Abstract
Introduction: Self-reported height and weight data have been used in several studies with the purpose of determining the prevalence of overweight and obesity. Despite being a simple methodology, little information exists about the reliability of these measures, namely, in university students. The objective of this study was to determine the sensitivity and specificity of self-reported body mass index (BMI) to evaluate the prevalence of overweight and obesity in university students. Methods: In a convenience sample of 380 university students (226 women and 154 men), weight and height were obtained by self-reported measures and anthropometrie assessment according to international standards methodology (objective). BMI was calculated from self-reported and direct measures. Results: The discrepancy between objective and self-reported weight was not significative. For height, this discrepancy was significantly different in women, in men, and between genders. The difference between BMI values was significantly different in women (0,8 ± 1,1 kg/m2), in men (0,4 ± 1,1 kg /m2) and between genders. Concerning overweight and obesity, according to the objective BMI, the sensitivity was only 50% in women, and 70% in men, while the specificity was 99% in women and 98% in men. Conclusion: Our results show a poor sensitivity of self-reported weight and height data, to estimate overweight and obesity, thus, this method might not be reliable for studies of prevalence of obesity in this population.

Abstract
The classical TCP/IP layered protocol architecture is beginning to show signs of age. In order to cope with problems such as the poor performance of wireless links and mobile terminals, including the high error rate of wireless network interfaces, power saving requirements, quality of service, and an increasingly dynamic network environment, a protocol architecture that considers cross-layer interactions seems to be required. This article describes a framework for further enhancements of the traditional IP-based protocol stack to meet current and future requirements. Known problems associated with the strictly layered protocol architecture are summarized and classified, and a first, solution involving cross-layer design is proposed.

Abstract
This paper demonstrates the application of Markov chain models to valuate generation assets within deregulated electricity markets. A new framework for modeling electricity markets with Markov chain model is proposed. The advantage of the Markov chain model is that it deploys fundamental approaches to identify the key economic forces underlying the electricity markets such as demand on electricity and supplied online generation capacity. Based on this new model, real option calculations are used to valuate generation assets. Markov chain model is combined with binomial tree to approximate the stochastic movement of prices on both electric energy and ancillary services, which are driven by the market forces. A detailed example is presented. This method is shown to provide optimal operation policies and market values of generation assets. This method also provides capability to analyze the impacts of demand growth patterns, competition strategies of competitors and other key economic forces.

Abstract
ReDiFlex is a system that supports the development of applications that use dynamically and partially-reconfigurable hardware. The hardware functionality is specified by the flow of data between mutable operators. The system automatically creates the physical implementation after partitioning the model to fit the hardware constraints; during application execution new computation-dependent partial configurations can be created. ReDiFlex provides run-time support for reconfiguration and data transfer scheduling.

Abstract
In this paper we describe an application of recommender systems to team building in a company or organization. The recommender system uses a collaborative filtering model based approach. Recommender models are sets of association rules extracted from the activity log of employees assigned to projects or tasks. Recommendation is performed at two levels: first by recommending a single team element given a partially built team; and second by recommending changes to a completed team. The methodology is applied to a case study with real data. The results are evaluated through experimental tests and one survey to potential users.

Abstract