections and then selects an optimal feature subset to represent each connection attribute. After feature selection, they are fed to Multi-Class Support Vector Machine (MC-SVM) based IDS for classification. The performance of MRFS-IDS is evaluated using a standard intrusion dataset, i.e., NSL-KDD and performance is measured in terms of accuracy and false alarm rate. The simulation results demonstrate that the proposed MRFS-IDS model has gained an accuracy of 95% approximately and false alarm rate of 0.70% which is much better than the counterpart methods.
Keywords: Intrusion Detection; Data redundancy; Mutual Redundancy; Sliding Windowing; Mutual Information; MC-SVM.
An Approach Using Heuristic Pheromones-Based ACO Modeling for Green Vehicle Routing Optimization
by Ravi Prakash, Shashank Pushkar
Abstract: A mathematical heuristic-based method was introduced for addressing the issue in Green Vehicle Routing Optimization (GVRO). It analyzes a large number of vehicles along with a limited refueling network. A standard solution to this problem is given in this paper. GVRO seeks to minimize travel time renewable fuel sources while ensuring fewer emissions from greenhouse gases. An effective algorithm relies on a branch/slice optimization algorithm that combines a variety of valid inequalities in exams to increase lower limits. Implementation of an optimization algorithm based on heuristic Ant Colony Optimization (ACO) to obtain the best routes. In addition, the GVRO is better able to handle an accident and eliminates pollution by using the best alternatives.
Keywords: Green vehicle routing; Greenhouse gases; Heuristic approach; Environmental pollution; Pheromones; Ant colony optimization.
Secure Exchange and Effectual Verification of Educational Academic Records Using Hyperledger Fabric Block chain System
by Suresh Babu Erukala, Srinu Mekala, Naganjaneyulu Satuluri, Srinivasa Sesha Sai M., RAJENDRA KUMAR G
Abstract: Education plays a vital role in the countrys economic development. It improves the quality of people lives and leads to broad social benefits to individuals and society. Education raises peoples productivity and creativity, and promotes entrepreneurship and technological advances. Academic certificates in education are the documents that serve as proof of an individuals achievements or skill sets. These certificates are the symbol of excellence and vouch for professional skill sets of an individual in various numerous fields of the education sector. However, traditional academic certificates disconnect lifelong learning records, an increase of frauds, and the fake certificates that pose a serious problem in todays world. Existing centralized certificate verification process creates difficulty in verifying the authenticity, and lack of transparency in issuance of the certificates. Blockchain is a promising technology that claims to provide distributed, transparent, secure and reliable solutions to various business use-cases, which are having multiple participants with transparency and enhance trust among them. In this paper, we provide a solution to the educational certification problem by employing the blockchain network. The proposed network is a permissioned blockchain infrastructure, which is implemented in hyperledger fabric. The proposed system provides various services to issuing institutions, verifying organization, identification and authentication of the issuer, verifier and securely share academic records to the recipients, stores the academic records in the blockchain in a distributed manner, ensuring the privacy of stored records of the recipient. When compared to Ethereum,hyperledger fabric provides additional functionalities like efficient parallelism, concurrency, multiple transaction executions, and efficient commitments of the transaction into the ledger. The proposed system is a secure and trusted system that can be applied to any education sector, which provides the learners and education institutions to safeguard their brands and reputations, transparent sharing of certificates, and easyverification of academic achievements. The experimental analysis of the proposed system has been executed to test the performance of invoking and query transactions (certificates) using Hyperledger Caliper. We analyze the throughput and transaction latency of the proposed work as well. The experimental results exhibit that the proposed system achieves better transaction processing power and security compared to existing systems.
Keywords: Permissioned Blockchain; Academic Certificates; Hyperledger Fabric; Distributed Ledger.
A Novel Bio-Inspired Approach for VM Load Balancing and Efficient Resource Management in Cloud
by Purshottam J. Assudani, Balakrishanan P
Abstract: In cloud, balancing the load on Virtual machines and Efficient utilization of resources is very crucial and challenging, which becomes complex due to heterogeneous nature of virtual machines and users tasks in distributed environment. To maintain the service quality, it is mandatory that users tasks should be scheduled efficiently with immediate response, while satisfying QoS needs mentioned in SLA (Service Level Agreement). Concerning these issues, researchers have designed bio-inspired algorithms to solve optimization problems for resource scheduling. In this paper, we have proposed a bio-inspired method namely Inventive Particle Swarm Optimization (IPSO), which not only schedules users task efficiently, but also uniformly distributes the load among different VMs. Additionally, we have also designed another algorithm named as Merge Sort with Divide and Conquer (MSDC) approach to allocate the resources in cloud dynamically in an efficient manner. The experimentation is done on CloudSim simulator, which shows that proposed algorithms give better response time, VM utilization and execution time.
Keywords: Bio-Inspired Algorithms;Cloud Manager; VM Load Balancing; Scheduling of Resources; Resource Utilization; Dynamic Resource Allocation.
Efficiency Evaluation of HRF mechanism on EDoS attacks in Cloud Computing Services
by Rajendra Kumar G, Veeraiah Duggineni, Suneetha Bulla, Nageswara Rao Jarapala, J. Sunny Deol . G.
Abstract: Cloud computing is one of the most notable innovations in the IT Industry. It gives computing resources, software, web benefits for all cloud users on a rent basis. Security on the cloud is a blend of virtual assets, specialized and informational security issues. Denial of Service is one of the renowned compared to other known assaults and it produces from many sources contradicted to one single victim, these are called Distributed Denial of Service (DDoS) assaults. The conventional DDoS assaults can be transformed into EDoS (Economic Denial of Sustainability) assaults because of cloud elasticity. This EDoS assault uses the cloud assets for creating administration inaccessibility to the clients. There is a mandate to diminish EDoS assaults. HRF is the most suitable and effective mechanism to identify and diminish such assaults, in which assailant requests are recognized and dropped preceding arriving at the webserver. This paper assesses and examines the cost and performance sway using queuing theory and assesses the experimental model in terms of key performance metrics which incorporate QoS and cost metrics. Different scenarios appropriate to the HRF mechanism are taken into consideration and examined. Performance is compared with existing approaches using the game-theoretical methodology. To get the systematic solution and calculation of game value, various probabilities of defending techniques and assaulting strategies through numerical outlines are done lastly conclusions are drawn.
Keywords: HRF mechanism; Queuing theory; Game theory; Web Application Firewalls; EDoS attacks.
