*3.3. Sensors*

The general architecture of an intelligent managemen<sup>t</sup> system consists of readings (sensors), gateways (communication), and workstations (instructions, analytics, software, and user interface) [41–43].

#### 3.3.1. Sensors for Health Monitoring

Healthcare has become a prolific area for research in recent years, given that new sensor technology allows real-time monitoring of the patients' state. Smart healthcare

provides healthcare services through smart gadgets (e.g., smartphones, smartwatches, wireless smart glucometer, etc.) and networks (e.g., body area and wireless local area network), offering different stakeholders (e.g., doctors, nurses, patient caretakers, family members, and patients) timely access to patient information and the ability to deploy the right procedures and solutions, which reduces medical errors and costs [44].

Biosensors are fundamental when monitoring health, and different applications can be identified in medical diagnoses [45], and antigen detection [46], among others. Inorganic flexible electronics have witnessed relevant results, including E-skin [47], epidermal electronics (see Figure 4) [48], and eye cameras [49]. Some common materials used to create these sensors are carbon-based or conductive organic polymers, which present poor linearity [50,51]. However, more reliable, and flexible sensors have been created at lower cost, better linearity, and shorter response time, such as piezoresistive sensors integrating nano-porous polymer substrates [52].

New sensor developments are creating relevant opportunities in the health industry. Current procedures for sensing proteins are commonly based on noisy wet-sensing methods. A more robust procedure is carried out by means of graphene sensors that avoid the drifting of electrical signals, resulting in more stable and reliable signals. These sensors also reduce detection times [53]. Nonetheless, when having these robust sensors connected to the human body, one critical challenge is their communication through the wireless sensor network, since the IEEE 802.15.4 standard can hardly be adapted to multi-user interfaces [54]. Still, some solutions have been proposed, such as replacing the ultra-wideband (UWB) with a gateway, so sensor nodes stop and switch to 'sleep mode' until new information transmission is needed again. This allows lowering the energy consumption and collisions and increase the speed and number of users [55].

**Figure 4.** (**a**) Epidermal electronic system (EES) made of a skin replica created from the forearm, before and; (**b**) after application of a spray-on-bandage; (**c**) Colourized microscopy image of the EES with conductive gold films of 100 μm and; (**e**) its magnified view. (**d**) Microscopy image of EES with gold films of 10 μm and; (**f**) its magnified view [48].

3.3.2. Sensors for Mobility Applications

Three main systems of urban mobility: vehicles, pedestrians, and traffic, are considered in this section. Due to the increasing number of vehicles every year in urban settlements, traffic jams, pollution, and road accidents tend to increase as well [56]. These problems sugges<sup>t</sup> that there is an emergen<sup>t</sup> need for intelligent mobility solutions. One such solution is intelligent traffic control, oriented to avoid traffic jams and optimize traffic flow [18].

Due to repetitive starts and stops, fuel consumption and carbon emissions increase during traffic jams [18]. Therefore, providing solutions for traffic jams represents a direct positive impact in terms of urban mobility and air quality in cities. Moreover, heavy-duty vehicles (HDVs) and freight traffic release into the atmosphere large quantities of carbon emissions [57]. The automotive industry has put significant effort into developing more energy efficient powertrains (for example, hybrid electric vehicles). However, most HDVs are still fueled by diesel and providing optimal solutions to reduce the carbon emissions produced by these types of vehicles becomes a fundamental task [58].

Due to the COVID-19 pandemic, urban mobility underwent significant changes, such as a noticeable decrease in collective and individual transport, and with that a reduction in air pollution and carbon emissions [59]. This phenomenon has made governments and citizens consider future changes in post-lockdown mobility, to maintain cleaner environments in their cities. Applications in smart mobility include vehicle–vehicle (V2V), vehicle–infrastructure (V21), vehicle–pedestrian (V2P) [60], and vehicle–everything (V2X) connections (see Figure 5) [61].

