IoT Edge Solutions for Cognitive Buildings

A cognitive environment (CE) is a smart environment having self-learning and self-adaptation capabilities. It is obtained by augmenting a physical environment by using IT equipment and artificial intelligence technologies. The goal is to furnish the environment’s dwellers and the environment itself with advanced services devoted to (i) improving the quality of life of the people, (ii) optimizing the use of shared resources and spaces, (iii) increasing security and safety, (iv) assisting people in daily life activities, (v) extending the lifetime of devices and infrastructures, and (vi) enforcing and actuating policies promoting sustainability, green-aware behaviors, and energy-saving management of the whole system. Realizing a CE is a complex and multidisciplinary task. It requires transversal skills among which those related to distributed systems, IoT and artificial intelligence technologies, and networking. In addition, suitable methodological approaches and platforms could be leveraged in order to deal with the complex process of designing and implementing a CE. In this chapter, as enabling technology, the COGITO platform is introduced. COGITO is an agent-based IoT platform tailored to the development of CEs in a heterogeneous continuum computing environment comprising cloud, fog, and edge resources. The practical use of the platform is demonstrated through some use cases developed at the ICAR-CNR headquarter at Rende (Italy).

Cognitive buildings are structures that aim to produce meaningful results for occupants with the help of technology. To understand the cognitive buildings and harness their potential, it is important to know network technologies such as Internet of Things, cloud computing, fog computing, and edge computing. In this chapter, these key concepts are discussed within the scope of cognitive buildings. The design of cognitive buildings with various services and components requires a systematic approach. A modular design framework for smart buildings is proposed in the study. In accordance with the proposed framework, sample scenarios using key concepts are presented.

In the next few years, smart environments are expected to originate billions of raw Internet of Things (IoT) data that need to be stored and processed in order to implement a variety of control and monitoring services. While complex and long-term processing typically relies on remote cloud facilities, low-latency and interactive cognitive services may highly benefit from caching and computation resources, as well as artificial intelligence (AI) components, deployed at the network edge, close to where data are produced. Therefore, edge caching will play a pivotal role for the efficient and effective deployment of smart and cognitive environments, including houses and buildings. In this chapter, we scan the literature related to edge caching for IoT smart environments and identify the most promising decision policies together with the key benefits and open challenges. Conventional caching techniques are first scanned, before delving into more disruptive in-network caching solutions built upon the named data networking (NDN) paradigm. Focus will be then on the possible interplay of NDN-based edge caching policies with software-defined networking (SDN), as well as on the opportunities to leverage edge caching powered by AI techniques as a prominent sixth-generation (6G) enabler.

Cognitive buildings are given by the integration of the Internet of Things (IoT) with cognitive dynamic systems (CDS) with the aim of improving the management of public services and residential buildings with cognitive and self-adaptive functions. The research work has seen the contribution of jurists, geographers, engineers, and anthropologists who have crossed their competences to support the design of cognitive buildings. The aim was to make the research results related to the knowledge of the regulatory framework on sustainable living and the efficient use of energy in living spaces exportable, using advanced technologies. It was conducted an analysis of the territorial context, since the new technologies change the habits and the ways that citizens and city users must enjoy the urban habitat while affecting their perception of well-being. The attention was focused on the relationship between people, living, well-being, and technology. All of this has allowed us to identify and analyze specifically the needs of customers, translating them into a set of variables and parameters essential to the ex ante design and ex post evaluation of a cognitive system. It has been developed, also, a supranational (international and EU) and national normative analysis on innovative technological models for the improvement of living comfort.

Communication has been always a vital need for society, because it is a carrier of exchange of experiences that has allowed a more and more rapid evolution in every field of knowledge. As a prerogative of living beings, the “communication paradigm” in recent years has been evolved and adapted to new application domains like those tied to the cognitive environments by exploiting the advancements coming from the fields of artificial intelligence, edge computing, and Internet of Things. In this chapter, we will focus on several approaches used in some research projects with the aim of ensuring effective communications in the context of cognitive IoT environments (CoIoTEs). A CoIoTE is an environment augmented with IoT devices and technologies that autonomously learns how to act on itself and how to adapt based both on the surrounding context and on the needs of people living in it. The aforementioned approaches have been applied in some real operational scenarios.

The buildings are natural spaces for people where they live and work. In this context, the security and safety of dwellers are a necessity.

This is not however a simple endeavor: feeling either too little or too much security increases the levels of psychological stress that dwellers are subject to, especially in the cases in which, to increase security, people are monitored through cameras. In this case, the feeling that someone is watching you is extremely detrimental to the at-ease felt in a space.