Privacy-Preserving smart contracts for Fuzzy WordNet Based Document Representation and Clustering using Regularized K-Means Method
by Venkata Nagaraju Thata, Sudhir Babu A, Haritha D
Abstract: Key technology for unsupervised intelligent classification of any textual content is the clustering of documents. Prior document knowledge is not required for document clustering, which is an unsupervised method of learning as compare with document classification. For clustering rather than classification, little prior knowledge of the data is needed. The crucial challenges of document clustering are the high dimensionality, measurability, preciseness, extraction of semantic relationships from texts, and meaningful cluster labels. Fuzzy wordnet based document representation and clustering using the regularized k-means method as an efficient framework is introduced in the present paper with the purpose of improving the quality of document clustering.To estimate the performance of this framework we carried out experiments on different datasets. Experimental results show that this framework improves the quality of document clustering when compared to other existing methods. Furthermore, this system gives generalized and concrete labels for documents and improves the speed of clustering by reducing their size.
Keywords: Document Clustering; Regularized K-Means; WordNet; Fuzzy Weighting Score,TF-IDF.
A DYNAMIC PRIVACY PRESERVING BASED WSN FRAMEWORK FOR MEDICAL DISEASE PREDICTION
by B. Madhuravani, Murthy DSR, ViswanadhaRaju S
Abstract: Wireless sensor networks (WSNs) play a vital role in the real-time data communication process. Cloud based WSNs is used to improve the processing speed and the storage capacity in the real-time applications. Most of the conventional approaches are based on sensor resources with limited cloud services due to high computational cost. Also, these models are not efficient in processing large volumes of data in real-time applications due to computational memory and time. In order to improve these limitations, a hybrid machine learning based sensor network is developed to predict the medical disease patterns using the cloud servers. In this work, a hybrid PSO, support vector machine and sensor security algorithms are implemented on the real-time medical sensor devices. Experimental results show that the machine learning based sensor framework has better efficiency in terms of sensor processing time, accuracy and error rate than the conventional approaches.
Keywords: Machine learning; medical disease prediction; sensor network.
A Secure New HRF Mechanism for Mitigate EDoS Attacks
by Suneetha Bulla, Basaveswararao B, Gangadhara Rao K, Chandan K
Abstract: This paper proposes HTTP Request Filtering (HRF) mechanism to detect and mitigate EDoS attacks and compare the performance with existing mechanism through game theoretical approach. The HRF mechanism was implemented with three stages and hosted on Web Application Firewalls (WAF). The performance of these mechanisms with cost analysis is done using finite queuing model. The efficiencies are compared with the formation of two player non cooperative zero-sum game and gains are calculated based on loss probability as a QoS metric. To obtain the analytical solution and computation of game value, different probabilities of defending strategies and attacking strategies through numerical illustrations are carried out. The results are discussed and finally conclusions are drawn.
Keywords: HTTP Request Filtering; Cloud Security; Web Application Firewalls; Honeypot; Game Theory.
Improving QOS in flow controlled CR-Adhoc Network with Multi criteria Routing assisted with cooperative Caching and information redundancy
by Subaskar Reddy C V, Subramanyam M V, Ramana Reddy P
Abstract: Cognitive Radio Adhoc Network with intrinsic capabilities of Cognitive Radio solves the problem in wireless network caused due to limited available spectrum and inefficiency in spectrum usage. Due to spectrum availability, the network topology changes frequently and routing is disrupted. Packet delivery becomes a big problem due to frequent routing disruptions in Cognitive Radio Networks. In this work, a flow controlled multi criteria routing assisted with cooperative caching and packet reconstruction with information redundancy in packet is proposed to improve the QOS in Cognitive Radio adhoc network. The multi path routing path selection is guided with four parameters of current buffer occupancy in nodes, predicted link availability time, spectrum availability and cumulative path delay. Different from conventional multipath path routing, in this work propagation is controlled in proportion to the prediction based on observed channel quality thereby packet delivery ratio is increased without causing additional network overhead and at lower delay. The performance of the solution is tested against different speed and compared with state of art existing solutions.
Keywords: CR-Adhoc Network; QOS in flow control; Secure computing; IoT.
Design and Implementation of a Mobility Support Adaptive Trickle (MSAT) Algorithm for RPL in Vehicular IoT Networks
by N. V. R. SWARUP KUMAR JAVVADI, Suresh D
Abstract: With the development of the "Internet of Things," the scope and research for stable mobile assistance in Low-Power Wireless Networks (LPWNs) are growing. The components like single radio transceivers have been included in the low-power and low-cost products, interacting at very low TX / RX power to make basic electronics and antennas. These features result in insecure, asymmetric, and poor wireless communications, which affects the LPWN's quality of operation. In specific application to risk-benefit mobility nodes, vehicular networks, and preventive/corrective maintenance in manufacturing settings, this operation level is often harder to obtain. In this context, the proposed paper ensures efficient and timely contact in LPWNs under node mobility. RPL as a routing protocol is used in wireless low-power networks (LPWNs). This protocol is intended for the static wireless sensor network, so RPL needs to be changed to suit with Vehicular Networks' extremely dynamic topology. In the RPL protocol, the Trickle algorithm establishes routes between nodes in the network at different intervals. The Trickle algorithm is designed to reduce RPL's disadvantages. This paper proposes the vehicle's location as the RPL metric (MSAT-RPL), enabling RPL to be respond timely and propose RPL tuning trickle algorithm strategies in vehicular networks. A simulation was set up, "Cooja 3.0," is used to show the output of MSAT-RPL and compare it with existing models. MSAT-RPL has a high Packet Delivery Ratio (PDR), fair OverHead (OH), low latency (EED), and less power consumption (PC) as a result of the simulation.
Keywords: IoT; Vehicular Networks; RPL; DODAG; Trickle; Cooja 3.0; Listen Only Period; Adaptive DIO Period; Mobility Support Adaptive Trickle.
Analytical Review on Secure Communication Protocols for 5G and IoT Networks
by Premalatha J, Iwin Thanakumar Joseph S, Harshavardhanan Pon, Anandaraj S.P, JeyaKrishanan V
Abstract: Internet of Things (IoT) provides an exciting future for the interaction of devices utilizing computing and sensorial capabilities. Among the existing technologies, fifth generation (5G) schemes is expected to be the motivating force for the realization of IoT concept. This research paper reviews existing state of art on the secure communication for 5G and IoT networks. We have created taxonomy and categorized the research works based on the protocols such as Light weight authentication protocol, trust-aware routing protocol, service-oriented authentication protocol, and application-oriented protocol. The research gaps and the challenges faced for secure communication in 5G and IoT networks are listed for further enhancement in the security protocols. We further analyzed the research work based on the classification methods, performance metrics, and the year of publication. Our analysis reveals that the most commonly used classification technique is Trust-aware routing protocols, and the most frequently used performance metrics is execution time, the execution time of most of the research work lies between 10ms to 300ms, and the communication overhead ranges from bits to bits.
