International Journal of Communication Networks and Distributed Systems (30 papers in press)
An Investigation on Teletraffic Attributes for Channel Selection of IoT Objects in Cognitive Radio Internet of Things Networks towards 5G
by ARNAB KUNDU, Wasim Arif
Abstract: This paper investigates multiple attributes and related comparative issues regarding channel selection of IoT objects for cognitive radio IoT networks towards 5G. CU or SU as IoT nodes may intelligently manage their activity in a licensed spectrum by accessing vacant spectrum through dynamic spectrum access methodology. So that IoT is a booming developing concern to organise digital accessories. It will also provide the fastest data connections through sensor nodes by producing more data than any other emerging technology. With the continuous evolution in CRN and cognition capability, the IoT objects may think, learn, and make decisions by perceiving outside worlds. To frame the IoT with CR-based architecture in the coming future, some attributes such as intracell/intrapool handoff latency, intercell/interpool handoff latency, link continuation probability, link failure probability, switching cost, awaited number of spectrum handoff, non-execution probability, blocking probability, dropping probability, and throughput to learn the overall network attributes.
Keywords: IoT; CRN towards 5G; CRIoT; DSA; handoff; channel selection attributes.
Research on an Optimized Encryption Algorithm for Network Information Security Communication
by Ju Li
Abstract: At present, one of the commonly used encryption algorithms is the block cipher AES method, but in the design, we often only consider the bounded attack opponent, but in the face of the needs of the development of artificial intelligence, it is difficult to meet the secure communication of network information. This time, an optimised and improved GANs encryption algorithm based on neural network is proposed. The encryption algorithm can improve the objective function and learning model, so as to achieve better algorithm security performance. Through simulation analysis, it can also be seen that with the increase of training times, the neural network training effect of Bob, Alice and Eve is better. The proposed optimisation algorithm can realise face generation in the case of non artificial knowledge, which has significant advantages compared with the traditional encryption algorithm.
Keywords: network information; safety; encryption algorithm; GANs model; neural network.
Utilization Aware VM Placement Policy for Workload Consolidation in Cloud Data Centers
by Dipak Dabhi, Devendra Thakor
Abstract: In recent years, the demand for cloud services has risen. Data centres must have a growing number of servers to accommodate rising demand for cloud services, and data centres consume a lot of energy. Virtual machine consolidation (VMC) is a strategy for reducing energy consumption in data centres by shutting down underutilised servers while maintaining service levels (SLA). The VMC process is separated into four policies: overloaded host detection, underloaded host detection, virtual machine selection, and virtual machine deployment. The utilisation aware VM placement (UAVMP) technique is presented in this research work, which efficiently selects the destination host for VMs migrating from overload/underload hosts based on the hosts utilisation and resource skewness. The performance is assessed using the CloudSim simulator. When we compare UAVMP results with power-aware best fit decreasing (PABFD), modified best fit decreasing (MBFD), first fit (FF) and least fit (LF), we find that UAVMP outperform all.
Keywords: cloud computing; VM consolidation; quality of service; QoS; service level agreement; SLA; VM selection; overload host detection; VM placement; underload host detection.
Cat swarm optimization-based mobile sinks scheduling in large-scale wireless sensor networks
by Srinivasulu Boyineni, Kavitha K., Sreenivasulu Meruva
Abstract: In wireless sensor networks (WSNs), the hotspot problem is one of the major challenging issues because it isolates some network parts and interrupts the data routing. The hotspot problem is mitigated through a mobile sink, where it visits a set of nodes in the network called rendezvous points, whereas the remaining nodes traverse their data to it. In large-scale WSNs, the travelling distance of MS is longer, and it increases the delay of reaching an RP. So, the data overflow may occur due to a limited buffer of sensor nodes. This problem is avoided by increasing the number of mobile sinks in the WSNs. In this context, a cat-swarm optimisation algorithm is used to decide the optimal set of mobile sinks and a simple geometric method to determine the optimal visiting order for each mobile sink. The proposed work is compared with start-of-art literature, and the proposed work outperforms them.
