Mihai Mocanu

QoS optimization in congested multimedia networks

In multimedia networks, maintaining the agreed QoS (Quality of Service) level is fundamental. VoIP (Voice over IP) and other multimedia applications such as video conferencing, video-on-demand, media streaming, data traffic etc. impose strict requirements and arise various factors that may alter the QoS in the network. The QoS may easily be affected by delaying or dropping packets, as a consequence of parameters such as latency, jitter, bandwidth and packet loss. The wide range of multimedia applications on the current market, impose various and strict requirements for multimedia networks to deal with. For better managing such requirements and improving the service process, it is worth classifying traffic with respect to the bit rate. According to the service requirements of source applications and to the arrival characteristics, the traffic on high-performance networks can be divided into four categories: Constant Bit Rate, Variable Bit Rate, Available Bit Rate and Unspecified Bit Rate. The paper presents the results of a performance evaluation experiment conducted within our Competence Center for Network Management. The performance measurements involve several scenarios which combine different traffic types, queuing mechanisms and resource reservation protocols. The goal of the experiment is to identify the most suitable scenario to provide qualitative services and to efficiently utilize resource given the conditions imposed by each traffic type.

Building an advanced dense classifier

Classification of items (i.e. students, genes, etc.) on data sources provided by various software systems represents an important task that needs to be performed for many reasons related to the business goals of the system. This paper presents several approaches that need to be taken into consideration by a data analysis system designer who aims to obtain an advanced classifier that implements several extra functionalities. The engine presented in this paper is preliminary tested on real data obtained from Tesys on-line educational environment in an attempt to determine the most suitable tutors for currently existing students. Our goal is to build a Decision Tree classifier that accommodates data. This new data structure extends the functionality of a Decision Tree and is called DenseJ48. This new classifier implements efficiently several extra functionalities besides the core ones that may be used when dealing with data.

Fluoroscopy servoing using translation/rotation decoupling in an A/P view

This paper presents a fluoroscopy servoing algorithm for automatic alignment of a needle using a medical robot during interventional procedures. The goal of this work is to provide physicians with assistance in needle alignment during minimally invasive procedures under fluoroscopy imaging. This may also help reduce radiation exposure for the physician and provide more accurate targeting of internal anatomy. The paper presents the overall concept and describes our implementation along with the initial laboratory results and studies in the interventional suite. The algorithm is based on a single anterior/posterior fluoroscopic image. Future work will be aimed at demonstrating the clinical feasibility of the method.

Enhancing QoS/QoE in multimedia networks

In a multimedia context, QoS (Quality of Service) is a measure of success. It represents the guarantee of successfully delivering packets, without delaying or dropping any, therefore, the guarantee of providing high quality experience for end-users. Depending on the content being delivered, there exist various parameters which need to be considered when designing QoS architectures, such as latency, jitter, bandwidth and packet loss. There is a direct and high impact of such parameters on voice quality, video streaming, data traffic etc. On the other hand, high-performance networks are required to support different traffic types, alternating from continuous data streams for real-time communication to burst traffic for best-effort communication. The traffic on high-performance networks can be divided into four categories: Constant Bit Rate, Variable Bit Rate, Available Bit Rate and Unspecified Bit Rate, according to the service requirements of source applications and to the arrival characteristics. The aim of the research work presented in this paper is to identify the most appropriate QoS-related techniques to be applied in order to provide qualitative services and to ensure the efficiency of resource consumption, under the constraints imposed by each traffic type. The research results are based on performance measurements conducted within our Competence Center for Network Management, taking into consideration various traffic types, queuing mechanisms and resource reservation protocols.

Needle targeting under C-arm fluoroscopy servoing

This paper describes a method for translational and orientational alignment of a robotic needle driver based on image servoing and x-ray fluoroscopy. The translational process works by segmenting the needle in a frame-grabbed fluoroscopic image and then commanding the robot to automatically move the needle tip to the skin entry point. The orientational alignment is then completed based on five different positions of the needle tip. Previously reported fluoroscopy servoing methods use complex robot-image registration algorithms, fiducial markers, and two or more dissimilar views that included moving the fluoroscope. Our method aligns the needle using one setting of the fluoroscope so that it does not need to be moved during the alignment process. Sample results from both the translational and orientational steps are included.

