Fusion of different sensor modalities has proven very effective in numerous remote sensing applications. However, in order to benefit from fusion, advanced feature extraction mechanisms that rely on domain expertise are typically required. In this paper we present an automated feature extraction scheme based on deep learning. The feature extraction is unsupervised and hierarchical. Furthermore, computational efficiency (often a challenge for deep learning methods) is a primary goal in order to make certain that the method can be applied in large remote sensing datasets. Promising classification results show the applicability of the approach for both reducing the gap between naive feature extraction and methods relying on domain expertise, as well as further improving the performance of the latter in two challenging datasets.

Human behavior modeling enables many applications for smart cities, smart homes, mobile phones and other domains. We present a hierarchical hidden Markov model for human activity recognition that uses semi-supervised learning to automatically learn the model parameters using only labeled data of the top-layer of the hierarchy. This significantly reduces the annotation requirements for such a model and simplifies the design of such a model, since the inherent structure of the activity is automatically learned from data. The design consideration that remains is the number of states used for representing the actions that an activity consists of. Using multiple real world datasets we show that the same model works both for the recognition of activities of daily living in a smart home and for recognizing office activities from audio data. We show how a variable number of action states per activity can result in a significant increase in performance over using a fixed number per activity. Finally, we show how the use of Bayesian and Akaike information criterion results in models using a sub-optimal set of action states, since a model using intuitively chosen set states is able to outperform them.

In this paper we discuss the use of the random forest algorithm for automatic fusion and classification of hyperspectral and LiDAR data. We demonstrate how relative feature relevance can be used in random forests to perform automatic and unsupervised feature selection. This allows using a large number of features without suffering from the curse of dimensionality. The effectiveness of the proposed approach is demonstrated on two datasets. The first dataset features a combination of hyperspectral and LiDAR data for urban classification whereas the second dataset is the well-known Indian Pines dataset featuring pure hyperspectral imagery. We show that by using the proposed approach classification accuracies can be improved significantly.

The Internet of Things (IoT) is understood as a major embodiment of the convergence between device-oriented sensor networks and data-oriented applications that is facilitated through the Internet portfolio of technologies. From an architecture perspective, the wide range of operational parameters entailed by multiple application domains myriads of combinations of sensors and applications is the most significant challenge brought on by the IoT vision. To this end, the design blueprint of a platform that enables the seamless integration of multiple dissimilar devices and their efficient use by independently contributed applications is an essential architecture concern. Herein we address this concern by introducing a Unified Sensing Platform (USP) designed to accommodate an open set of sensor types and to expose their functional capabilities to applications in an efficient, reusable and context-aware way.

Realizing good system performance in IEEE 802.11 deployments requires thorough network planning, a sophisticated task, which, more often than not, requires expert knowledge. In cases where IEEE 802.11 networks operating under different administrations reside in the same geographical area (as defined by the bounds of radio signal transmission and reception), the problem of efficient radio resource allocation arises. Particularly for IEEE 802.11a/b/g systems, efficient radio resource allocation in a dense urban environment can be quite a challenge, due to the finite number of interference-free channels available. To address this shortcoming, we study the mechanisms provided by the IEEE 802.11k standard upon the performance of radio resource management strategies in WLAN type systems under a dense deployment. To this end, we present a Learning With Communication (LWC) Distributed Channel Assignment (DCA) algorithm that exploits local communication among access points to realize a rapidly converging self-organizing channel assignment scheme. We evaluate the performance of LWC DCA algorithm through simulation scenarios.

