JMM Abstracts 

Vol.13 No.3&4 Dec 30, 2017

Next Generation Networks (NGN) and Services


Editorial (181-182)
Brahim Ouhbi


Enhanced QoS Management SDN-Based in IMS with QoE Evaluation (183-196)
Sara Khairi, Brahim Raouyane, and Mostafa Bellfakih
The increase of mobile subscribers' requests and the explosion of multimedia services such as provided by the IP Multimedia Subsystem IMS) are making QoS management more and more complex. In fact, with the emergence of the Software Defined Network (SDN) paradigm which enables dynamic configuration of network resources and flexible QoS management, Next Generation Networks (NGN) have found necessary to integrate this new concept. The aim of this paper is twofold. First, to present a new architecture SDN-based for Next Generation Networks (NGN) to enhance the QoS management. The architecture is implemented and evaluated for a Video on Demand (VoD) in IMS network. Second, to demonstrate the implementation of the proposed architecture by testing the Quality of Experience (QoE) which is evaluated in terms of the Video Mean Opinion Score (VMOS).

A Novel Anomaly Intrusion Detection Based on SMO Optimized by PSO with Pre-Processing of Data Set (197-209)
Mehdi Moukhafi, Khalid El yassini, and Seddik Bri
Current IDSs are mainly based on techniques based on heuristic rules called signatures to detect intrusions in a network environment. These approaches based signature could only detect a known attacks and referenced above. Since there is no signature for new attacks, other approaches must be taken in consideration, such as algorithms learning machine. However, the major problem of IDSs based on learning machine is the high rate of false positives. This study proposes a novel method of intrusion detection based on pre-processing of training data and a combination PSO (Particle Swarm Optimization) -SMO (Sequential minimal optimization) to develop a model for intrusion detection system. The simulation results show a significant improvement in performances, all tests were realized with the kdd99 data set. compared with other methods based on the same dataset, the proposed model shows high detection performances.

GENAUM:  New Semantic Distributed Search Engine (210-221)
Ichrak Saif, Abdelaziz Sdigui Doukkali, Adil Enaanai, and El Habib Benlahmar
The rapid development of services based on distributed architectures is now emerging as important items that transform mode of communication, and the exponential growth of the Web makes a strong pressure on technologies, for a regular improvement of performance, so it’s irresistible to use distributed architectures and techniques for the search and information retrieval on the Web, to provide more relevant search result, in minimum possible time. This paper discuss some solutions researchers are working on, to make search engines more faster and more intelligent, specifically by considering the semantic context of users and documents, and the use of distributed architectures. This paper also presents the overall architecture of GENAUM; the collaborative, semantic and distributed search engine, based on a network of agents, which is the core part of the system. The functionality of GENAUM is spread across multiple agents, to fulfill user’s performance expectations. At the end of this paper, some preliminary experimental results are presented, that attempts to test the user modeling process of GENAUM, using reference ontology.


Coupling and Annotated Corpus and a Lexicon for Amazigh POS Tagging (222-232)
Samir Amri, Lahbib Zenkouar, and Mohamed Outahajala
This paper investigates how to best couple hand-annotated data with information extracted from an external lexical resource to improve part-of-speech tagging performance. Focusing mostly on Amazigh tagging, we introduce a decision tree and Markov model using TreeTagger system. This system gives 92.3 % accuracy on the Amazigh corpus, an error reduction of 15 % (18.45 % on unknown words) over the same tagger without lexical information. We perform a series of experiments that help understanding how this lexical information helps improving tagging accuracy. We also conduct experiments on datasets and lexicons of varying sizes in order to assess the best tradeoff between annotating data versus developing a lexicon. We find that the use of a lexicon improves the quality of the tagger at any stage of development of either resource, and that for fixed performance levels the availability of the full lexicon consistently reduces the need for supervised data.

Sentiment Classification of Arabic Tweets: A Supervised Approach (233-243)
Naaima Boudad, Rdouan Faizi, Richard O. Haj Thami, Raddouane Chiheb

Social media platforms have proven to be a powerful source of opinion sharing. Thus, mining and analyzing these opinions has an important role in decision-making and product benchmarking. However, the manual processing of the huge amount of content that these web-based applications host is an arduous task. This has led to the emergence of a new field of research known as Sentiment Analysis. In this respect, our objective in this work is to investigate sentiment classification in Arabic tweets using machine learning. Three classifiers namely Na´ve Bayes, Support Vector Machine and K-Nearest Neighbor were evaluated on an in-house developed dataset using different features. A comparison of these classifiers has revealed that Support Vector Machine outperforms others classifiers and achieves a 78% accuracy rate.


A Map-Matching Based Approach to Compute and Modelize NLOS and Multipath Errors for GNSS Positioning in Hard Areas (256-269)
Bassma Guermah, Tayeb Ssadiki, Hassan el Ghazi, Serge Reboul, and Esmail Ahouzi
In Global Navigation Satellite systems (GNSS), the performances of classical localization methods show a significant degradation in constrained environments (urban and indoor environments), due to Non-Line-of-Sight(NLOS) and Multipath phenomena affecting GNSS signal. In order to improve positioning accuracy in hard environment, this paper aims to propose an approach to compute and adapt the NLOS and Multipath error model to GNSS signal reception conditions. The approach aims firstly to propose a Map-Matching based-technique to compute Multipath and NLOS errors in real time positioning, secondly, to test adequacy of these errors with the most used models in the literature and finally to model the Multipath and NLOS errors using Gaussian mixture noise. As a result, we have shown that a Gaussian, Rayleigh and Uniform model were not be able to model effectively Multipath and NLOS errors and we have demonstrated that a Gaussian mixture model can approximate these errors and improve positioning accuracy in urban environment.

Back to JMM Online Front Page