Cristian Dragana

Cloud challenges for networked embedded systems: A review

Networked embedded systems have been increasingly deployed in key application areas such as healthcare, home automation, industrial machines, large scale military and environmental monitoring. This has enabled the cyber-physical systems paradigm through wireless sensor networks (WSN) which collect, store, process and transmit information about a focused physical process or system to evolve. One of the key challenges with this approach is handling in an efficient manner the large quantities of data that are being generated in real-time, and extracting the high-level information pieces that are to be presented to the stakeholders and end users. Both the data storage and communication infrastructure also raise security issues at both the decentralized, node-level, and at the central levels. To this extent, cloud-based services and frameworks offer a promising alternative to storing, analyzing and securing sensor data. The paper surveys the main developments in cloud services for handling and securing sensor data at a large scale with quality of service constraints in order to deliver the intended data to the end users.

Large scale heterogeneous monitoring system with decentralized sensor fusion

Monitoring large areas, critical infrastructure systems and surveillance operations require the design of advanced embedded systems which cooperate for achieving mission objectives within a decision support framework. As fundamental building blocks of this new paradigm, we focus on the integration at the communication and data processing levels of wireless sensor networks with unmanned aerial vehicles. WSNs are seen as collections of embedded computing and communication devices which are able to measure and report continuous and discrete values to a central server and allow dynamic interactions through dedicated middleware. At the same time, UAVs are a vital resource for rich field-level information such as picture and video stream capture along with great flexibility for deployment and operation. The paper discusses the general architecture of a large scale heterogeneous monitoring system and the application of decentralized sensor fusion mechanisms for efficient information extraction and data reduction. This promises significat impact on reducing the energy and communication constraints of the embedded sensing nodes. Experimental results are provided stemming from real-world deployment of the UAV platform and ground sensor network subsystems.

Interlinking unmanned aerial vehicles with wireless sensor networks for improved large area monitoring

Wireless Sensor Networks (WSNs) comprising a large number of sensing nodes deployed within the area of interest, are able to measure, process and share specific parameters. Besides enabling effective area coverage, recent research has proven that unmanned aerial vehicles (UAVs) represent a viable addition to large area monitoring through remote sensing and data collecting functions. The proposed interlinking between autonomous UAV and on-ground WSNs overcomes the limitations that prevent sensing nodes from adequately managing large scale applications. This paper presents an overview regarding state of art WSNs and UAVs technologies used for large area monitoring, introduces a novel approach for smart data collecting and further explores the in-network data paradigm. Various simulations have been implemented and analyzed from a comparative standpoint.

Evaluation of continuous consensus algorithm in border surveillance missions

Large scale wireless sensor networks require intelligent and reliable distributed information processing mechanisms which can effectively delegate decision at the field level. Consensus algorithms have been extensively studied and deployed in many generic multiagent systems framework and are able to provide localized agreement among sensing entities. The paper discusses the evaluation of a local consensus algorithm for a border surveillance system, focused on detection of military terrain vehicles, starting from a system architecture for large scale monitoring systems previously proposed. Simulation scenario of a ground local WSN was developed using real data as reference for initial states of the sensor nodes. Performance indicators such as speed of convergence and accuracy of consensus were discussed.

An approach for weighted average consensus in event detection

Large scale monitoring systems require reliable and efficient in-network information extraction mechanisms able to effectively track events at the field level. The study of consensus algorithms for distributed data processing has gained a lot of interest in the last decade. Average consensus algorithms used for decentralized sensor fusion in wireless sensor networks, iteratively compute the global average value, in a completely distributed manner through local information exchange among neighbors. In the first instance, it is mandatory to pursue the reduction of convergence time, for energetic reasons, but it is also essential to lead the convergence to a reliable value. In this paper we propose a new weighted average consensus algorithm, tailored for event detection where each sensor selects its own weights on the basis of some local information regarding number of direct neighbouring nodes and estimated distances to each neighbour. Various simulations have been implemented and analysed from a comparative standpoint.

In-network stochastic consensus for WSN surveillance applications

Surveillance applications require reliable monitoring system architectures and cost-efficient innetwork data processing mechanisms able to provide effective information extraction for event tracking. Wireless Sensor Networks (WSNs) appear to be the most suitable technology due to some well known benefits and the rapid development of embedded devices. Current research is mostly focused on improving and developing new decentralized sensor fusion schemes able to overcome the limitations introduced by the battery operated sensor nodes, but most of the time evaluating the proposed algorithms is performed with a serious lack of precise channel modeling. To close this gap, we provide a stochastic consensus implementation for lossy wireless networks comprising fixed and mobile sensor nodes, as a heterogeneous surveillance system based on ground sensor nodes and UAVs (Unmanned Aerial Vehicles). We developed a simulation framework tailored for fixed and mobile sensor nodes and we are able to evaluate the performance of consensus algorithms from a comparative standpoint, considering deterministic and probabilistic packet propagation models described by common models such as Free Space and Nakagami models. In this paper, we consider a complex in-network adaptive estimation using the distributed least-mean square algorithm tailored for WSN. We perform a mean-square error (MSE) performance analysis for both deterministic and probabilistic propagation models.

Comparison of model reference adaptive control and cascade PID control for ASTank2

This paper presents Model Reference Adaptive Control (MRAC) compared to Cascade PID control in order to evaluate their performances. The cascade control has an outer loop for the level control and an inner loop for flow control that can make the response faster. There the linearization technique is applied to obtain a new linear model. MRAC is designed based on Lyapunov theory and Barbalets lemma that will make the system stable and obtain convergences. The reference model can be selected as the third order system with relative degree two to satisfy rise time, settling time, peak time, and overshoot requirements. General adaptive control laws are developed for tuning three gains based on the reference model. These control gains cover level tracking performance. MRAC also help to remove dynamic uncertainties and modeling errors. Simulation results on Matlab/Simulink indicate that the MRAC can improve the control performances compared to those of the classical PID control effectively.

Data-driven methods for smart building AHU subsystem modelling

Modern, densely instrumented, smart buildings generate large amounts of raw data. This poses significant challenges from both the data management perspective as well as leveraging the associated information for enabling advanced energy management, fault detection and control strategies. Networks of intelligent sensors, controllers and actuators currently allow fine grained monitoring of the building state but shift the challenge to exploiting these large quantities of data in an efficient manner. We discuss methods for black-box modelling of input-output data stemming from buildings. Using exploratory analysis it is argued that data mining inspired approaches allow for fast and effective assessment of building state and associated predictions. These are illustrated using a case study on real data collected from commercial-grade air handling units of a research building. Conclusions point out to the feasibility of this approach as well as potential for data mining techniques in smart building control applications.

Hierarchical cloud storage engine

Cloud computing and cloud storage systems are being used more and more in a variety of domains, from everyday user applications like healthcare monitoring systems and intelligent buildings to military devices. The deployment of these frameworks also enables modern control and automation paradigms found in cyber-physical systems and Industry 4.0. Driven by exponential decreases in computing and storage costs along with high bandwidth, low-latency communication networks, cloud-based infrastructures have increasingly been adopted in large-scale industrial applications. Our paper proposes hierarchical or key-value databases for systems that gather a very big number of data items from a variety of sensors, with focus on smart building and smart city scenarios. We discuss GT.M as a key-value database engine, optimized for transaction processing with a very high throughput. Results of a simulation-based study comparing hierarchical and relational database performance for several types of operations are presented.