Abstract
Wireless sensor networks show great potential to successfully address the timeliness and energy-efficiency requirements of different cyber-physical system applications. Generally, these requirements span several layers of the stack and demand an on-line mechanism capable of efficiently tuning several parameters, in order to better support highly dynamic traffic characteristics. This work presents a cross-layer QoS management framework for ZigBee cluster-tree networks. The proposed framework carries out an on-line control of a set of parameters ranging from the MAC sub-layer to the network layer, improving the successful transmission probability and minimizing the memory requirements and queuing delays through an efficient bandwidth allocation at the network clusters. Through extensive simulations in a real datacenter monitoring application scenario, we show that the proposed framework improves the successful transmission probability by 10%, and reduces the end-to-end delay by 94%.

Abstract
A reflection configuration setup for long-period fiber gratings is presented. It permits to obtain a unique band with attenuation double than that obtained in transmission configuration, which is interesting for applications where this value is reduced (e. g., the mode transition phenomenon). The method is based on the deposition of a silver mirror at the end of the optical fiber, which permits to absorb the power transmitted through cladding modes and to avoid the generation of interferometric bands. The method also solves the requirement of a precise cleave or to polish the end of the grating, a drawback present in other publications. The versatility of the setup has been proved by application of the cladding etching technique until the attenuation band corresponding with the first guided mode in the cladding is visualized in an optical spectrum analyzer. The experimental results are supported by the numerical data obtained with a method based on the exact calculation of core and cladding modes and the utilization of coupled mode theory.

Abstract
Innovation has a cost. It is often the case that museums trying to innovate, simply cannot afford what it takes to build truly memorable exhibitions. Lack of human resources and high-tech equipment makes it hard to create quality contents that could be shown to the general public. Nevertheless, universities' museums are usually embedded in an environment that has the potential to provide all the tools and human resources required. The only thing needed is to establish a proper strategy and an interaction facilitator platform - U.OpenLab - which enables creating, building and sharing knowledge about the museums' collections and the academic population. This will make it easier to distribute the aforementioned knowledge to the general public, in a truly sustainable, systematic, integrated and articulated manner. In this paper we are going to present the University of Porto (U.Porto) OpenLab prototype that is being built as the stepping stone of this project, providing students with learning in a project environment.

Abstract
The ANTAREX 1 project aims at expressing the application selfadaptivity through a Domain Specific Language (DSL) and to runtime manage and autotune applications for green and heterogeneous High Performance Computing (HPC) systems up to Exascale. The DSL approach allows the definition of energy-efficiency, performance, and adaptivity strategies as well as their enforcement at runtime through application autotuning and resource and power management. We show through a mini-App extracted from one of the project application use cases some initial exploration of application precision tuning by means enabled by the DSL.

Abstract
Built-in sensors in most modern smartphones open multipleopportunities for novel context-aware applications. Although the HumanActivity Recognition field seized such opportunity, many challengesare yet to be addressed, such as the differences in movement by peopledoing the same activities. This paper exposes empirical research onOnline Semi-supervised Learning (OSSL), an under-explored incrementalapproach capable of adapting the classification model to the userby continuously updating it as data from the users own input signalsarrives. Ultimately, we achieved an average accuracy increase of 0.18percentage points (PP) resulting in a 82.76% accuracy model with NaiveBayes, 0.14 PP accuracy increase resulting in a 83.03% accuracy modelwith a Democratic Ensemble, and 0.08 PP accuracy increase resultingin a 84.63% accuracy model with a Confidence Ensemble. These modelscould detect 3 stationary activities, 3 active activities, and all transitionsbetween the stationary activities, totaling 12 distinct activities

Abstract
Smart Grid (SG) concept is defined as an electricity network operated intelligently to integrate the behavior and actions of all energy resources connected to the network to ensure efficient, sustainable, economic and secure supply of electricity. This concept emerged in recent decades not only for economic reasons but also ecological and even political. SG have been the subject of major studies and investments and continues to represent an area of enormous challenges. Some of the problems of intelligent systems connected to the managed SG are: the real-time processing optimization algorithms and demand response programs; and more accurate predictions in the management of production and consumption. This paper presents a case study for evaluating the performance and accuracy of energy consumption forecast with use of SVM (Support Vector Machines) in different frameworks.