Carlos Renato Lisboa Francês

Strategies for improving the modeling and interpretability of Bayesian networks

One of the main factors for the knowledge discovery success is related to the comprehensibility of the patterns discovered by applying data mining techniques. Amongst which we can point out the Bayesian networks as one of the most prominent when considering the easiness of knowledge interpretation achieved. Bayesian networks, however, present limitations and disadvantages regarding their use and applicability. This paper presents an extension for the improvement of Bayesian networks, treating aspects such as performance, as well as interpretability and use of their results; incorporating genetic algorithms in the model, multivariate regression for structure learning and temporal aspects using Markov chains.

Performance evaluation based on system modeling using Statecharts extensions

Abstract This paper presents two extensions for Statecharts: the Stochastic Statecharts, which use the original statecharts notation with a minor modification in the formal semantics and the Queuing Statecharts, which do not follow the pure Statecharts notation, but a join between Statecharts and queuing network representations. Some basic elements of Statecharts are redefined such as events and conditions, besides some concepts referring to the dynamic system behavior. The specification approaches show the basic behavior of a generic queuing system by means of templates and standard events. It is presented the PerformCharts, a new simulation environment based on Statecharts specification, which allows model solution using either Markov chains or the Network Simulator (NS).

Technical Communication: Performance analysis of multi-service wireless network: An approach integrating CAC, scheduling, and buffer management

Traffic management (TM) mechanisms such as Call Admission Control (CAC), Scheduling, and Buffer Management (BM) play a key role in the design of multi-service wireless network by providing service differentiation from diverse applications and assigning the network resources (radio channels or buffer) according to the quality of service (QoS) requirements of each service class. We propose in this work two new models that integrate CAC, Scheduling, and BM in the design of multi-service wireless network. By presenting their Markovian models and their performance metrics, we investigate their respective design tradeoffs and compare their performance.

SensorBus: a middleware model for wireless sensor networks

The use of middleware eases the development of distributed applications by abstracting the intricacies (communication and coordination among software components) of the distributed network environment. In wireless sensor networks, this is even trickier because of their specific issues such as addressing, mobility, number of sensors and energy-limited nodes. This paper describes SensorBus, a message-oriented middleware (MOM) model for wireless sensor networks based on the publish-subscribe paradigm and that allows the free exchange of the communication mechanism among sensor nodes allowing as result the capability of using more than one communication mechanism to address the requirements of larger number of applications. We intend to provide a platform which addresses the main characteristics of wireless sensor networks and also allows the development of energy-efficient applications. SensorBus incorporates constraint and query languages which will aid the development of interactive applications. It intends with the utilization of filters reduces data movement minimizing the energy consumption of nodes.

A New Parallel Approach for 3D Ray-Tracing Techniques in the Radio Propagation Prediction

A new computational parallel model based on 3D ray-tracing for radio propagation prediction is presented. This approach considers that the main tasks in a 3D ray-tracing technique can be evaluated in an independent and/or parallel way. The workload distribution among the participant nodes of the parallel architecture (cluster of PCs) is performed through a random assignment of the initial rays and the field points for them. Simulations are realized in order to validate and evaluate the performance of the proposed model. The presented results show that the scalability of the model is obtained naturally due to independence of the involved processes. The efficiency of the model presents behavior above the ideal for cases with ostensible processing of rays. These characteristics favor to the increase of the prediction precision through the increase of the density of launched rays and the possibility of incorporation of new propagation mechanisms.

Decision support in power systems based on load forecasting models and influence analysis of climatic and socio-economic factors

This paper presents a decision support system for power load forecast and the learning of influence patterns of the socio-economic and climatic factors on the power consumption based on mathematical and computational intelligenge methods, with the purpose of defining the future power consumption of a given region, as well as to provide a mean for the analysis of correlations between the power consumption and these factors. Here we use a linear modelo of regression for the forecasting, also presenting a comparative analysis with neural networks, to prove its efectiveness; and also Bayesian networks for the learning of causal relationships from the data.

The Influence of Interference Networks in QoS Parameters in a WLAN 802.11g : a Bayesian Approach

In spite of the significant increase of the use of Wireless Local Area Network (WLAN) experienced in the last years, design aspects and capacity planning of the network are still systematically neglected during the network implementation. For instance, to determine the location of the access point (AP), important factors of the environment are not considered in the project. These factors become more important when several APs are installed, sometimes without a frequency planning, to cover a unique building. Faults such as these can cause interference among the cells generated by each AP. Therefore, the network will not obtain the QoS patterns required for each service. This paper proposes a strategy to determine how much a given network can affect the QoS parameters of another network, by interference. In order to achieve this, a measurement campaign was carried out in two stages: firstly with a single AP and later with two APs using the same channel. A VoIP application was used in the experiment and a protocol analyzer collected the QoS metrics. In each stage 46 points were measured , that are insufficient for statistically characterize the environment. For expanding this data, an Artificial Neural Network (ANN) was used. After the measurement, an analysis of the results and a set of inferences were made by using Bayesian Networks, whose inputs were the experimental data, i.e., QoS metrics like throughput, delay, jitter, packet loss, PMOS and physical metrics like power and distance.

The influence of interference networks in QoS parameters in a WLAN 802.11g environment

This paper proposes a strategy to determine how much a given network can affect the QoS parameters of another, by interference. In order to achieve this, a measurement campaign was carried out in two stages: firstly with a single AP and later with two APs separated by a distance less than three meters, using the same channel. After the measurement, an analysis of the results and a set of inferences were made by using Bayesian Networks, whose inputs were the experimental data, i.e. QoS metrics such as: throughput, jitter, packet loss, PMOS and physical metrics like power and distance.

Computational parallelization strategy applied in full 3D ray-tracing wireless channel modeling

This paper presents a computational parallelization strategy applied in propagation models based on 3D ray-tracing techniques. This approach considers that the rays are independent from each other, what allows a uniform division of the tasks by equal and random distribution of rays among the parallel computer nodes. The strategy efficiency is proved by simulation where the results are discussed.

Restoration in optical cloud networks with relocation and services differentiation

Optical cloud networks allow for the integrated management of both optical and IT resources. In this paradigm, cloud services can be provisioned in an anycast fashion; i.e., only the source node asking for a service is specified, while it is up to the cloud control/management system to select the most suitable destination data center (DC) node. During the cloud service provisioning process, resiliency is crucial in order to guarantee continuous network operations also in the presence of failures. On the one hand, a survivability strategy needs to be able to meet the availability requirements of each specific cloud service, while on the other hand it must be efficient in using backup resources. This paper proposes a restoration-based survivability strategy, which combines the benefits of both cloud service relocation and service differentiation concepts. The former is used to enhance the restorability performance (i.e., the percentage of successfully restored cloud services) offered by restoration, while the latter ensures that critical services are given the proper consideration while backup resources are assigned. The paper proposes both an integer linear programming (ILP) formulation, which guarantees optimal results, and a heuristic, which trades the optimality of the solution achieved by the ILP for faster processing times. Simulation results show that the average service availability and restorability performance obtained by both the ILP and the heuristic are very close to that achievable using a protection-based strategy, but with the inherent benefit, in terms of efficient use of resources, offered by a restoration-based approach.