Keywords: Authentication Protocol; Internet of Things; Privacy Preservation; and Secure Communication.
F-CHILS MAPS - A NOVEL ENCRYPTION SCHEME FOR SECURED MEDICAL DATA TRANSMISSION USING FOG-BAN ENVIRONMENT
by Swaminathan Amudha, Murali M
Abstract: With an Internet of Things, remote health monitoring has significantly increased and playing an effective role in human disease diagnosis. Patients clinical data are collected from variety of tiny sensors and are transmitted to the medical physician, Care givers and medical center in remotely. Different solutions have been proposed to monitor health condition in wireless Body Area Network. To prevent security issues, malicious users involvement, timely delivery and also to ensure data integrity, a high-end security algorithms are needed. Internet of things is now integrated with secured Fog Gateways to provide high end security solution with less latency. Many traditional algorithms were incorporated in the IoT network, due to its light weight, low power and low memory requirements. But these algorithms can be easily broken and affected by several attacks due to the poor mathematical operations. This research work propose a new hybrid chaotic maps FoG based Chaotic Henon Integrated Logistic-Tents Schemes which uses the 3D-Chaotic maps for key generation and used as symmetric key for crypto graphical operations. These 3D hybrid chaotic maps are hard and have been evaluated by extensive experimentations which show the impact of the proposed 3D chaotic maps has significantly increased the security of the clinical data when compared with the other conventional algorithms.
Keywords: AES; ECC; F-CHIL Map;3D-Chaotic Maps; Logistics Tent algorithms; Lorenz algorithms; Bifurcation; Diffusion and Permutation.
A FEATURE RANKING BASED DEEP LEARNING SECURE FRAMEWORK FOR MULTI-CLASS LEAF DISEASE DETECTION
by Nagageetha M
Abstract: Multi-class leaf disease prediction is one of the challenging tasks in large image databases due to uncertainty and high dimensional feature space. Most of the traditional deep learning framework are used to classify a single class disease prediction with limited feature space. However, these frameworks have high false positive rate and error rate due to various background, noisy appearance and semantic high- and low-level features for classification problem. Also, feature extraction and classification are the major problems in traditional convolution neural network (CNN) on multi-class leaf datasets. In this work, a novel image feature extraction based deep learning classifier is designed and implemented on large multi-class leaf datasets. In this framework, a hybrid statistical leaf shape extraction method is used to find the essential features in the conditional probability based principal component analysis (BPCA) approach. A novel deep learning classifier is proposed to improve the leaf disease prediction rate with high true positivity and accuracy on the multi-class leaf disease datasets. Experimental results show that the present framework has high computational performance than the traditional deep learning frameworks for multi-class classification.
Keywords: Leaf disease classification;extreme learning; bayesian estimators; principal component analysis.
A Filter based Machine learning classification framework for cloud based medical databases
by Devi Satya Sri Velivela, Srikanth Vemuru
Abstract: Machine learning tools and techniques play a vital role in the medical field and cloud computing applications. Most of the traditional machine learning models use static metrics, limited data size and limited feature space due to high computational processing time. In this work, a hybrid outlier detection and data transformation approaches are implemented on the cloud based medical databases. Proposed data filtering module is applicable to high dimensional data size and feature space for classification problem. In the classification problem, an advanced boosting classifier is implemented on the filtered data in order to improve the true positive and error rate. Experimental results are simulated on different medical datasets such as tonsil and trauma databases with different feature space size and data size. Simulation results proved that the proposed boosting classifier has better error rate and statistical accuracy than the conventional approaches.
Keywords: Cloud computing Medical databases; Machine learning.
CUCKOO SEARCH ASSISTED FUZZY LOGIC ALGORITHM FOR SMART WSN ROUTING SYSTEM
by Mohana Sundaram K, Nageswari D, Prakash J
Purpose This paper aims to provide the secured communication between the networks of WSN and this security is provided by creating the secrete keys between the neighboring keys. In this paper, the location based key (LBK) management is utilized. This paper also addresses the performance of the network more efficiently by cuckoo search assisted Fuzzy logic algorithm.
Design/Methodology/approach In this paper, each node is analyzed to find the occurred threats and to find the security levels. This paper uses the cuckoo search assisted fuzzy logic algorithm to improve the network performance and the security level is enhanced by location based key (LBK) management. By this method, the load at each node is balanced with the improved security level and also the lifespan of the network is improved.
Findings This proposed method demonstrates the level of security in communication between the nodes. By using this algorithm the delay, performance, error and the security level of the network are identified.
Originality The Cuckoo search assisted fuzzy logic algorithm is the novel method, which is used to get better performance of the network and to improve the communication security.
Key words: WSN, Fuzzy logic, Cuckoo search algorithm, LBK.
Keywords: WSN; Fuzzy logic; Cuckoo search algorithm; LBK.
FORECASTING THE DECISION MAKING PROCESS OF SUPREME COURT USING HIERARCHICAL CONVOLUTIONAL NEURAL NETWORK
by SIVARANJANI NAMASSIVAYAM BALAS, Jayabharathy J
Abstract: Artificial Intelligence is one of the most energizing innovations applied in many fields like text processing, image processing. Every case, which is present in the courtroom, is inspired to get justice. Since every individual has their own view on a particular topic, the irregularities in the views of the people lead to the conflict and make them seek justice. In this paper, we propose a decision forecasting model of cases in the Supreme Court of India. The model interprets the legal cases in a similar fashion as a lawyer and forecasts a decision based on the information given. The model aims to predict whether the filed case in the Supreme Court of India will win or not by considering the past similar cases from the years 2000-2019. The proposed model does not only considers the cases filed in the Supreme Court but also the cases with an unsatisfied decision from the lower court. This is to be able to better predict if the current case will win if an appeal is chosen.
In this paper, two algorithms have been proposed (i) Bi-SVM, it is used to classify the nature of the cases as civil or criminal. (ii) C-XGB is used to predict the chances of whether the case will win or not. When an out-of-sample case, for which a decision is to be made is given as input, the model yields 96% of accuracy which is higher than the accuracy of the existing models.