Surveillance cameras are often necessary and useful, but, sometimes, it is important to have an alternative approach that could be much more unobtrusive. Such an approach can consist in audio analytics, as microphones are most often concealed enough to avoid increasing the stress levels of inhabitants while keeping an adequate amount of security.

This chapter aims to show a system of security for cognitive buildings based on microphones. In particular, we will analyze the information content of raw recording data obtained from the microphones and their processability into audio events, with detailed, actionable human-readable information.

We also propose a completely edge-based processing paradigm with special safeguards for data filtering and information control to obviate any privacy concern that might arise.

We then propose a complete implementation, and we apply it in two case studies that are significant: a residential apartment, with security necessities, and a free access room with event monitoring priorities.

Cognitive buildings are equipped with sensors and actuators to customize the indoor conditions according to users’ needs and preferences. The eLUX lab, at the Smart Campus of the University of Brescia (Italy), is the first national cognitive building where educational spaces are monitored and dashboards promote the users’ awareness. There, a fixed IoT network allows gathering data to perform analytics for prompt fault detection and fine-tuning of the conditions and possibly of the energy management. However, the users’ involvement is paramount for correct tuning and customization, as indoor conditions can be measured by a sensor (e.g., temperature, humidity, illuminance, CO₂ concentration) located in a single point in the indoor space, while the parameters can strongly vary in the presence of, e.g., a fan coil or the distance from the windows and doors. To address this issue, an RTLS (Real-Time Location System) can be connected to the users’ direct and indirect feedback to support a precise definition of the indoor conditions and boost the improvement of the cognitive behaviour (i.e., automatic window opening or lighting dimming). In this chapter, we illustrate how the eLUX Lab can be enhanced to support the aggregate programming paradigm for offering resilient distributed services (exploiting an RTLS) that run on wearable devices without relying on the connection to a central server.

This work involves the development of an advanced solar shading control algorithm with the aim of reducing energy requirements and improving visual comfort. The proposed control system is based on IoT devices that sense the environment and interact with it following real-time intelligence that allows adaptation to changing situations. The designed control strategy is aimed at adjusting the tilt angle of movable Venetian blinds to take the greatest advantage of natural light in the presence of occupants, avoiding glare, and ensuring energy savings. The study integrates the use of an artificial light management system, which is necessary to reach the setpoint illuminance. The results show that the control system can halve cooling energy demand and it can reduce electricity for artificial lighting by up to approximately 30%. In the scenario based on scheduled occupancy and LED lamps, an economic saving of 30.8% is achieved. The research findings highlight how the use of devices with greater intelligence can guide towards better-managed buildings, which are more responsive to user preferences.

Energy consumption of heating, cooling, ventilation, lighting, and appliances is deeply influenced by human presence in buildings. Accurate room occupancy prediction is a key to making buildings cognitive and self-adapting in order to achieve energy efficiency and wastage cut. Instead of using cameras or human tracking devices, a predictive model based on indoor non-intrusive environmental sensors allows mitigating privacy concerns. In such direction, this study aims to develop a data-driven model for occupancy prediction using machine learning techniques based on a combination of temperature, humidity, CO2 concentration, light, and motion sensors. The approach has been designed and realized in a real scenario by leveraging the COGITO platform. The experimental results show that the proposed long short-term memory neural network is well suited to account for occupancy detection at the current state and occupancy prediction at the future state, respectively, with an overall detection rate of 99.5% and 92.6% on a literature dataset and 99.6% and 94.2% on a real scenario. These outcomes indicate the ability of the proposed model to monitor the occupancy information of spaces both in a real-time and in a short-term way.

The COVID-19 pandemic has impacted the lifestyle of people in every community and workplace, including universities. There, places like cafeterias where people are expected to not wear a mask for the majority of time, i.e., while eating or drinking, are potentially very risky. In such scenarios, the Internet of Things (IoT) technological stack and Edge Intelligence paradigm represent really useful solutions for the safe provision of essential services by predicting, monitoring, and contrasting potentially dangerous situations. Therefore, in this chapter, we present an example of cognitive building denominated as Smart Cafeteria: it is a highly sensor-and-actuator-augmented environment, aimed at monitoring the users’ presence in order to detect those dangerous situations for COVID-19 virus spreading. Driven by the development guidelines of the ACOSO-Meth methodology, the Smart Cafeteria exploits a set of heterogeneous edge devices, IoT technologies, cloud services, and neural networks for acquiring, gathering, analyzing, and predicting temperature and humidity values, since the latest studies have recently suggested that cold, dry, unventilated air contributes to virus transmission, especially in the winter season. The Smart Cafeteria has been designed within the campus of the University of Calabria, in Italy, as the specific target, but it can be adapted to any popular building or workplace. The obtained prototype testifies the suitability of approaches based on the Edge Intelligence paradigm for the development of effective and cheap solutions aimed at safer living spaces, within and beyond emergency situations.