Keywords: wireless sensor networks; WSNs; data acquisition; multiple mobile sinks; cat swarm optimisation; ant colony optimisation.
A Rapidly-exploring random tree-based intelligent congestion control through an alternate routing for WSNs
by P. Suman Prakash, D. Kavitha, Chenna Reddy P.
Abstract: In wireless sensor networks (WSNs), congestion is a challenging issue, and it degrades the efficiency in terms of packet loss, energy wastage, throughput, etc. The primary cause of the congestion in WSNs is the data routing of the many-to-one pattern. It means multiple nodes can send their data to a single sink using multi-hop transmissions. To control the congestion in WSN, we use a rapidly-exploring random tree (RRT)-based mechanism to divert the data packets from the congested nodes. Initially, we use a mathematical model to determine the congested nodes in the WSNs. Further, we identify the routing path using the RRT algorithm in which the algorithm can construct a dynamic routing while avoiding the congested nodes in the routing path. We estimate the efficiency of our approach using simulation runs and compare the results using the recently published algorithm. We notice the improved performance in our method.
Keywords: wireless sensor networks; WSNs; congestion factor estimation; intelligence congestion control; rapidly-exploring random tree; routing; quality of service.
A search ranking algorithm for web information retrieval.
by Shanshan Zhi, Huanhuan Wang
Abstract: The development of the internet has seen an explosion in the amount of information, which has increased the scope of queries for users but greatly increased the difficulty of searching for valid information. In order to retrieve effective information faster, search ranking algorithms are needed to rank the retrieved information and return it to the user. This paper briefly introduced the RankNet algorithm among web information search ranking algorithms and optimised the loss function to improve its retrieval ranking performance. Simulation tests were carried out with Microsoft public data set MSLR-WEB30K. The improved RankNet algorithm was compared with the ranking support vector machine (SVM) algorithm and the traditional RankNet algorithm. The results showed that as the number of returned retrievals increased, the retrieval ranking performance of all three search ranking algorithms tended to decrease; under the same number of returned retrievals, the improved RankNet algorithm had the best performance.
Keywords: search ranking; rank learning; RankNet; pairing loss; support vector machine; SVM.
Grey Prediction-based Energy-Aware Opportunistic Routing in WSN
by NAGADIVYA S, R.Manoharan Rajendiran
Abstract: Opportunistic routing (OR) protocol is widely applied for wireless sensor network (WSN) for maximising energy as well as network lifetime. OR selects the capable forwarder set of node for multi-hop forwarding, based on residual energy and transaction history. This selection process using the forwarding set will continue for every hop, till the destination is reached. This paper proposes a new routing protocol, namely grey prediction-based energy-aware OR protocol for WSN. The grey-prediction model is a proven model which needs minimum data for prediction and is used to select the nodes for the forwarding set used in every hop. The new protocol is simulated and compared with existing ones for performances. The observations suggested that the advocated protocol shows superior performance as against the already accepted protocols about network lifetime, throughput, and energy consumed and remaining.
Keywords: opportunistic routing; wireless sensor network; WSN; grey prediction; energy efficiency.
A Novel High-efficiency Searchable Encryption Scheme Under Robot Cloud Computing Environment
by Zhongli Wang, Aiyun Ju
Abstract: This paper proposed a high-efficiency searchable encryption scheme under cloud computing environment. Through the matching calculation of keyword index set and keyword trapdoor generated by the data user, the searchable encrypted mechanism is realised. By utilising the powerful computing resources of the cloud server, the pre-decryption operation is introduced to reduce the computing time cost of the data user. The cloud server does not know anything about the original file in the cloud through a trapdoor corresponding to the user-provided keyword (the trapdoor retrieves files containing a specific keyword from a large number of encrypted files). Security analysis shows that the new scheme cannot leak data privacy information and has stronger security than other state-of-the-art schemes.
Keywords: searchable encryption; cloud computing; keyword search; pre-decryption operation.