Keywords: Neural Networks; Machine Learning; Feature Engineering; chi2 (?2); CNN; error metrics.
Architecture and Routing Protocols for Internet of Vehicles: A Review
by Farhana Ajaz, Mohd. Naseem, Sparsh Sharma, Gaurav Dhiman, Mohammad Shabaz, S. Vimal
Abstract: Modern vehicles should be able to commute a tremendous amount of data and information within their neighborhood. To incorporate the requirements of modern vehicles, the conventional Vehicular Ad-hoc Network (VANETs) are emerging to the Internet of Vehicles. IoV keeps all the smart vehicles connected with the help of Sensors, GPS, Entertainment System, Brakes and throttles. These devices send and store their data with the help of cloud. This paper intends to contribute to the review of IoV, its challenges, characteristics and application. A detailed discussion on architectures and routing protocols along with its classification is also discussed. This paper ought to guide and motivate researchers working in the area of IoV to develop scalable and efficient routing protocols.
Keywords: Internet of Vehicles; Internet of Things; Routing protocols; Architecture; VANETs; MANETs; Cloud Computing; FOG Computing.
Special Issue on: Convergence of Soft Computing for Smart Cities
SHA-AMD: Sample-Efficient Hyper-Tuned Approach for Detection and Identification of Android Malware, Family and Category
by Aamir Rasool, Abdul Rehman Javed, Zunera Jalil
Abstract: Smart cities provide smart security solutions against cyberattacks to the communities. Lately, Android-based smartphones emerge as the best selling product in the market. Android technology is advancing in both usage and popularity as more people are using it to connect with every passing day. Due to this popularity and extensive usage, the android OS in mobile technology has become a lucrative target for attackers as they are coming up with new types of viruses, malware, and threats. The malware of Android devices is emerging in different forms, and each time with a different signature. Conventional approaches for detecting/classification malware need continuous improvement to perform well against these changing malicious intentions. Specific malware features can be picked for detection, and performance can be improved using deep ensemble techniques. In this paper, we propose a solution for the detection and identification of novel malware. We have utilized deep learning models for malware detection, identification, and classification with feature reduction to optimize model performance and use ensemble techniques for better results. Experimental results demonstrate better performance than state-of-the-art techniques.
Keywords: Smart Cities; Smart Security; Malware; Family; Category; Deep learning; Machine learning; Cyberattack.
A CA-GRU Based Model for Air Quality Prediction
by Jingyang Wang, Xiaolei Li, Tingting Wang, Jiazheng Li, Qiuhong Sun
Abstract: Smart cities aim to maximize the optimization of urban functions, promote economic growth, and use smart technology and data analysis to improve the life quality of urban residents. The air quality index (AQI) is an important evaluation index of air pollution, describing the degree of air pollution and its impact on the health of human beings. Therefore, it is particularly important for smart cities to accurately predict AQI. The accuracy of AQI prediction remains a challenge for current methods. In this paper, an AQI prediction model based on CA-GRU is proposed, which includes convolutional neural networks (CNN), attention mechanism, and gated recurrent unit (GRU). In this model, the feature extraction of input data is realized through CNN, weights are assigned according to the states of old data with attention mechanism, and the AQI is predicted with GRU. To prove the validity and accuracy of the CA-GRU prediction model, we take experiments using air quality data and weather data from 1st January 2017 at 00:00 to 30th September 2020 at 23:00 in Shijiazhuang city of China and compare this model with other models. Finally, mean absolute error (MAE), mean square error (MSE), explained variance score (EVS), and R2 are used to evaluate the performance of the model in this experiment. The results shows that this model has the highest performance than the others, with the MAE of 6.099281, MSE of 90.781522, EVS of 0.972560, and R2 of 0.972495.
Keywords: CNN; Attention; GRU; AQI Prediction.
Evaluation of Functional Maturity for a Network Information Service Design & Case Analysis
by Mohammad Al Rawajbeh, Vladimir Sayenko, Issam AlHadid, Fadi Al-Turjman, Lakshmana Kumar Ramasamy
Abstract: The current study discusses methodological issues for monitoring process to control the state of developing information system. The main goal of this evolutionary process is to achieve the optimal value for some criteria. The proposed approach is considered as a conception for the evolution of properties for system (object) at the planning phase. The study has proposed the evolution conception of system properties for planning (development) phase, concept of improving the system properties, the indicators of the functional maturity, and the technique of evaluation of functional maturity. the planning phase; the quality indicators named functional maturity; a concept of the evaluation of functional maturity; estimates of the functional maturity indicators and a technique for functional maturity evaluation monitoring conception. The implementation of this technique has developed a tier-based procedure. The results have demonstrated that network planning should follow integrated work practices along with the implementations of solid system design.
Keywords: COBIT; Development; Functional Maturity; Information System,; IT System; Methodological Issues; Monitoring Process; Network Information Service,; Network Structure; Planning Phase.
Lion plus Firefly Algorithm for Ternary based Anomaly Detection in Semantic Graphs in Smart Cities
by M. Sravan Kumar Reddy, Dharmendra Singh Rajput
Abstract: The detection of suspicious or abnormal entities in massive data sets is really a complex task in homeland security in smart cities. Even though there were several technologies for data mining and social network analysis; detecting the abnormalities in huge semantic graphs is often the complex aspect as the nodes are deeply linked with countless links. A node is termed to be the abnormal node when it carries unique or abnormal semantics in network. This paper aims to introduce a new idea of finding the abnormal or suspicious nodes, and this is done by modeling the graph structure using diverse nodes and links associated to the node in a particular distance through edges. Once the graph structure is framed, the ternary list formation is done, and the list of all nodes is based on the nearby list of neighborhood nodes. Subsequently, the logic is induced in this work to detect the abnormal nodes via optimizing the node pairs. For this, a new hybrid algorithm termed as Threshold-based LA Update (T-LAU) that hybridizes the concepts of Lion Algorithm (LA) and Firefly (FF), respectively. The similarity among mined features in terms of rules recognizes the abnormal behavior of nodes in network. Finally, the analysis is carried out for validating the enhancement of the presented model.
Keywords: Semantic graph; Anomaly detection; Ternary List; Lion Algorithm; Firefly Algorithm; Smart Cities.