Structural health monitoring (SHM), together with condition monitoring (CM), nondestructive evaluation (NDE), statistical process control (SPC), and damage prognosis (DP), through the most recent techniques of survey and data processing, allows to identify, evaluate, and monitor with ever-greater clarity the structural characteristics and the level of damage of any building and, therefore, to predict its trend over time.

The use of traditional and experimental sensor networks and the processing of the data obtained from them allow to identify anomalies in the behavior of structures in operation, as well as to implement early warning systems.

The use of accelerometric sensors is helpful for identifying the representative parameters of the structural behavior; the measurements of the displacements, on the other hand, allow a quick estimate of the magnitude strictly correlated to any damage suffered by the structure during a seismic event or a failure.

In particular, the connection of SHM with AI and with building information modeling (BIM) can make the system cognitive, making it capable of managing (e.g., ensure, predict, assess) the healthiness of a building.

The article also presents a case study to highlight how the proposed methodology is applicable to concrete cases.

The Internet of Things paradigm envisions a world where every physical object is equipped with sensing/actuation capabilities and computing power and acquires its own digital identity. These objects are referred to as “smart” and have the goal of collecting and processing information about the environment surrounding them. One of the fields of interest in IoT applications concerns the intelligent management of activities in indoor environments, even if affected by unusual restrictions due to special conditions, such as those posed by the Covid-19 pandemic. This study focuses on the development of an IoT application based on the COGITO platform for the intelligent management of meeting rooms. By processing data collected from a set of IoT devices, cameras, and cognitive microphones, the developed prototype is able to autonomously monitor and make decisions about aspects that continuously affect environmental comfort, event management, and assessment of compliance with anti-contagious regulations throughout the time the room is occupied. After a brief review of the state of the art, the chapter describes the developed application. Furthermore, it highlights the features that make meeting room environments more comfortable for users and effective in managing events such as meetings and lectures.

In cognitive buildings (CBs), intelligent IoT edge devices do more than gather data. They also aggregate, analyze, and stream data at the edge of the network, where cognitive controllers based on machine learning algorithms enable new levels of control and security while significantly improving the overall user indoor comfort and safety. As a result, these CBs will be more intelligent, self-learning, innovative, and simple to manage. CBs also promise to heighten their dwellers’ comfort (and, as a consequence, their performance) by optimizing, e.g., lighting, temperature, and humidity where needed. The reinforcement learning (RL) method is becoming more and more attractive for the designing of cognitive controllers. This chapter presents a human-centered reinforcement learning control for visual comfort management in cognitive buildings. A satisfaction-based visual comfort model is coupled with RL to adapt the boundaries of the comfort zone in the presence of a group of occupants. Compared with the traditional controls, it is personalized and human-centric since users’ perceptions of the surroundings are exploited as the feedback loop. A case study of an office room and its performance are also presented.

In the last few years, many appliances are spreading into our houses and are daily used. Such equipment significantly improves the quality of life of people, but their use, when not well regulated, can bring a needless increment in the electricity bill. Such an increment could be mitigated by using intelligent scheduling policies that guide the users toward correct exploitation of electric devices so optimizing their use while, at the same time, saving energy, money, and time. This chapter proposes a case study in which a cognitive scheduling approach is used. Such a case study, implemented in the context of the COGITO project, is devoted to automatically scheduling electric loads in houses according to user preferences, self-produced energy, and variable energy costs.

The management of thermal comfort in a building is a challenging and multi-faced problem because it requires considering subjective parameters, such as human preferences and behaviors, and also objective parameters, which can be related to other environmental aspects like the reduction of energy consumption. This chapter exploits cognitive technologies, based on deep reinforcement learning (DRL), for automatically learning how to control the HVAC system in an office. The goal is to develop a cyber-controller able to minimize both the perceived thermal discomfort and the needed energy. The learning process is driven through the definition of a cumulative reward, which includes and combines two reward components that consider, respectively, user comfort and energy consumption. Moreover, a human reward, inferred by the frequency of user interactions with the HVAC system, helps the DRL controller learn the requirements of users and readily adapt to them. Simulation experiments are performed to assess the impact that the two components of the reward have on the behavior of the DRL controller and on the learning process.