CRLMDA: CRL Minimization and Distribution Algorithm in Cluster-based VANETs
by Dinesh Singh, Ashish Maurya, Ranvijay Ranvijay, R.S. Yadav
Abstract: This paper proposes an algorithm called CRLMDA (CRL Minimization and Distribution Algorithm) to minimize CRL in cluster-based VANETs. The algorithm monitors the use of certificates during communication. A cluster head (CH) vehicle informs regarding the certificate to the certificate authority to avoid lazy processing of safety applications. The CH vehicle that identifies a vehicle as malicious immediately initiates a local and global revocation process to avoid mishappening in network. The proposed algorithm compresses the CRL using Bloom filter and distributes it to the other vehicles via roadside units and the CH vehicles. Also, we give a message authentication algorithm of the received event reporting messages at the CH vehicle. The performance of our proposed CRLMDA is compared with two well-established algorithms, HHL and SCRLE, in varying transmission ranges and vehicle densities. The reported results show that the CRLMDA performs better in revocation overhead and CRL distribution delay than existing algorithms.
Keywords: Vehicular Ad-hoc Network (VANET); Cluster Head (CH); Malicious Vehicle; Safety Applications; Certificate Authority.
An Energy Efficient Dynamic Small Cell On/Off Switching with Enhanced k-means Clustering Algorithm for 5G Hetnets
by Janani Natarajan, B. Rebekka Issac
Abstract: The massive growth in the current and envisaged cellular traffic lead to innovations in 5G heterogeneous networks (HetNets) and implementation technologies. The small cells (SCs) or small base stations (SBS) aided macro base station (MBS) topology in HetNets effectively accomplish capacity growth and spectrum reuse at the cost of network complexity, power consumption and energy efficiency. In this paper, we propose a mechanism to maximise the system energy efficiency jointly by enhanced k-means clustering and dynamic load based SC switching algorithm for HetNets. The clustering algorithm optimises the initial centroids for maximum inter-cluster separation. Within each cluster, SC switching is decided based on a permissible threshold of inter-cluster interference. Further, the intra cluster SC coordination is formulated as a cooperative game for load balancing. Simulation results illustrate performance improvement of the proposed scheme up to 20% in energy efficiency and 16% in system throughput compared to conventional k-means clustering approach.
Keywords: small cells; heterogeneous networks; HetNets; 5G; cluster; energy efficiency; macro base station; MBS.
Cyber Defense Using Attack Graphs Prediction and Visualization
by Shailendra Mishra
Abstract: The use of the internet and other related technologies has increased dramatically in recent years. Since sensitive and critical data is readily available on these systems, this information can easily be accessed. Information leaks or attacks on networked devices are becoming more common every day. This research explores the visualisation of attack graphs in public cyberspace to predict exploit paths across networks. Vulnerability analysis reveals various aspects of the system that are exploited. By combining graph adjacency matrices cyber-attack graphs are created. With the attack graph, grey areas and research points can be easily identified. Cybersecurity and network administration can be achieved by analysing M-steps. Moreover, machine learning algorithms such as SVM, RF, KNN, LR, and multilayer perceptron (MLP) are used to detect the attack and analyse the performance of the proposed system. In terms of accuracy, recall, precession, and F-score, RF and MLP were the best classifiers.
Keywords: IDS network security; attack graph; adjacency matrix; intrusion detection system; machine learning; cyber defence.
IEEE 802.15.4 MAC protocol optimisation in body sensor networks: a survey, outlook and open issues
by Abdulwadood Mohamed Othman Alawadhi, Mohd. Hasbullah Omar, Noradila Nordin
Abstract: Body sensor networks (BSNs) have sparked a surge in interest and demand. BSNs is made up of miniature biosensors that gather and send data over a wireless network, allowing medical professionals to watch patients as they go about their regular lives and provide real-time opinions for medical diagnosis. This paper describes network architecture by BSN standard with many nodes that serve as biosensors in medical research to monitor patients health. Furthermore, IEEE 802.15.4 based on networks are often executed near other wireless networks operating in the same industrial, scientific and medical (ISM) band. In addition, this paper explained the media access control (MAC) protocol based on the IEEE 802.15.4 with operation modes distinguished by a variety of ruggedness which has led to its acceptance in many BSNs, and several restrictions recreate the crucial role in weakening its execution. This paper highlights, outlines, and critically evaluates two different MAC protocol optimisation approaches. A research plan will be demarcated for each approach for MAC protocol that has been rigorously studied the research challenges, previous solutions, and open research issues in this article.