User Participation Behavior in Crowdsourcing Initiatives: Influencing Factors, Related Theories and Incentive Strategies
by Xu Zhang, Zhanglin Peng, Qiang Zhang, Xiaonong Lu, Hao Song
Abstract: Crowdsourcing is a powerful paradigm that leverages collective intelligence to solve problems. Good performance in crowdsourcing initiatives depends on energetic user participation. This paper reviews and analyzes the current existing research works that are related to user behavior in crowdsourcing initiatives. Particular attention is paid to the following aspects. First, we summarize the influencing factors of user participation behavior in crowdsourcing initiatives. Second, we review the related behavior theories to further understand the relationship between these factors and user behavior. Third, we generalize about incentive strategies from the perspective of requesters and crowdsourcing platforms. Finally, the research directions of user behavior in crowdsourcing initiatives are discussed in this study.
Keywords: Crowdsourcing; user participation; user behavior; motivation; incentive strategies.
GSFI_SMOTE: A Hybrid Multiclass Classifier for Minority Attack Detection in Internet of Things Network
by Geeta Singh, Neelu Khare
Abstract: The Internet of Things (IoT) network is prone to several cyber-attacks, due to many obligations of IoT devices. Minority attack detection in the IoT network is a challenge. This paper proposed three multiclassifier models to address this challenge at different IoT layer. Random Forest (RF) classifier is the main component in proposed models. RF hyperparameters are tuned with Grid-Search Cross-Validation (GSCV) to build an initial GS model that achieves a 100% normal traffic detection rate. It can efficiently separate normal and anomalous traffic at the IoT network layer. GS is extended with Feature-Importance (FI) based feature reduction technique and the Synthetic Minority Oversampling Technique (SMOTE) successively to realize GSFI and GSFI_SMOTE models that achieve better minority attack detection rates and applicable to the resource-limited fog/edge infrastructure, and the critical IoT infrastructure respectively. GSFI_SMOTE outperforms the existing methods. The UNSW-NB15 is used as a use case for experimenting with proposed models.
Keywords: Security; Internet of Things; Network Monitoring; Attack Detection; Anomaly; Machine Learning; Random Forest; Feature Importance; Oversampling; Parameter Tuning.
Optimization of K-means algorithm based on sample density Canopy
by Guoxin Shen, Zhongyun Jiang
Abstract: Since the random selection of the initial centroid and the artificial definition of the number of clusters affect the experimental results of K-means, therefore, this article uses sample density and Canopy to optimize the K-means algorithm. This algorithm first calculates the sample density of each data, and selects the data point with the smallest density as the first cluster centroid; then combines the Canopy algorithm to cluster the original sample data to obtain the number of clusters and each cluster center. As initial parameters of the K-means; finally combines the K-means algorithm to assemble the original samples. UCI dataset and self-built dataset were used to compare simulation experiments. The results show that the algorithm can make clustering results more accurate, run faster, and improve the stability of the algorithm.
Keywords: Clustering; K-means algorithm; Density; Neighborhood; Initial centroid.
AN UNCERTAINTIES EVALUATION AND ANALYSIS USING QUANTITATIVE TECHNIQUE FOR A SOFTWARE PROJECT
by Harvinder Singh, Adarsh Kumar, Kiran Kumar Ravulakollu, Manoj Kumar, Thompson Stephan
Abstract: Opportunities cannot be converted into achievements without assessing risk. Software developers and project managers presume that activities would proceed smoothly and according to plan in software project development. However, in reality, this is not always true due to the existence of unseen risks that inrnuence the development process and thereby performance. Hence, there is a need for more realistic risk evaluation criteria that can address the project managers need. This paper proposes a Project Risk Evaluation Technique (PRET) to determine the attractiveness of a software project. The proposed technique used normal distribution (DN) approach for risk understanding and thereby use the activity network to assess them. Using the graph theory method, with nodes as phases of the development process and edges as the duration of phases and the whole network as a software project, the algorithm systematically represented to extend various influential aspects of risk analysis. Hence, using quantitative analysis, the proposed algorithm is evaluated against negative exponential distribution (DNE) and is found to have a slight degree of uncertainty associated with risk and attractiveness. This signifies that there is still a scope of improvement for stabilizing the proposed technique with minimal improvements in the project factors.
Keywords: Risk Analysis; Software Risk Management; PRET Algorithm; Normal Distribution; Negative Exponential Distribution.
Recent advances in Blockchain Technology: Asurvey on Applications and Challenges
by Saqib Hakak, Wazir Zada Khan, Gulshan Amin , Mamoun Alazab, Sweta Bhattacharya, G. Thippa Reddy, Basem Assari
Abstract: The rise of blockchain technology within a few years has attracted researchers across the world. The prime reason for worldwide attention is undoubtedly due to its feature of immutability along with the decentralized approach of data protection. As this technology is progressing, lots of developments in terms of identifying new applications, blockchain-based platforms, consensus mechanisms, etc are taking place. Hence, in this article, an attempt has been made to review the recent advancements in blockchain technology. Furthermore, we have also explored the available blockchain platforms, highlighted and explored future research directions and challenges.
Keywords: Blockchain technology; Blockchain platforms; Healthcare; Cybersecurity; Finance; Smart city; Smart Grids; Logistics; Supply Chain; Ownership services; Hyperledger.
Device and Method for Dynamic Image Display of Financial Transaction Operation Data
by Jingdong Yan, Wuwei Liu
Abstract: With the booming economy, financial transactions have also become very active, but there are often situations where operational data is not obvious in the process of financial transactions. This paper aims to improve the visibility of data in the process of financial transactions, provide data support for transaction decision-making, and ensure the security of the operation process. This research mainly discusses the devices and methods for dynamic images to display financial transaction operation data. The device is divided into an early warning module, a user interaction module, an intelligent management module and a monitoring module. The early warning module is mainly based on a large amount of financial transaction data to warn customers against credit risks. The user interaction module is mainly to provide users with the display of dynamic images during financial transactions. The intelligent management system provides an interface for model meta data management to the model manager, and provides an interface for obtaining requirements to the requirements reporting staff. The monitoring module mainly realizes the monitoring of transactions, and promptly warns illegal transactions that occur during the transaction. At the same time, the user's transaction data displays the calculation results of the calculation engine through the result display page, which includes the chart of the sub-sequence to be returned and related information of the sequence. After referencing the results, the user can return to the home page to perform a new search. Based on the brightness perception function of HDR images, combined with the conversion chain reaction in the form of Open EXR, appropriate decimal floating point data is embedded in each pixel to hide the user's transaction information. During the test, the average transmission efficiency of archive logs was 77%, and the average transmission efficiency of active logs was 79%. The research results show that the dynamic image display system device of financial transaction operation data has excellent performance, which can provide real-time transaction information to users.
