Volume 10, Issue 6, December 2019

Efficient Normalization of Features Extracted from Vocal Signal for Support Vector Machine
Pages: 1-5 (5) | [Full Text] PDF (343K)
Boutkhil SIDAOUI and Kaddour SADOUNI
Mathematics and Computer Science Department, University Center Salhi Ahmed-Naâma, Algeria
Computer Science Department, University of Science and Technology Mohamed Boudiaf-Oran, Algeria

Abstract -
In this paper we introduce and investigate the performance of an efficient normalization of features extracted from signals of vowels, into fixed-length features vectors, in the context of vowels classification by kernel machine learning methods. The proposed models capture the speech signal and extract features vectors of the same size, over the entire vowel segment. In this work, we propose three approaches based on signal segmentation to windows, in the extraction of MFCC coefficients. In the aim to produces a compact size, yet wealthy and informative training dataset, also to give state of the art vowels classification results. To improve the performance of the proposals approaches, we are opted to use support vector machine method for vowels classification, due its performance in machine learning. Recognition rates of 58.76%, 58.85% and 61.19 %, on the 20 vowels of TIMIT corpus was achieved by the three approaches, for handling the multi category nature of vowels classification, using SVM method with one-versus-one and one-versus-all strategies. Encouraging results have been achieved with 20 vowels of TIMIT datasets.
 Index Terms - Features Extraction, MFCC, Classification and Support Vector Machine Multiclass
An Efficient Algorithm for Energy Management in Smart Home Including Renewable Energy
Pages: 6-11 (6) | [Full Text] PDF (486K)
Halim Halimi, Florim Idrizi, Festim Halili, Burim Baftijari
Department of IT, Faculty of Natural Sciences and Mathematics, University of Tetova, Tetova, R. N. Macedonia

Abstract -
Energy management strategies are instrumental in the performance and economy of smart homes integrating renewable energy and energy storage. Home energy management system (HEMS) in the smart home allows the customer to control, optimize and monitor the energy consumption and the energy conservation. In this paper, a brief overview on the architecture and functional modules of smart HEMS is presented. Then, the advanced HEMS infrastructures and home appliances in smart houses are thoroughly analyzed and reviewed. For management and monitoring the energy consumption of home appliances and lights is used ZigBee based energy measurement modules while for renewable energy is used a Power Line Communication (PLC) based renewable energy gateway. The home server monitors and controls the energy consumption and generation and controls the home energy use to reduce the energy cost. The remote energy management server aggregates the energy information from the home servers, compares them and creates statistical analysis information. We propose the control algorithm to efficiently manage the renewable energy and storage to minimize grid power costs at individual home. The proposed HEMS architecture is expected to optimize home energy use and result in home energy cost saving.
 Index Terms - Home Energy Management, Home Server and Renewable Energy
Use of 5G Network and Standardization of Frameworks to Enhance Security of IoT Systems
Pages: 12-19 (8) | [Full Text] PDF (645K)
Ajwang, Stephen Oloo
Department of Informatics and Information Science, School of Information Communication and Media Studies, Rongo University, P.O. Box 103-40404, Rongo

Abstract -
Internet of Things (IoT) has advanced over the years as a result of convergence of various technologies brought about by the increasing availability of embedded systems, pervasive computing and big data. IoT uses sensor technology and data analytics capabilities to transform data to actionable intelligence for business, transport, manufacturing, automotive and smart cities. Despite its widespread adoption, IoT devices have increasingly become a target for cyberattack because of the huge amount of valuable data they carry, lack of standard development framework, closed loop functioning ability, open and concurrent access of the system by data consumers and data controllers, and the inherent memory constraints. This paper therefore evaluates the strength of the available security protocols provided by AWS, Azure and Arm Mbed IoT frameworks and recommends enhancement of the security of the IoT devices through standardization of frameworks at technology and regulatory level to manage common security goals of confidentiality, integrity and availability. The paper also recommends the use of 5G 3GPP standards to secure IoT devices by leveraging its network capabilities such as ultra-low latency, reliability, reusability modularity and self-containment of network functions for consumer protection and network resilience.
 Index Terms - Internet of Things, Security, Frameworks, 5G Networks and Convergence
Machine Learning based Internet Traffic Classifiers: A Review
Pages: 20-25 (6) | [Full Text] PDF (445K)
J. Zaib Bhatti, A. Umar
Department of Computer Science and Engineering, University of Engineering and Technology

Abstract -
Internet traffic classification plays an important role in network management, bandwidth usage, network security and network speed. The amount and types of data flowing in the networks are increasing with increasing number of applications. So classification of such data is becoming very challenging. Many methodologies and techniques have been proposed to classify Internet traffic in various categories. These techniques and algorithms have different advantages and disadvantages over the others. Some classifiers use multi objective approach using minimized training data while other use multi-stage and multi clustering classifiers for accurate recognition of internet traffic. But, studies show that machine learning based classifiers have a good performance in comparison with other techniques. Best feature subset selection causes better accuracy and efficiency in machine learning algorithms. We discuss and analyze some commonly used approaches on the basis of their classification accuracy and efficiency for internet traffic of different applications.
 Index Terms - Ensemble Method, Internet Traffic Classification, Analyze and Machine Learning
Pages: 26-32 (7) | [Full Text] PDF (256K)
----

Abstract -
 Index Terms -