Keywords: duty cycle; media access control protocol; body sensor network; BSN; IEEE 802.15.4; backoff.
Performance Analysis of the Hard-Decision Cooperative Multi-Stage Spectrum Sensing for Cognitive Radio.
by Mohammad Hallaq, Wissam Altabban, Omran Abbas
Abstract: The number of applications that depend on wireless communications is increasing consistently. Therefore, the spectrum, which is a scarce resource, is becoming more crowded. Recently, cognitive radio (CR) technology has stood out to be a solution for this issue. However, this technology still faces problem that drag its performance such as the hidden terminal problem which occurs while CR is performing the spectrum sensing process. This problem can be solved by using cooperative sensing, where we rely on multiple cognitive radios rather than using a single one. In this paper, the authors propose a novel cooperative spectrum sensing model with multi-stage local sensing. First, each CR applies two-stage local sensing, the first stage is energy detection, and the second is maximum-minimum eigenvalues detection. Afterwards, each CR sends its detection result to a fusion centre to make the final decision. The approach is analysed in light of different rules and parameters under IEEE 802.22 standard. Finally, the results demonstrate that the best fusion rule for this system is the majority rule.
Keywords: cognitive radio; multi stage sensing; cooperative spectrum sensing; CSS; hard decision rules; hidden terminal problem; fusion centre; energy detection.
Deep Reinforcement Learning Empowered Energy Efficient Task-Offloading in Cloud-Radio Access Networks
by Naveen Kumar, Anwar Ahmad
Abstract: Mobile applications often demand computationally heavy resources to attain high quality on the other hand running all programs on a single mobile device still consumes a lot of energy and generates a lot of delay. By breaking a single operation into distinct components, partial computational offloading may be intelligently investigated in mobile edge computing (MEC) to minimise energy consumption and service latency of mobile device. Some of the components run on the mobile device itself, while the rest are sent to a mobile edge server. Current task offloading systems, primarily focus on average-based performance indicators, failing to satisfy the deadline constraints. This paper offers a deep reinforcement learning (DRL) empowered energy efficient task offloading method, to optimise the reward under task deadline constraints. Simulation results show that the proposed technique can efficiently transfer traffic from cloud-radio access networks to next-generation node B, while also saving energy by turning off underutilised baseband processing units.
Keywords: C-RAN; deep reinforcement learning; DRL; mobile edge computing; Q-learning; resource allocation; task offloading.
Privacy and Security Improvement in UAV Network using Blockchain
by Hardik Sachdeva, Shivam Gupta, Anushka Misra, Khushbu Chauhan, Mayank Dave
Abstract: Unmanned aerial vehicles (UAVs), also known as drones, have exploded in every segment present in todays business industry. They have scope in reinventing old businesses, and they are even developing new opportunities for various brands and franchisors. UAVs are used in the supply chain, maintaining surveillance and serving as mobile hotspots. Although UAVs have potential applications, they bring several societal concerns and challenges that need addressing in public safety, privacy, and cyber security. UAVs are prone to various cyber-attacks and vulnerabilities; they can also be hacked and misused by malicious entities resulting in cyber-crime. The adversaries can exploit these vulnerabilities, leading to data loss, property, and destruction of life. One can partially detect the attacks like false information dissemination, jamming, grey hole, blackhole, and GPS spoofing by monitoring the UAV behaviour, but it may not resolve privacy issues. This paper presents secure communication between UAVs using blockchain technology. Our approach involves building smart contracts and making a secure and reliable UAV ad hoc network. This network will be resilient to various network attacks and is secure against malicious intrusions.
Keywords: unmanned aerial vehicles; UAVs; blockchain; data privacy; network security; smart contract; Ethereum.