Keywords: Dynamic Images;Financial Transactions;Operating Data;Intelligent Management.
A Machine Learning Approach for Celebrity Profiling
by Durga Prasad Kavadi, Fadi Al-Turjman, Adi Narayana Reddy K, Rizwan Patan
Abstract: The celebrity profiling is a type of authorship profiling, which is used to predict the sub profiles like gender, fame, birthyear and occupation of a celebrity for a given textual content. Knowing the details of the celebrities by analysing their written text became one interesting area of research to the researchers. The task of Celebrity Profiling is introduced in PAN competition 2019. The researchers from different countries participated in the competition and proposed different types of solutions to the Celebrity Profiling. Most of the researchers showing interest on the stylistic features like most frequent terms, most frequent word n-grams, character n-grams, part of speech tags n-grams to differentiate the writing styles of the celebrities. In this work, the experiment started with a set of stylistic features for predicting the accuracies of sub profiles like gender, fame and occupation. The experiment continued with the most frequent terms used in the corpus for sub profiles prediction. In both experiments, we used frequency measure in the representation of a vector value in the document vector representation. It was observed that the accuracies of two approaches are not satisfactory. A new approach named as sub profile based weighted approach is proposed to improve the accuracy of celebrity profiling. In this approach, most frequent terms are used to compute the document weight. The document weights were used to represent the document vectors instead of weights of the features. Two different term weight measures are used to calculate the weight of the terms and a document weight measure is proposed to compute the document weight. The document vectors are given to machine learning algorithms as input to build the training model. Three algorithms such as random forest, support vector machine and naive bayes multinomial are used in the experiment to generate the model. This model is used to predict the sub profile of a celebritys text. The accuracies of proposed approach for sub profiles prediction are improved when compared with existing approaches for celebrity profiling.
Keywords: Celebrity Profiling; Author Profiling; Gender Prediction; Fame Prediction; Occupation Prediction; Machine Learning Algorithms; Accuracy.
Computational Trust Evaluation Algorithm (C-FTEA) for Cloud Models Using Fuzzy Logic Approach
by VAISHALI THAKARE
Abstract: Cloud computing era has brought us many services on the pay-as-you-go business paradigm. And offers several advantages over traditional computing models in industries, so applications which are developed for smart cities requires large amount of storage. Hence, smart cities applications are using cloud computing technologies to manage their applications with great flexibility and availability. But, cloud computing has many challenges to be addressed and one of the most trending challenges is security and trust. After several research by various researchers, they have designed trust evaluation algorithms and trust evaluation models that can be used in a cloud computing environment. But a satisfactory level of trust is not achieved yet and hence we propose a computational trust evaluation algorithm using a fuzzy logic approach. Fuzzy theory is used by researchers in the area of resource optimization, scheduling, and service dependability in cloud computing security. But data storage, security, and trust using fuzzy logic have been ignored. Moreover, there is no special trust evaluation algorithm for cloud service users. In this paper, C-FTEM is designed to strengthen the security at data storage for smart cities a computational trust evaluation algorithm is designed by using a fuzzy logic approach and evaluated to prove its effectiveness.
Keywords: Cloud computing; cloud service provider; cloud service user; cloud security; computational trust models; fuzzy logic; smart cities.
Development of Image-guided Puncture Robot used in Trigeminal Neuralgia Treatment
by Bin LIU, Hongbo YANG, Kai GUO, Caijun LUO, Chang LIU, Senhao ZHANG, Yingying ZHANG, Zhenlan LI
Abstract: Recently, invasive surgery had developed rapidly and was widely used in the fields of tumor treatment, spine, and joint disease treatment. At present, most of the analgesic puncture operations were performed manually by doctors in the CT environment. High skills are required during such complex surgery, it also increases the doctors risk due to exposure to the radiation environment. This paper studies image-guided puncture robots. In this work, through the use of binocular vision cameras, real-time medical image data was collected during the surgery using CT. Also, we set up the model for the spatial solution of the manipulator and integrated with the image data, to make a set of graphical operation interface. Under the guidance of the doctors, the posture of the manipulator was controlled to achieve the specific position required by the doctors. Through the development of software, hardware, and algorithm, we designed and developed an image-guided, minimally invasive surgical real-time robotic system to improve the safety and accuracy of the surgeon during the puncture procedure.
Keywords: Image-guided; CT-compatible; Puncture; Surgical robot.
A Recommendation Algorithm Based on Modified Similarity
and Text Content to Optimize Aggregate Diversity
by Shuhao Jiang, Hongyun Zhao, Zhenzhen Li
Abstract: With the popularity of smart phones, many people use mobile phones to provide personalized recommendations in a smart city. Aggregate diversity is defined as recommending different categories of items to different users. This paper proposes a personalized recommendation method based on modified similarity and text content. The algorithm optimizes the similarity value through modified similarity algorithm, solves the problem of unclear item category by extracing the text features of user browsing. And it clusters according to user category preference, and research and practice personalized recommendation algorithm based on aggregate diversity optimization. Experimental results show that the proposed algorithm can improve the aggregate diversity of recommendation results while ensuring the accuracy of recommendation.
Keywords: Personalized Recommendation; Aggregate Diversity; Similarity Calculation; Text Features.
An ICT-Based Solid Waste Management System for Smart Cities: A case of Municipality in India
by Rutvij Jhaveri, Prerak Shah, Neel Patel, Dhrumil Patel, Shashank Thanki, Akash Kumar Bhoi, Jitendra Bhatia
Abstract: Smart cities are technology enabled and data driven intelligent cities with prominent use of ICT based technologies integrated with sensor devices. This paper addresses this issue by proposing a smart solid waste management system for municipalities in urban localities. This low-cost, dynamic, energy-efficient and easily deployable embedded system is proposed to effectively manage solid waste. A kit comprising of a micro-controller, a pair of ultrasonic sensors and global system for mobile communication (GSM) module is installed into the garbage container which can sense the level of garbage in the container and automatically informs the municipality authority about the same whenever the container gets full or empty. At the same time, the kit stores and updates the status of the container on a cloud storage for further analysis. The data analysis on the cloud dataset through diverse parameters provides a dashboard view via web application to infer the waste collection process.
Keywords: Smart cities; Solid waste management; Information and communications technologies; Linear regression.
