Teresa Riesgo’s research while affiliated with Universidad Politécnica de Madrid and other places

The design of border surveillance systems is critical for most countries in the world, having each border specific needs. This paper focuses on an Internet of Things oriented surveillance system to be deployed in the Sahara Desert, which is composed of many unattended fixed platforms, where the nodes in the edge have a Forward Looking InfraRed (FLIR) camera for field monitoring. To reduce communications and decentralise the processing, IR images should be fully computed on the edge by an Automated Target Recognition (ATR) algorithm, tracking and identifying targets of interest. As edge nodes are constrained in energy and computing capacity, this work proposes two ATR systems to be executed in low-power microprocessors. Both proposals are based on using Bag-of-Features for feature extraction and a supervised algorithm for classification, both differing in segmenting the InfraRed image in regions of interest or working directly with the whole image. Both proposals are successfully applied to infer about a dataset generated to this end, getting a trade-off between computing cost and detection capacity. As a result, the authors obtained a detection capacity of up to 97% and a frame rate of up to 5.71 and 59.17, running locally on the edge device and the workstation, respectively.

This paper proposes a novel readout methodology for improving the performance of Refractive Index (RI) based photonic transducers. Specifically, the authors focus on an optical transducer reported recently in the literature, the so-called Resonant Nano-Pillars (RNPs) transducer. The readout signal for this transducer is usually obtained based on the Wavelength Shift of the Resonant Mode (WSRM), which identifies a single point from the signal, such as the minimum of a resonant mode, whose wavelength shift or intensity value has a correlation with the RI of the media, and, therefore, with the monitored chemical component. This work proposes a novel spectral analysis through Principal Component Analysis (PCA), later inferring the property of interest by regression techniques. To evaluate the performance of the proposal, the authors mimic an agro-food experiment emulating a fermentation process as a proof of concept by measuring the ethanol concentration over time in two liquids: Deionized Water (DIW) and White Wine (WW). The authors compare both methods by inferring the ethanol concentration in the two experiments. As a result, the authors demonstrated experimentally that the proposal significantly outperformed the WSRM method, reporting an improvement of the Limit of Detection (LoD) of up to 140 times. Moreover, this PCA method can be also applied to many other biochemical sensing systems and transducers.

The Internet of Things is being established nowadays and the deployment of devices to be interconnected with each other and to the Internet is becoming a more common issue. The Internet of Things is structured in several levels and, in this regard, depending on the layer in which these devices are located, the complexity range and specific constraints arise. In the lowest level of this structure, the so-called edge is found, where data are gathered from the environment by tiny electronic devices which are very limited in energy, computing and memory resources. These devices, in turn, present different levels of complexity, which lead to a distinction between the edge layer and the sensor or extreme edge layer. In the last one is where the most essential and limited tasks are carried out, these are sensing and sending data. Meanwhile on the edge, in addition to sensing, some processing can be performed in order to offload upper layers and to save precious energy by not communicating useless data. In this review paper, the structure of the Internet of Things is presented and the extreme edge is detailed, presenting the implementation options and requirements, with the purpose of having a better understanding of the needs when a specific application has to be developed and a set of devices has to be deployed. Special emphasis receives the bottom layer, focusing on devices oriented to last unattended for tens of years.

The explosion of the Internet of Things has dramatically increased the data load on networks that cannot indefinitely increment their capacity to support these new services. Edge computing is a viable approach to fuse and process data on sensor platforms so that information can be created locally. However, the integration of complex heterogeneous sensors producing a great amount of diverse data opens new challenges to be faced. Rather than generating usable data straight away, complex sensors demand prior calculations to supply meaningful information. In addition, the integration of complex sensors in real applications requires a coordinated development from hardware and software teams that need a common framework to reduce development times. In this work, we present an edge and fog computing platform capable of providing seamless integration of complex sensors, with the implementation of an efficient data fusion strategy. It uses a symbiotic hardware/software design approach based on a novel messaging system running on a modular hardware platform. We have applied this platform to integrate Bluetooth vehicle identifiers and radar counters in a specific mobility use case, which exhibits an effective end-to-end integration using the proposed solution.

Cyber-Physical Systems are experiencing a paradigm shift in which processing has been relocated to the distributed sensing layer and is no longer performed in a centralized manner. This approach, usually referred to as Edge Computing, demands the use of hardware platforms that are able to manage the steadily increasing requirements in computing performance, while keeping energy efficiency and the adaptability imposed by the interaction with the physical world. In this context, SRAM-based FPGAs and their inherent run-time reconfigurability, when coupled with smart power management strategies, are a suitable solution. However, they usually fail in user accessibility and ease of development. In this paper, an integrated framework to develop FPGA-based high-performance embedded systems for Edge Computing in Cyber-Physical Systems is presented. This framework provides a hardware-based processing architecture, an automated toolchain, and a runtime to transparently generate and manage reconfigurable systems from high-level system descriptions without additional user intervention. Moreover, it provides users with support for dynamically adapting the available computing resources to switch the working point of the architecture in a solution space defined by computing performance, energy consumption and fault tolerance. Results show that it is indeed possible to explore this solution space at run time and prove that the proposed framework is a competitive alternative to software-based edge computing platforms, being able to provide not only faster solutions, but also higher energy efficiency for computing-intensive algorithms with significant levels of data-level parallelism.

Once specific Smart Sensors are designed and manufactured in the newest nanoelectronics technology, and a Wireless Sensor Network is designed for being used on wearable applications (Cyber-Physical System) with optimum performance (data rate, power consumption, comfortability, etc.) the next step is the treatment applicable to the large amount of data collected. This can be a very general, and sometimes an unaffordable problem, but considering a system collecting physiological data from smart sensors on a human body, the range of possibilities is restricted to health or leisure but also to safety. In this case, a finite, and well-located, number of physiological sensors are producing few data per unit of time, which are locally processed for obtaining a reduced set of characteristics that are globally analyzed. In this paper, an analysis on different approaches for combining data from smart sensors attached to human body, with the purpose of determining the main emotion present in the person, is presented. Machine learning, selection of the best characteristics from raw sensor data, data bases for system training, etc. are the key aspects in this problem. The conclusions of the analysis will help in the design of a new application, where emotion detection can be used for personal safety (domestic violence, sexual violence, bullying, etc.). Attention is paid on the locally and globally data processing in terms of hardware and software, together with low-power behavior.