A Random Forest Algorithm under the Ensemble Approach for Feature Selection and Classification
by Ankit Kharwar, Devendra Thakor
Abstract: Over the many previous years, research analysts have proposed diverse intrusion detection systems (IDS) tactics to manage computer threats expanding number and complexity. IDS takes all the data over the network and analyses the data using machine learning for finding the attacks. It is tough to find attacks on the network because it contains fewer records than standard data. It is significantly challenging to design an IDS for high accuracy. It also foregrounds different feature selection methods to select the best feature subset. We use the random forest feature importance for finding the best features. Single classifiers can mislead the find result, so we use random forest as classification with the help of best features. The proposed model is assessed on standard datasets like KDD99, NSL-KDD, and UNSW-NB15. The experimental result shows that the proposed model outperforms the existing methods in terms of accuracy, detection rate, and false alarm rate.
Keywords: intrusion detection; anomaly detection; machine learning; ensemble methods; random forest; feature selection; feature importance; classification; cybersecurity; network security.
An ECC-based Enhanced and Secured Authentication Protocol for IoT and Cloud Server
by Bhanu Chander, Kumaravelan Gopalakrishnan
Abstract: Due to the tremendous development in various domains, both internet of things (IoTs) and cloud-based computing are recognised as emerged technologies of the 20th century. IoT collection of heterogeneous interconnected embedded devices, then cloud computing provides the possible infrastructure which could be used at any-time and from anywhere. In recent times, due to their intelligence, competence ability to provide remote services, IoT and cloud computing employed into various real-life domains. The integration of IoT with the cloud infrastructure may lead to security issues. Moreover, these IoT-based devices prerequisite to link thru high-resource hubs such as cloud server (CS) for communication and resource sharing. Such combination comprises security hazards and injects dishonest data that principal to system fiasco. Especially, IoT policies accommodate on a public web links which perturbs the secrecy of trustworthy operators. Hence, this article planned an authentication idea on ECC for IoT-based embedded devices with CS. We discover the fuzzy structure facility for channel non-reciprocal, then ECC with one-way hash functions for secure validation. In conclusion, we scrutinise the planned procedure thru the AVISPA and related the resource constraints costs with existing authentication procedures.
Keywords: internet of things; IoTs; cloud server; one-way hash function; embedded device; fuzzy logic; ECC; security and privacy.
Survey on Wait Free Consensus Protocol in Distributed Systems
by Radha Rani, Dharmendra Prasad Mahato
Abstract: Computer applications are transitioning from centralised to decentralised automation in the modern industrial era. The consensus algorithm is a critical component in decentralised applications. Wait-free consensus is an unresolved issue in asynchronous systems. Deterministic protocols are known to be incapable of solving the wait-free consensus problem. Wait-free consensus protocol implementation occurs when all processors complete their predefined steps regardless of the execution speed of the other processors. Many randomised algorithms for wait-free consensus have been proposed, but no deterministic algorithm is possible. In this paper, we present a survey of wait-free consensus algorithms that have been studied and are currently being used in some well-known applications. There is also discussion of open issues and challenges in deploying various consensus mechanisms. This survey will provide a detailed understanding of the wait-free consensus protocol and will aid in the direction of research in the field of designing consensus algorithms.
Keywords: distributed systems; consensus; asynchronous model; fault tolerance; consensus; message passing.
Deep Learning based on Multimedia Encoding to Enhance of Video Quality
by Nagendra Panini Challa, Shanmuganathan C, Shobana M, Ch.Venkata Sasi Deepthi, Bharathiraja N
Abstract: Over the years, video compression has become a key method of sharing information and communicating in the world of media. The future image quality of the ISO standards groups will expect high quality images. Also, visual video codecs (VVC) have optimised encoding decisions that use the same average square error, or total square difference, for decades. At the same time, the sudden increase in deep learning (DL) methods raises the issue of whether DL could actually profoundly change the way the clip was programmed. We developed a new visual quality measurement (VQM) to find ways to make it better. The proposed approach could significantly reduce the encoding compute load while maintaining almost the same enhanced rate distortion optimisation (ERDO) efficiency as the previous encoder, based on experimental measurements. To further improve efficiency, the proposed methodology could be combined with fast motion search algorithms and filtration methods.