Vehicles include several sensors needed for their proper operation, measuring several operational parameters of the vehicle, such as speed, energy consumption, atmospheric pressure, and ambient temperature [62,63]. Such parameters are used to optimize speed profiles to minimize vehicle energy consumption considering traffic condition and geographical information. To achieve this aim, a cloud architecture is implemented that retrieves information from vehicle sensors and external services. In [64], an eco-route planner is proposed to determine and communicate to the drivers of heavy-duty vehicles (HDVs) the eco-route that guarantees the minimum fuel consumption by respecting the travel time established by the freight companies. Additionally, in this case, a cloud computing system is proposed that determines the optimal eco-route and speed and gear profiles by integrating predictive traffic data, road topology, and weather conditions. Vehicle weight and speed regulation are also important to ensure road and passengers safety, helping in the avoidance of serious accidents [65]. Efforts in increasing pedestrian's safety are valuable contributions in improving urban mobility [66]. Regarding traffic, conventional traffic light systems are defined in a non-flexible structure, such that light transitions have defined delays and onsets [67]. Dynamic changes in traffic volume, congestions, accidents, and pedestrian confluence, should have been considered to provide an optimized traffic control [67].

Pedestrian´s movement and behavior in urban settings have been monitored mainly using cell phones, by monitoring call detail records (CDRs) [68], social media checkins [37,69], MAC address reading [70], and smart cards detection in public transport [71]. Vehicle detection have been achieved by cement-based piezoelectric, induction loop sensors, measuring vehicle's weight-in-motion (WIM), and performing vehicle type classification [65], and ferromagnetic sensors buried in the asphalt for smart-parking solutions [72]. Vision-based sensors, such as infrared (IR) [67] and light detection and ranging (LiDAR) [73] have been used to detect the position of vehicles, pedestrians and buildings within a given proximity. Pedestrian–vehicle (P2V) oriented sensors also exist, such as the "smart car seat", a contact-free heart rate monitoring sensor oriented to ensure driver's well-being and safety [74]. Mobile phone apps have allowed P2V and V2P applications for collision prediction [66,75].

**Figure 5.** Two types of V2V connections: (**a**) from vehicles to vehicles with an intermediate transfer connection and; (**b**) directly from vehicles to other vehicles [61].

Recently, virtual sensors have been used to enhance innovative solutions especially in the electro-mobility sector. VSs have been introduced for operating in the sensor-cloud platform as an abstraction of the physical devices. In particular, a VS can logically reproduce one or more physical sensors, facilitating and increasing their functionalities, performing complex tasks that cannot be accomplished by physical sensors [76]. Differently from a real sensor, the VS is equipped with an intelligent component based on data processing algorithm to derive the required information elaborating the available input data from heterogeneous sources. Indeed, VS is typically used in services in which it is necessary to derive data and information that are not available or directly measurable from physical sensing instrumentation [77,78]. Although the use of such sensors has been explored in different domains or verticals of the smart city, the mobility sector is the one where they find large application. In the electric mobility domain, for instance, they are used to predict the personal mobility needs of the driver to estimate the duration and cost of the battery charging, to predict the energy demand of medium or long-range trips, etc. All these predictions are performed through ad-hoc algorithms able to process available input data from the electric vehicles, the users, and the charging stations [76–78].

#### 3.3.3. Sensors for Security

The human and environmental security approaches are a very crucial ingredient to achieve sustainable development in smart cities. Security is referred to as a state of being free from danger or threat and for maintaining the stability of a system. Safety is a dynamic equilibrium, which consists in maintaining the parameters important for the existence of the system within the permissible limits of the norm. According to the United Nation's Human Security Handbook and Agenda 2030 Sustainable Development Goals (SDGs) [79] the types of insecurities endangering the sustainable development of humans and, hence, future cities are: food, cybernetic, health, environmental, personal, community, economic, and political, as the main core of a smart city.