Special Issue on: New Trends in Security and Privacy for Mobile Internet of Things Sensing
Research on spam filtering algorithm based on mutual information and weighted naive Bayesian classification
by Xincun Yang, Haiyun Yu, ZhiYi Jia
Abstract: In this paper, N-gram algorithm is used to construct the characteristic database of e-mail virus. Based on the characteristic database, the transmission and immune automatic behavior of e-mail virus are constructed based on the disease transmission model. Based on a large number of samples of five different virus types, a mail virus feature library is generated, and the automatic transmission and immune process of mail virus are analyzed according to the mail virus feature library. Then, aiming at the low accuracy and recall rate of the traditional spam filtering algorithm, an improved mutual information feature and weighted naive Bayesian classification algorithm is proposed to complete the spam filtering. The algorithm improves the mutual information calculation by introducing the word frequency factor and the difference factor between categories, and takes the calculation result as the attribute weight of naive Bayesian classification to complete the spam filtering. Experiments on trec06c open source data set show that the feature library generated by this method has a good performance of email virus detection, and the analysis of email virus behavior can better meet the actual work of email virus prevention. The algorithm proposed in this paper has better robustness than the traditional naive Bayesian classification, and is significantly better than the traditional algorithm in spam filtering accuracy and recall rate, and has better feasibility and effectiveness in practical application.
Keywords: mail virus feature library; spam automatic behavior; mutual information feature; Weighted Naive Bayes; spam filtering.
AI and Machine Learning for the Analysis of data flow characteristics in industrial network communication security
by Zhi Xu, Jun Lu, Xin Wang, JiaHai Zhang, Mamoun Alazab, Vicente García Díaz
Abstract: AI and machine learning are a revolutionary technology being explored by the communication industry to integrate them into communication networks, provide modern services, improve network efficiency and user experience. The innovative industrial control systems are more connected to the internet connection, such that the rich resources on the network are used to their overall effectiveness. The intrusion detection system is essential for ensuring the security of the industrial control system. Hence, in this paper, a Machine learning assisted Intrusion detection system (MLAIDS) has been proposed to analyze data flow characteristics in industrial network communication security. The progressive use of proposed machine learning algorithms will improve IDS functionality, especially in Industrial Control Systems. The analysis of data flow characteristics involves the method of ensuring an adequate degree of security for a dispersed industrial network concerning some main elements, including system features as well as the present state of requirements and the implementation of suitable controls (countermeasures) that may lead to reducing the security risk under a predefined, acceptable threshold. The numerical results show that the proposed MLAIDS method achieves high detection accuracy of 98.2%, a performance ratio of 97.5%, a prediction ratio of 96.7%, F1-score of 95.8%, and least root mean square error of 10.5% than other existing methods.
Keywords: Artificial Intelligence; Machine learning; Data flow Characteristics; Industrial Network Communication Security.
Special Issue on: Machine Learning and Deep Learning Methods for the Applications in Ad Hoc and Ubiquitous Computing
An Adaptive Stochastic Central Force Optimization Algorithm for Node Localization in Wireless Sensor Networks
by Pei-Cheng Song, Shu-Chuan Chu, Jeng-Shyang Pan, Tsu-Yang Wu
Abstract: Node localization in wireless sensor networks is a common and important practical application problem. Among the many localization algorithms, the MDS-MAP algorithm is a more effective one. However, the positioning effect of the MDS-MAP algorithm is not accurate in some cases, so the metaheuristic algorithm is implemented to further optimize the estimation results of the MDS-MAP algorithm in this paper. The improved central force optimization algorithm uses adaptive parameters to achieve randomness, while adding the restart strategy and accelerate strategy so as to avoid getting stuck in a local optimum. The CEC2013 and CEC2014 benchmark test suites used to verify the proposed algorithm are more competitive than some other existing algorithms. The improved central force optimization algorithm is applied to the MDS-MAP localization algorithm. The experimental results show that the improved central force optimization algorithm has a further optimization effect on the position estimation results of MDS-MAP.
Keywords: Central Force Optimization; Restart strategy; Adaptive; Stochastic; MDS-MAP; WSN.
A Compact GBMO Applied to modify DV-Hop based on layers in wireless sensor network
by Jeng-Shyang Pan, Min Gao, Jian-po Li, Shu-Chuan Chu
Abstract: Gases Brownian Motion Optimization (GBMO) has been shown a useful optimization method. The compact concept is implemented to the GMBO named Compact Gases Brownian Motion Optimization (CGMBO) so as to improve the efficiency and effectiveness of the GMBO. Simulation results based on the 23 test functions consisting of the unimodal, multimodal, fixed-dimensional functions and composite multimodal functions demonstrate the superior of the proposed CGMBO. The idea of layer concept is also proposed to implement the Distance Vector-Hop (DV-Hop) by modifying the original average distance of each hop called Layer DV-Hop (LDV-Hop), experimental results also show the proposed LDV-Hop really improve the average positioning accuracy of each node for wireless sensor network. Finally, the proposed CGMBO is combined with the proposed LDV-Hop so as to greatly reduce the position error compared with the DV-Hop. The actual error per-hop distance between nodes is large. When the calculated average hop distance of the nodes does not reach the ideal value, the actual distance between the nodes and the calculated distance will have a large deviation.
Keywords: Gases Brownian Motion Optimization; Wireless Sensor Network; Compact Gases Brownian Motion Optimization; LDV-Hop.
Auto Insurance Fraud Identification Based on CNN-LSTM Fusion Deep Learning Model
by Huosong Xia, Yanjun Zhou, Zuopeng (Justin) Zhang
Abstract: The traditional auto insurance fraud identification method relies heavily on feature engineering and domain knowledge, making it difficult to accurately and efficiently identify fraud when the amount of claim data is large and the data dimension is high. Deep learning models have strong generalization abilities and can automatically complete feature extraction. This paper proposes a deep learning model for auto insurance fraud identification by combining Convolutional Neural Network (CNN), Long Short Term Memory (LSTM), and Deep Neural Network (DNN). Our proposed method can extract more abstract features and help avoid the complex feature extraction process that is highly dependent on domain experts in traditional machine learning algorithms. Experiments demonstrate that our method can effectively improve the accuracy of auto risk fraud identification.
Keywords: auto insurance fraud; deep learning; CNN-LSTM.