Keywords: multimedia; deep learning; visual quality measurement; VQM; quality; encoding optimiser; visual video codecs; VVC; enhanced rate distortion optimisation; ERDO.
Effect of Different Attack Strategies on Controllability Robustness of Directed Complex Networks
by Peng Geng, Annan Yang, Lai Wei, Rui Chen, Ziyu Pan
Abstract: This article summarises the controllability robustness of various directed complex networks under different attack strategies. Based on the theory of node-degree, edge-degree, node-betweenness and edge-betweenness, the controllability robustness evaluation criterion for complex networks is proposed. Considering directed complex networks, we describe the construction of seven network models (RGN, SFN, OLN, MCN, QSN, RTN and RRN). Based on six attack strategies (NABR, NABB, NABD, EABR, EABB and EABD), we conduct simulated attacks on the above seven network models and analyse the results. All the attacks are classified into node-based and edge-based. Through simulation experiments, we can see that under the same network environment, the damage caused by the betweenness-based attack to the network is greater than that of the degree-based attack. The controllability robustness of scale-free network and onion-like network is almost the same regardless of the attack. Compared with other networks, random rectangle network has the best controllability robustness. Therefore, the simulation results can also draw the conclusion that the multi-ring structure is helpful to improve the controllability robustness.
Keywords: directed complex networks; controllability robustness; attack strategy.
Internet of Things Producer Mobility Management in Named Data Networks: A Survey, Outlook, and Open Issues
by Ahmad Abrar, Suki Arif, Khuzairi Bin Mohd Zaini
Abstract: The sustainability of the current internet architecture is a challenging task under consideration of tremendous and dramatic growth of smart mobile devices and content demands. The mobility support plays an essential role in growing trend of smart mobile devices, making mobility more intriguing and widely discussed domain. Therefore, this study provides broad discussion on mobility support. Also, describes consumer and producer mobility in internet of things-based wireless body area networks (IoT-WBANs) using named data networks (NDN). NDN is a significant paradigm for data distribution in IoT networks, but it confronts considerable challenges in terms of producer mobility. We discuss producer mobility extensively and categories into different approaches based on their characteristics and functionalities. Moreover, we constructed a clear research roadmap regarding IoT in NDN and critically reviewed each approach to identify the various research concerns and long-standing challenges. Furthermore, we investigated various challenges in IoT-WBANs caused by producer mobility and suggested intuitive solutions.
Keywords: named data networking; internet of things; IoT; wireless body area networks; WBANs; information centric networks; consumer mobility; producer mobility.
ENERGY EFFICIENT CONGESTION CONTROL IN WIRELESS SENSOR NETWORKS USING META-HEURISTIC ALGORITHMS
by Abdul Ali, M. Vadivel
Abstract: Wireless sensor network (WSN) is an infrastructure-less wireless network that is used to monitor the physical or environmental conditions. In WSN, energy is the main constraint in congestion control that decreases the networks lifetime. This papers main goal is to propose a novel clustering-based routing protocol to mitigate the congestion issue in the network and to maximise the lifespan of WSN. To control the congestion problem in WSN, an energy-efficient ultra scalable ensemble clustering is introduced in this paper. In addition, the flamingo search algorithm-based fuzzy inference system is applied for cluster head (CH) selection. Rat swarm optimisation is used to select the route between CH and BS. The experimental results demonstrated that the proposed methodology gives better results than the existing techniques and it enhances the energy efficiency of WSN. The residual energy of the proposed approach (5 J at 2,000 rounds) is greater than the existing methods.
Keywords: wireless sensor network; WSN; ultra-scalable ensemble clustering; flamingo search algorithm; FSA; rat swarm optimisation; RSO; cluster head selection.