Wireless Sensor Network (WSN) nodes are often deployed in harsh environments where the possibility of permanent and especially intermittent faults due to environmental hazards is significantly increased, while silicon aging effects are also exacerbated. Thus, online and in-field testing is necessary to guarantee correctness of operation. At the same time, online testing of processors integrated in WSN nodes has the requirement of minimum energy consumption, because these devices operate on battery, cannot be connected to any external power supply, and the battery duration determines the lifetime of the system. Software-Based Self-Test (SBST) has emerged as an effective strategy for online testing of processors integrated in nonsafety critical applications. However, the notion of dependability includes not only reliability but also availability. Thus, in order to encase both aspects we present a methodology for the optimization of SBST routines from the energy perspective. The refined methodology presented in this paper is able to be effectively applied in the case that the SBST routines are not initially available and need to be downloaded to the WSN nodes, as well as the case that the SBST routines are available in a flash memory. The methodology is extended to maximize the energy gains for WSN architectures offering clock gating or Dynamic Frequency Scaling features. Simulation results show that energy savings at processor level are up to 36.5 percent, which depending on the characteristics of the WSN system, can translate in several weeks of increased lifetime, especially if the routines need to be downloaded to the WSN node.

Autonomic networking and cooperative communications paradigms have emerged as promising technologies that can pave the way for next generation networks. CONSERN project is focusing on cooperation and collaboration enablers, cooperative decision and control algorithms that will maximise the expected gains by utilising cooperative problem solving for improving energy efficiency in wireless networks. In this paper, cooperative and/or autonomous solutions are captured and a mapping to the envisaged technologies is made. Based on this mapping an initial analysis of the requirements for energy-efficient cooperation is presented. The cooperation deals with information exchange, actions coordination and decision making within and between different types of network nodes. Based on the requirements analysis a set of key cooperation enablers are identified and further elaborated upon.

Wireless Sensor Network (WSN) applications often need to be deployed in harsh environments, where the possibility of faults due to environmental hazards is significantly increased, while silicon aging and wearout effects are also exacerbated. For such applications, periodic on-line testing of the WSN nodes is an important step towards correctness of operation. However, on-line testing of processors integrated in WSN nodes has to address the additional challenge of minimum energy consumption, because these devices operate on battery, which usually cannot be replaced and in the absence of catastrophic failures determines the lifetime of the system. In this paper initially we derive analytically the optimal way for executing on-line periodic test with adjustable period, taking into account the degrading behavior of the system due to silicon aging effects but also the limited energy budget of WSN applications. The test is applied in the form of Software-Based Self-Test (SBST) routines, thus we proceed to the power optimized development of SBST routines targeting the transition delay fault model that is well suited for detecting timing violations due to silicon aging. Simulation results show that energy savings for the final SBST routine at processor level are up to 35.4% and the impact of test in the battery life of the system is negligible.

Software-Based Self-Test (SBST) has emerged as an effective method for on-line testing of processors integrated in non safety-critical systems. However, especially for multi-core processors, the notion of dependability encompasses not only high quality on-line tests with minimum performance overhead but also methods for preventing the generation of excessive power and heat that exacerbate silicon aging mechanisms and can cause long term reliability problems. In this paper, we initially extend the capabilities of a multiprocessor simulator in order to evaluate the overhead in the execution of the useful application load in terms of both performance and energy consumption. We utilize the derived power evaluation framework to assess the overhead of SBST implemented as a test thread in a multiprocessor environment. A range of typical processor configurations is considered. The application load consists of some representative SPEC benchmarks, and various scenarios for the execution of the test thread are studied (sporadic or continuous execution). Finally, we apply in a multiprocessor context an energy optimization methodology that was originally proposed to increase battery life for battery-powered devices. The methodology reduces significantly the energy and performance overhead without affecting the test coverage of the SBST routines.

In next generation systems and networks, the incorporation of mechanisms achieving robust, predictable and self-adaptive behavior with minimum cost will be a key requirement. Towards this goal we introduce the notion of the "self-growing network". The latter, in its initial deployment stage, is limited to a single dedicated purpose (energy-efficient networking, spectrum efficient communications, control and surveillance use, etc) but can evolve/grow into a multi-purpose, versatile infrastructure, that serves a broader range of applications by utilising combinations of self-x and cooperating features.

Software-based self-test (SBST) has emerged as an effective strategy for non-concurrent on-line testing of processors integrated in embedded system applications. It offers the potential for on-line testing without any hardware overhead. However, test generation is usually based in a semi-automated approach and gate-level information is required for effective test program generation.In this paper we present a methodology for converting processor validation tests in order to perform automated directed random SBST routine generation, based on templates that are developed utilizing a combination of functional and high-level structural pattern generation approaches. The methodology is applied on the OpenRISC 1200 processor, easily achieving test coverage of 86.43%, using only low-effort gate-level independent code generation.