Optimal Dense Convolutional Network Model for Image Classification in Unmanned Aerial Vehicles based Adhoc Networks
by Hephzi Punithavathi, Dhanasekaran S, Duraipandy P, Laxmi Lydia E, Sivaram M, K. Shankar
Abstract: Unmanned aerial vehicles (UAVs) has the potential of generating an ad hoc communication network on the fly. They are presently employed in different application areas like security, surveillance, rescue operations, and so on. Aerial image classification gains more importance in the remote sensing community and several studies have been carried out in recent days. On the other hand, deep learning (DL) is currently exhibiting excellent performance in several processes namely object detection, tracking, image classification, and so on. In this view, this paper presents an optimal Dense Convolutional Network (DenseNet) with bidirectional long short term memory (Bi-LSTM) based image classification model called optimal DenseNet (ODN)-BiLSTM for UAV based adhoc networks. The proposed model involves two major processes namely feature extraction and classification. Firstly, DenseNet model is applied as a feature extractor, where the hyperparameters of DenseNet are tuned by the use of Adagrad optimizer. Secondly, the Bi-LSTM model is applied as a classifier, which classifies the aerial images captured by UAV. Detailed performance analysis of the proposed model takes place using UCM aerial dataset and the results are investigated under several dimensions. The experimental outcome ensured the goodness of the presented ODN-BiLSTM model on the applied UCM aerial dataset over the compared methods. The ODN-BiLSTM model has provided effective image classification results over the other methods with the maximum accuracy of 98.14% and minimum execution time of 80s.
Keywords: Adhoc networks; Deep learning; Image classification; Unmanned aerial vehicle;.
Two-Stage Adaptive Weight Vector Design Method for Decomposition Based Many-Objective Evolutionary Algorithm
by Xiaofang Guo, Yuping Wang, Xiaozhi Gao
Abstract: For many-objective problems, in which the Pareto Front (PF) is complex, e.g., incomplete and degenerate, it is challenging to use the conventional weight vector method to obtain a set of uniformly distributed weight vectors on the target PF. In order to acquire the searching information of the irregular shape of the PF accurately, this paper proposes a two stage adaptive weight vector design approach for decomposition based many objective evolutionary algorithms. Firstly, a preset initial weight vector generation strategy based on the crowding information is proposed to capture the effective area of the distribution of population. Next, the Self-Organized Mapping (SOM) weight vector design method using the feedback of weight vectors in the first stage as the initial weight neurons is adopted to generate weight vectors to capture the topological structure of the distribution of PF. An external archive is further employed to store the newly generated cumulative offspring and update the weight vectors of the SOM, which can help to adaptively adjust the weight vectors. In addition, a new composite aggregation function combined metric distance with angle-distance is proposed to improve both convergence and diversity in dealing with the many objective optimization problems. The performance of our algorithm is examined using a total of 10 test problems with both regular and irregular PF. The experimental results show that the proposed method can significantly improve the convergence and diversity performance with irregular shape of PF.
Keywords: Many-objective evolutionary algorithm; two stage; crowding; self-organized mapping; Lp-metric.
Behavior-based Grey Wolf Optimizer for Wireless Sensor Network Deployment Problem
by Yu Qiao, Hung-Yao Hsu, Jeng-Shyang Pan
Abstract: The existing Grey wolf optimizer doesnt perform well in convergence and diversity of population. This paper investigates the grey wolf optimizer and proposes a behavior-based grey wolf optimizer (BGWO) based on the real behaviors of the wolf pack. In BGWO, it mainly consists of two strategies: Lost wolf strategy and mating strategy. Lost wolf strategy benefits from the phenomenon that wolves with weak adaptability will get lost during the migration of wolves. The mating strategy comes from the competition in the wolf pack for mating. In the BGWO strategy, abandoning the low-adapted individuals in the wolf pack and competing with the wolf pack for mating behavior not only increases the population diversity in wolf pack, but also reduce the possibility of getting stuck in a local solution during optimization. Eighteen benchmark functions in CEC2017 are used to test the performance of BGWO and the result shows that the performance of BGWO is better than existing algorithms in the literature such as GWO, PSO, FA and PSOGWO. Moreover, In the WSN problem, a combination of coverage rate, connectivity rate and total network energy consumption is proposed as the objective function and optimized by BGWO. The experimental results show that BGWO perform well than other algorithms.
Keywords: BGWO; Lost Wolf Strategy; Mating Strategy; WSN Deployment.
Learning Stereo Disparity with Feature Consistency and Condence
by Liaoying Zhao, Jiaming Li, Jianjun Li, Yong Wu, Shichao Cheng, Zheng Tang, Guobao Hui, Chin-chen Chang
Abstract: Most of the existing stereo matching methods have been formulated into four regular parts: feature extraction (FE), cost calculation (CC), cost aggregation (CA), and disparity refinement (DF). They can obtain high precision results in most regions through modifying parts of the four methods, but still have problems in some ill-posed regions. This paper focuses on feature consistency and confidence (FCC), discovers the new attributes of the feature, and proposes a novel neural network structure for stereo matching by measuring the consistency and confidence of features. Base on this method, the paper fuses the cost volume and calculates the pixel confidence map for cost calculation and cost aggregation. The experimental results show the proposed method outperforms most of the state-of-the-art methods on both SceneFlow and Kitti benchmarks and lowers the estimation error of stereo matching down to 1.82\% ranking at the 7th position in the Kitti 2015 scoreboard six months ago. http://www.cvlibs.net/datasets/kitti/.
Keywords: depth estimation; stereo matching ,confidence measure; feature consistency; multi-distance metrics.
Research on Multi-feature Fusion Entity Relation Extraction Based on Deep Learning
by Shiao Xu, Shuihua Sun, Zhiyuan Zhang, Fan Xu
Abstract: Entity relation extraction aims to identify the semantic relation category between the target entity pairs in the original text and is one of the core technologies of tasks such as automatic document summarization, automatic question answering system, and machine translation. Aiming at the problems in the existing relation extraction model that the local feature extraction of the text is insufficient and the semantic interaction information between the entities is easily ignored, this paper proposes a novel entity relationship extraction model. The model utilizes a multi-window convolutional neural network (CNN) to capture multiple local features on the shortest dependency path (SDP) between entities, applies segmented bidirectional long short-term memory (BiLSTM) attention mechanism extracts the global features in the original input sequence, and merges the local features with the global features to extract entity relations. The experimental results on the SemEval-2010 Task 8 dataset show that the model's entity relation extraction performance is further improved than existing methods.
Keywords: deep learning; multi-feature fusion; entity relation extraction; shortest dependency path; attention mechanism.