In search of clusters Based routing Protocol for WSN using consensus algorithm.
by Vikram Dhiman, Manoj Kumar
Abstract: The wireless sensor network (WSN) is a resource-constrained network type with two important challenges: energy consumption and network management. The goal of our design is to research and develop the WSN routing protocol, which is based on clusters and Open Flow controllers, in order to improve overall network performance. The results reveal that the proposed protocol is extensible and substantially more streamlined than standard soft computing-based implementations. These developments enable sensor networks to have a higher level of flexibility and programmability when placing sensors to the target. Our simulation results demonstrate that SDN extends network life by 20% when compared to traditional clustered-based routing by sending fewer control messages and using less processing resources once the initial processing is complete. The controller performs route computations rather than usual routing sensor nodes, which saves energy and extends the network's total lifetime.
Keywords: software-defined networking; SDN; wireless sensor network; WSN; soft-computing; clustering; OpenFlow controller.
Leveraging capsule network to learn content text for collaborative filtering
by Ji Li, Suhua Wang
Abstract: At present, most of the building components, technologies and frameworks of deep learning are based on convolutional networks. However, some deep learning studies on image processing have shown that the capsule network can be more representational because it can capture various posture changes, including translation, rotation and scaling, and can remember the position relationship between parts. Despite the intriguing nature of capsule network and their potential to open up entirely new natural language processing architectures, little work has been done in this area. In this work, we use the capsule network to learn the content text of the item (such as the plot text of the movie or the description document of the product), so as to obtain a better representation of the item and help achieve a more accurate recommendation. We proposed leveraging capsule network to learn content text for collaborative filtering (CCCF). This model combines capsule network and neural matrix factorisation to effectively model text data and user-item ratings. Experiments conducted from different perspectives on two popular datasets show that CCCF achieves good performance in common recommendation tasks, which proves the effectiveness of capsule network in recommendation.
Keywords: capsule network; content text; generalised matrix factorisation; collaborative filtering.
Cluster-Based Combined Hybrid Relay Vehicle Selection Approach for Improving Performance and Reliability in Vehicular Ad-hoc Network
by Puja Padiya, Amarsinh V. Vidhate, Ramesh Vasappanavara
Abstract: One of the main objectives of smart transport systems is to improve road safety. Ad hoc vehicle network uses vehicle accident alert systems that broadcast crashes to vehicles. Reliable and timely receipts of messages are two of the main intents for the creation of VANET security alert protocols. Existing techniques are hard to meet both objectives at the same time. With proposed suggestion, the vehicle transmits by employing the combined cluster and broadcast technique all the accident warning messages. The multi-hop message dissemination in VANET involving the selection process of relay vehicles in our proposed combined approach helps to improve performance in terms of reduced collision and reliability in terms of improved reachability and thus helps faster dissemination of emergency messages which in turn reduces the further impact of emergency situations.
Keywords: vehicular communications; reliability; relay vehicle; emergency messages; ad hoc networks; safety; broadcast; cluster; message dissemination; vehicular ad hoc network.
Time and position aware resource search algorithm for the mobile peer-to-peer network using ant colony optimisation
by Dharmendra Kumar, Ajay Kumar Dubey, Mayank Pandey
Abstract: The progressive growth of wireless mechanisms and the wide popularity of intelligent devices are attracting particular attention to peer-to-peer mobile ad hoc networks (P2P MANET). Increasing the efficiency to search the resources became an important research topic in the P2P MANETs. Most existing research gives more emphasis on position-based clustering techniques and does not give preference to time. Due to not giving any preference to time, the existing approaches are not so appropriate to improve the efficiency to search the resources in P2P MANETs. This paper suggests a navel resource search approach that first uses position-aware peer clustering strategy. Then, using an ant colony optimisation method with pheromones, it chooses the preeminent neighbour's peer that is of time-aware neighbouring peer's resource preferences and the time-aware neighbouring peer's availability. Based on the results of the experiment, this approach outweighs other approaches in terms of search delay, overhead traffic, and search success rate.
Keywords: resource discovery mechanism; peers resource preferences; ant colony optimisation; position aware; time aware; mobile P2P ad hoc network; P2P MANET.