Food security: The importance and technological challenges of the integration of urban food systems in smart city planning are discussed in [80]. High quality and sustainable production include smart hydroponics and gardening systems that gather information by sensors that measure pH, humidity, water and soil temperature, light intensity, and moisture [81]. Several methods are proposed to monitor the quality and safety of the food during production and distribution, including gas sensor array [82] for the analysis of chemical reaction occurred in spoiled food. Hybrid nanocomposites and biosensors have also been reported in food security context [83].

Cyber security: The main security challenges, including privacy preservation, securing a network, trustworthy data sharing practices, properly utilizing AI, and mitigating failures, as well as the new ways of digital investigation, are discussed in [11,40]. Design plane solutions are usually software-based and use diverse types of encryption techniques, including advanced encryption standard (AES) and elliptic curve cryptography (ECC) for crypto or level security and encryption, authentication, key management, and pattern analysis for the system-level security [84,85] (see Figure 6).

**Figure 6.** Smart city architecture defined in five planes: application (connects city and citizens), sensing (sensors measurements), communication (cloud services), data (processing and analysis) and security and privacy planes (assurance of security and privacy) [85].

Health security: In-body inserted devices are designed to communicate with healthcenters and hospitals [55]. The privacy, security, and integrity of these sensors and the information on the health record concerning legal and moral issues are of grea<sup>t</sup> interest which is widely discussed in works such as [86,87].

Environmental security: In recent decades, the use of satellite remote sensing and in-orbit weather observation, disaster prediction systems have risen drastically. These tools are an integral sensing part of the future smart cities [88,89]. There is a wide range of sensors, including earthquake early detection systems which use vibration detection and monitoring soil moisture and density of the earth [90], radiation level detectors [91], tsunami inundation forecast methods assimilating ocean bottom pressure data [92]. These sensors are connected in a wireless network, offering a global prognosis of environmental threats. Continuous emission monitoring systems (CEMS) have helped develop marketbased environmental policies to address air pollution [93]. CEMS are allowing better tracking of powerplant emissions in real time to inform decarbonization strategies for the grid [94]. Efforts to deploy cost-effective sensing capabilities so far have produced fragmented data, but new optimization and AI tools are being proposed to resolve this issue [95]. New smart sensing and visualization (e.g., satellite, LiDAR, etc.) capabilities are focusing on greenhouse gas emissions (GHG), for instance around the carbon capture potential of agriculture, forestry, and other land uses (e.g., natural climate solutions). Advanced monitoring, reporting, and verification (MRV) features will continue to play a major role in enhancing the transparency, environmental integrity, and credibility of subnational, national, and regional emissions trading systems (ETS) for the future integration of a global carbon market [96,97]. Regarding infrastructure and buildings, continuous monitoring to detect corrosion and minor damages to prevent a possible failure takes advantage of the integration of surveillance cameras, humidity, atmospheric, and stress sensors, among others [98]. A simple combination of vibration and tilt sensing devices provides one of the low-cost and high-efficiency techniques proposed for a wide range of structures [99].

Personal and community security: To detect anomalies, violence, and unauthorized actions, biometrics and surveillance cameras are widely used. Smart lighting systems are a useful and cost-effective tool that uses common sensors like light and motion detectors and can improve the security tasks [100]. Surveillance cameras, face-recognition systems, and global positioning systems (GPS), in combination with data handling systems, are increasingly common tools in the hand of law-enforcement agencies as smart public security strategy reported in [39]. Ref. [101] reviews the possible combination of different devices categorized in sensors, actuators, and network systems. The challenges presented by the growing use of such technologies and concerns for individual privacy is the topic of an emerging research area [102].