Generating Alerts for Drops in Student Activity Levels in a Social Learning Environment

Monitoring students’ activity in a social learning environment is an important issue both for students and teachers. Providing learners with notifications whenever their activity level drops has the potential to increase their motivation and engagement. This paper tackles the issue of accurately pointing when a student has a drop or an increase in activity in the context of a social learning environment. We designed and implemented a data analysis framework which generates statistical dashboards based on aggregated student activity on three social media tools (blog, wiki and microblogging tool); alerts are subsequently issued in case of a significant decrease in activity. Experimental results obtained on student data collected over the course of five years reveal a pattern regarding the average number of generated alerts. Therefore our system can be successfully used by the instructor to easily configure the number of alerts issued to the students.

Design of an Advanced Smart Forum for Tesys e-Learning Platform

This paper presents an application of Intelligent Data Analysis techniques in the area of online educational environments and more exactly, the discussion forums within them. The research area is also referred as Educational Data Mining and have many tools and techniques already developed. This work concentrates on the improvement that can be provided by the design and implementation of a forum that has “smart” capabilities and aims to be proactive to the user’s needs. The main issues addressed are the interaction design, student’s academic performance and the achievement of better models by completing the already gathered data with the logged data offered by the forum. We present here three methods that can solve the above mentioned issues: recommending subjects of interest, computing trends and offering smart alerts for users that are at risk for academic failure. Every method represents a tool that will be integrated in the forum and will take benefit from the extra logged data.

Improving Peer-to-Peer Communication in e-Learning by Development of an Advanced Messaging System

This chapter presents an advanced messaging system, whose goal is to improve the peer-to-peer communication in e-Learning. The improvement is based on the ability of the developed system to produce information that is highly related to the informational needs of the person, who accesses it. The system is an intelligent one because it integrates a classification procedure for retrieval of the messages that have a high potential of being interesting to peers. It uses as input data activity logs obtained by monitoring the communication that takes place within the e-Learning platform. The main data analysis goal is to create a user’s model, for which derived classes are in close relation with specific set of messages. The outcome is in the form of a tool that allows learners to receive a set of recommended messages that is highly to be interesting for them. The tool analyzes the user’s features, classifies them and according with the class label obtained set of messages. The tool also acts as a message indexing system by storing messages in correlation with labels assigned to learners. A classical classification procedure is used for obtaining a labeling. The data used to train the classifier is gathered from the on-line educational environment and contains all the necessary information (i.e., the features) regarding the activities performed by learners on the platform. The high quality of the system is based also on a text-mining module that uses stemming, annotation, and concept detection for a proper assignment of messages to learner’s labels.

Coronary artery segmentation in cardiac angiography images based on a hybrid approach

A reduction of the size of the coronary artery openings, estimated by the visual inspection of the X-ray images of the transient radio contrast distribution during conventional angiography, can be used to formulate predictions about clinical symptoms and dynamic reductions in coronary blood flowing. Even with introduction of new noninvasive imaging modalities such as Cardiac CT and MRI, coronary angiography remains the gold standard for detecting ischemic coronary disease.

Obtaining Knowledge Using Educational Data Mining

Obtaining knowledge is one of the most important tasks in currently developed systems. Regarding this open problem, educational data mining is one of the areas that gather many efforts. This paper presents a custom methodology of obtaining knowledge about learners. The obtained knowledge is based on activity performed within e-Learning environment. The logged activity regards data about how the student answered to test questions. The main task of the procedure regards clustering students such that different pedagogical approach will be used for each cluster. K-means algorithm has been used as clustering method. The final goal is to create a model of analysis which may conclude whether or not an e-Learning platform is capable of classifying students depending on accumulated knowledge.