Software-based self-test (SBST) has emerged as an effective strategy for on-line testing of processors integrated in non-safety critical applications. Among the various systems that fall in the previous category, wireless sensor networks (WSN) are often deployed in harsh environments where the possibility of permanent and especially intermittent faults due to environmental hazards is significantly increased, thus on-line and in-field testing is necessary to guarantee the accuracy of the sensed values. At the same time, on-line testing of processors integrated in WSN has the extra requirement of minimum energy consumption, because these devices are operating on battery, cannot be connected to any external power supply, and the battery duration determines the lifetime of the system. In this paper we present a methodology for the optimization of SBST routines from the energy perspective. Techniques utilized for achieving energy reduction include energy aware loop synthesis, loop transformations, instruction substitution and register renaming. Simulation results show that the energy savings at processor level are up to 35.6%.

The vast proliferation in the continuously changing telecommunications environment of Next-Generation-Networks (NGNs) poses strict requirements for the dynamic adaptation of mobile devices according to the user needs, system resources and Radio Access Technology (RAT) availability. The reconfiguration concept emerges as a paradigm that offers an effective solution to this problem. On the other hand, the operation of mobile devices offering cognitive functionalities is characterized by a significant number of parameters, many of which are contradictive and cannot be simultaneously optimized, thus constituting reconfiguration a multi-objective optimization problem. This work proposes a solution to the problem, addressing efficient decision making mechanisms to avoid the ping-pong effect of multiple reconfigurations and exploiting fuzzy logic reasoning approach for terminal reconfiguration decision, focusing on resource management and protocol configuration.

Software-based self-test (SBST) has emerged as an effective strategy for on-line testing of processors integrated in non-safety critical embedded system applications. Among the most popular applications falling in this category are the various mobile devices. However, in-field testing of processors integrated in mobile devices has the extra requirement of minimum energy consumption, since these devices are operating on battery. In this paper initially we present the parameters contributing to energy consumption in order to set the scope of the problem and make a qualitative analysis about the efficiency of the SBST routines from a low energy perspective. Then we propose a power evaluation framework based on a combination of tools from the testing and computer architecture technical areas. Utilizing this framework we evaluate for the first time the most effective structural SBST strategies in terms of energy consumption for three representative processor configurations, showing the validity of our analysis. Finally, the SBST strategies are combined and an appropriate strategy for each processor component is selected, in order to minimize overall energy consumption during on-line testing up to an order of magnitude between the best and worst case combination of test routines.

Today's nanometer technology trends have a very negative impact on the reliability of semiconductor products. Intermittent faults constitute the largest part of reliability failures that are manifested in the field during the semiconductor product operation. Since software-based self-test (SBST) has been proposed as an effective strategy for on-line testing of processors integrated in non-safety critical low-cost embedded system applications, optimal test period specification is becoming increasingly challenging. In this paper we first introduce a reliability analysis for optimal periodic testing of intermittent faults that minimizes the test cost incurred based on a two-state Markov model for the probabilistic modeling of intermittent faults. Then, we present for the first time an enhanced SBST strategy for on-line testing of complex pipelined embedded processors. Finally, we demonstrate the effectiveness of the proposed optimal periodic SBST strategy by applying it to a fully-pipelined RISC embedded processor and providing experimental results

The continuous proliferation in modern communication technologies towards a pervasive composite environment prescribes a profound shift in telecommunication management and poses new challenges, in terms of proposing key solutions for the management and support of legacy as well as emerging systems. Driven by these advances, a direct requirement lies in providing an advanced configuration system that will embrace the demands for optimal and adaptive configuration and reconfiguration of the underlying layers. This objective can be achieved by upgrading the capabilities of mobile equipment with the introduction of a configuration system that incorporates advanced features such as policy-based decision making mechanisms and protocol layer adaptation and reconfiguration procedures. This contribution exploits protocol reconfiguration schemes for policy-based equipment management as addressed in the E2R environment.