#### 3.3.4. Sensors for Water Quality Monitoring

Water is an invaluable commodity and is necessary for any living being. Smart water managemen<sup>t</sup> focuses mainly on making water distribution systems more efficient by applying sensors and telemetry for metering and communication [103,104]. It applies in three broad areas: fresh water, wastewater, and agriculture. Moreover, more holistic perspectives around shared resource systems, such as the water–energy–food nexus are also benefiting from new sensing capacities and smart managemen<sup>t</sup> systems enabled by digital technologies to provide more sustainable, resource efficient use solutions [97]. The principal usefulness of smart water systems lies in controlling valves and pumps remotely [104] measuring quality [103], pressure, flow, and consumption [105].

Consumption monitoring includes metering and model applications to describe consumption patterns. Water loss managemen<sup>t</sup> encompasses leakage detection and localization [105]. For water quality the focus is on measuring, analyzing, and maintaining a set of pre-established parameters. It is an integral real-time managemen<sup>t</sup> involving stakeholders [106–108]. In the agriculture, the use of IoT devices is a common way to make irrigation more efficient and effortless [109–111]. Noise sensors and accelerometers are popular methods to detect leaks in water distribution infrastructure [105,106].

The use of electromagnetic and ultrasonic flow meters and sensors for measuring pressure are IoT technologies for water consumption rate analysis [103,108].

Sensors used to analyze the quality of the water are mainly applied for physical– chemical parameters such pH, temperature, electrical conductivity and dissolved oxygen [108,109], also oxidation-reduction potential and turbidity [112–114], and presence of toxic substances [115]. In some cases, novel probes, such as for residual chlorine [103] or nitrate and nitrite, were implemented (see Figure 7) [112]. Humidity sensors are applied to measure soil moisture and assist in managing the schedule programs of irrigation in agricultural lands [110,116,117].

#### 3.3.5. Sensors for Waste Monitoring

Smart waste managemen<sup>t</sup> consists of resolving the inherent problems of collection and transportation, storage, segregation, and recycling of the waste produced. Use of smartbins, solutions for the Vehicle Routing Problem (VRP) and waste managemen<sup>t</sup> practices have been reported [41,118]. The use of smartbins refers to the implementation of different kind of sensors in the bins used to collect waste, which provide quantitative and qualitative information about the bin content [119–121].

For the VRP, the proposals are algorithms for optimization of the routes, considering social, environmental, economic factors, peak hours, infrastructure, type and capacity of the collection vehicles and others, in an effort to save resources like money, time, fuel, and labor [121–123].

**Figure 7.** Outer and inner view of an integrated IoT sensor for water quality monitoring applications. The sensor consists of a nitrite and nitrate analyzer based on a novel ion chromatography method, used for detection of toxic substances [112].

With this, researchers aim for an integrated, real-time managemen<sup>t</sup> that involves the communities and all the stakeholders [124,125]. The principal use of sensors in smartbins is monitoring the volume, weight, and content of the bins. For monitoring the filling level of the container, the main approaches that have been used are ultra-sound (US) [43,119], and in some cases IR sensors [121,126]. A load cell is also used to detect the weight of the bin [124,127]. In the literature, various sensors are used to detect harmful gases [126], movements near the container [120], and metal sensors to separate metallic waste [128], and to measure humidity [126,128], as well as temperature [43,123]. The My Waste Bin IoT container presented in [43] is shown in Figure 8.

**Figure 8.** Front and back view of the My Waste Bin, an IoT smart waste container, enabling real-time GPS tracking and weight monitoring [43].

#### 3.3.6. Sensors for Energy Efficiency

Energy is an essential resource for the operation of the many activities occurring in cities [14]; therefore, the efficient use of this resource is paramount to reduce costs and promote environmental and economic sustainability [129].

The main sinks of energy consumption in urban communities are those associated with industrial and transport activities, buildings operations, and public lighting. In this section, we focus on the sensors used to monitor the usage of energy ground transport in buildings and public lighting. Since sensors for industrial activities were covered in previous sections.