Accuracy evaluation of supervised machine learning classification models for wireless network traffic
by Elans Grabs, Ernests Petersons, Dmitry Efrosinin, Aleksandrs Ipatovs, Janis Kluga, Valentin Sturm
Abstract: The article contains results of training and testing machine learning models with captured network traffic data. The main goal is to perform classification of video traffic in computer networks. Multiple performance metrics have been evaluated for commonly used classic supervised machine learning algorithms, as well as more advanced convolutional neural network model (for comparison). The article describes in detail the experimental setup, traffic pre-processing procedure, features extraction with different traffic window length and model parameters for training/testing. The article provides some experimental results in the form of tables and 3D surface plots. The conclusion of the article summarises the main findings and outlines the future study directions.
Keywords: accuracy; classification models; features extraction; network traffic; performance metrics; statistical parameters; supervised machine learning; traffic intensity; window length; wireless networks; convolutional neural network; CNN.
Load balancing algorithms with cluster in cloud environment
by S.B. Kshama, K.R. Shobha
Abstract: Load balancing is one of the important aspects of cloud computing. Its main goal is to improve system performance and to reduce its cost. Cloud computing has a dedicated load balancer to attain this. Sometimes, heavy traffic may overwhelm the load balancer and disrupt it to achieve its goal. The repeated disruption has an impact on customer service and results in poor performance. The clustering technique gives a solution to this by reducing the burden on a single machine and improves system performance. In this paper, an analysis has been made to test the performance of the load balancer with and without the cluster. The previously proposed capacity-based load balancing (CBLB) and artificial bee colony_CBLB algorithms are considered during this experiment. The results are compared by applying clusters on existing load balancing algorithms. In addition to this, the performance of the ABC_CBLB with cluster is also compared with two existing cluster load balancing algorithms. The results of CBLB and ABC with cluster are better than the other comparing algorithms.
Keywords: load balancing; cloud computing; cluster; capacity-based load balancing; CBLB; artificial bee colony; ABC; virtual machine; cloudlets; K-mean; utilisation capacity; nature inspired algorithm.
Utilisation-aware VM placement policy for workload consolidation in cloud data centres
by Dipak Dabhi, Devendra Thakor
Abstract: In recent years, the demand for cloud services has risen. Data centres must have a growing number of servers to accommodate rising demand for cloud services, and data centres consume a lot of energy. Virtual machine consolidation (VMC) is a strategy for reducing energy consumption in data centres by shutting down underutilised servers while maintaining service levels (SLA). The VMC process is separated into four policies: overloaded host detection, underloaded host detection, virtual machine selection, and virtual machine deployment. The utilisation-aware VM placement (UAVMP) technique is presented in this research work, which efficiently selects the destination host for VMs migrating from overload/underload hosts based on the host's utilisation and resource skewness. The performance is assessed using the CloudSim simulator. When we compare UAVMP results with power-aware best fit decreasing (PABFD), modified best fit decreasing (MBFD), first fit (FF) and least fit (LF), we find that UAVMP outperforms all.
Keywords: cloud computing; VM consolidation; quality of service; QoS; service level agreement; SLA; VM selection; overload host detection; VM placement; underload host detection.
Energy consumption profiles of wireless sensor nodes in smart cities using CupCarbon (V 5.0) simulator
by Hanshita Prabhakar, Asna Furqan
Abstract: In the development of smart cities, smart devices are integrated with sensors and embedded systems to provide innovative services which enhance the internet of things (IoT); also, the power consumed by these devices for specific operations will be a primary concern. Wireless sensor network (WSN) consists of enormous numbers of sensor nodes densely deployed over smart city sensor nodes that are efficient in collaborating and computing the condition of their surrounding environments. The proposed work of this paper is analysing how the different profiles of wireless sensor networks for intelligent cities consume energy as routing, communication of messages between two sensors, broadcasting of transmissions, and mobility is shown by the graph of energy consumption in joule (J) versus simulation time in seconds (S). We have performed and analysed the results on the network simulator called CupCarbon simulator. The novelty of this paper is that we could examine and estimate the errors while the transmission and reception happen in between the sensor nodes with the help of the console output messages.
Keywords: smart city; CupCarbon; V 5.0 simulator; wireless sensor network; internet of things; energy consumption.