*Ground transportation* represents the main sink of energy consumption ( ∼45%) and the major source of air pollutants in urban centers [130]. Car manufacturers report the specific fuel consumption (SFC measured as L/km) of their vehicles using laboratory test protocols. However, they do not report these data for heavy-duty vehicles. Furthermore, the real vehicles' energy consumption is affected by human (driving), external (traffic, road, and weather conditions), and technological factors. For gasoline and diesel-fueled vehicles, the common strategies to measure real-world fuel consumption on a representative sample of vehicles are: (i) measuring the fuel's weight before and after a specific distance being driven (gravimetric method), and (ii) measuring instantaneous fuel consumption through the on-board diagnostic system (OBD method). This second alternative uses optical sensors to measure the engine RPM, pressure sensors to measure the inlet air flow. The engine computer unit (ECU) uses these measured data to determine the engine fuel injection time. In addition, a global position system (GPS) determines the vehicle's speed. Using all this information, the ECU reports via OBD the vehicle's instant fuel consumption. Currently, there are commercially available readers that read the OBD data and send the collected information to the cloud. Using these technologies, telematics companies monitor thousands of vehicles in operation [131–133]. Similar systems are available for electric vehicles. We recommend this OBD-based alternative to measure the real energy consumption in ground vehicles.

*Buildings* represent 40% of total energy consumption [134] and 30% of greenhouse gas (GHG) emissions [129,134]. The main physical and non-physical factors involved in indoor environment quality are shown in Figure 9 [134]. These factors are measured using wireless sensors [135,136], virtual sensors [137], and artificial neural networks [16]. Energy consumption in buildings is associated mainly with (i) thermal comfort (operation of heating, ventilation, and air conditioning-HVAC systems); (ii) indoor lighting; (iii) various electrical loads (operation of electric equipment); (iv) thermal loads (use of fuels for heating and cooking), and (v) indoor air quality (pollutant concentration, odor, and noise) [138,139]. Table 2 shows the variables used to grade these five aspects and the sensors frequently used to measure these variables. However, additional variables influence energy consumption, such as the occupation level [140], and the building's structural design [141], and outdoor conditions (temperature, humidity, pressure, and solar radiation). Therefore, additional sensors are used to measure them. Some research works have focused on designing intelligent building managemen<sup>t</sup> systems (BMS) that use, in real-time, data from the sensors listed above and take actions oriented toward the reduction in energy consumption, such as turn lights off, closing doors and windows, etc. [142].

*Public lighting systems* represent almost 20% of world electricity consumption, and it is responsible for 6% of GHG [143]. Therefore, it is essential to centralize street lighting control and smart managemen<sup>t</sup> to reduce energy consumption, maintain maximum visual comfort and occupant requirements [139]. The variable most used in lighting systems is the lighting power density (LPD) [138]. Neural networks, wireless sensors, algorithms, and statistical methods are used to estimate the energy consumption and the corresponding costs [129,135,136]. Environmental factors, pedestrians' flow, weather conditions, and brightness levels influence light intensity [129,144]. The urban space adopts the most advanced Information and ICTs to support value-added services to manage public affairs, connecting the city and its citizens while respecting their privacy [20,145].

**Figure 9.** Physical and non-physical factors in IEQ studies [134].

Most of the time, the monitoring of the variables influencing energy consumption in buildings and public lighting is carried out wirelessly [18,129]. Various intelligent sensors [135], such as artificial vision, are used to measure temperature, relative humidity [17], electric current, gas flow, air quality [139], lighting, luminosity [129], solar radiation [139], and acoustic emission [146]. Measurement devices include light-dependent resistor (LDR) [139], IR radiation [140,147], semiconductors, magnets, and optic fiber. Intelligent sensors based on IoT are key to real-time monitoring of the many variables involved in energy management; these sensors can be adapted to microcontrollers and virtual sensors [129,137]. The main problem with wireless sensors is their battery life; therefore,alternative energy sources (thermal, solar, wind, mechanical, etc.) is vital, although these energies are usually available in minimal quantities [18].


