Computer scientists consider that we have entered the Centaur Era (humans and computers working together). People are interacting with the physical world in completely new ways, with sensory input devices, smartphones, wearables, smart applications, cars, robots and computers augmenting reality to assist humans [1
]. A rich variety of sensors are present in almost all smart phones: accelerometers, gyroscope, magnetometer, GPS technology, barometer, proximity sensor, ambient light sensor, sound sensor (microphone), touchscreen sensors, fingerprint sensor, pedometer, barcode/QR code sensors, heart rate sensor, thermometer, air humidity sensor and even Geiger counter which can measure the radiation level. The utility of sensors is ubiquitous: starting with embedded systems dedicated to air and water quality monitoring [2
], wearable sensors for remote health monitoring [3
], to wireless sensors networks used in smart transportation system (e.g., bus management routes) [4
], continuing with cars microcontrollers and its ecosystem of sensors used both for critical safety systems—engine, brakes, steering and suspension control—and for non-critical functions dedicated for automatic navigation, indoor air conditioning, central door locking, and entertainment devices. No matter where are applied, sensors provide input data for application processing, their common purpose being to improve life quality. Sensors can allow contextual sensing and understanding: e.g., automatic people gender detection by walking patterns, or mood detection based on different types of movements (slow, nervous or excited), activity detection (walking, running, sitting), automatic detection of the context (inside or outside), or detecting movement intensity based on heart rate and oxygen levels, also providing health status information. The advances in electronics, nanotechnology and biomedical sciences have allowed sensors to be used in monitoring applications of different targets at spatial and temporal resolutions that have never been possible so far providing big datasets.
The explosive increase in the number of sensors and applications in new cars has prompted a change in the role of vehicles, from a peripheral one—strictly for transport—to a real network that hosts clusters of connected devices. Millions of sensors used by people and their mobile devices enhance the advanced infrastructure, intelligent networks and platforms of smart cities [5
]. Smart public transport uses the technology to provide public transport users with a better user experience. The use of sensors and the GPS technology can provide real-time data about geospatial location of each bus, next-stop information relative to current location, and delay information relative to default timetables, about arrivals and departures of public transport. Online route planners which are able to provide much more useful information beside the travelled distance and crowded zones—such as air quality, road profile, and profit that might be obtained on each route—may help users choose the most efficient route from one location to another.
Devices and sensors communicate with each other like never before—more frequently, gathering more data—which in turn makes the communication more significant. Daily, more sensors are going online, increasing the wealth of useful information. Thousands of sensors are located on the highways and main streets collecting traffic parameters—occupancy, volume, and speed—at the rate of one reading per sensor per minute. Smart street lighting is based on enhancing lamp posts with sensors for detecting traffic—cars or pedestrians which are approaching—so that light intensity can be increased when it is actually needed. With this addition, lamp posts become smart assets that will help monitoring everything from the weather and air quality to traffic, even detecting and assisting police in locating gunshots or contributing to reduce the number of burglaries or crimes.
Accuracy, completeness, and timeliness of information, together with the speed of decision-making are of paramount importance in managing the mobility of goods and people. Smart traffic and the associated sensor networks are subjects that have recently attracted—and chances are they will continue to attract—a significant deal of research recently. A search in the Scopus database reveals that from 2010 until now more than 589,719 documents related to sensors in general appear on Scopus, and more than one thousand papers dealing with smart traffic have appeared in the last two years. Although several reviews have been published, none examines the research on smart traffic from the point of view of applications, adopting a multidisciplinary stance. This survey aims at filing that gap. Recent literature is critically reviewed and an interdisciplinary synthesis, which has been seen to be advantageous [6
] is provided. Some significant trends and challenges for future research are highlighted and discussed.
The need for a multidisciplinary approach to smart traffic also results from Figure 1
, which displays commonly found keywords and the relationship among them in terms of co-occurrence in scientific literature. The VOSviewer (http://www.vosviewer.com
) software determines how often each keyword occurred within the database as well as how often the elements were cited together. A systematic survey analyzing bibliographic data can be found in [7
]. Clusters are shown on the map in different colours. Basically, six clusters of terms can be isolated, counting 77 words all together, and three of these clusters are more significant in the context of this work. The first cluster is composed of terms related to the prediction, modeling and deployment of infrastructure that enables smart traffic—wireless networks, smart grids, VANETs, LTE networks—as well as performance analysis metrics, energy efficiency, load balancing. The second cluster is centred around the smart city concept and includes intelligent transport, autonomous vehicles, electric vehicles, smart card, smart devices, cyber–physical systems. The third cluster is organized around solution for the avoidance of traffic congestion: use of mobile phones, video analysis and image processing applications, RFID, user participation, and driver history-based applications. The other clusters are organized around the much broader concept of Internet of Things (IoT), Web, or challenges such as privacy and security of data collected from modern vehicles and hardware and software applications opportunities.
In the next section, a wide perspective on smart traffic is presented. Section 3
is then dedicated to aspects concerning the gathering, fusion and transmission of information. A discussion of the principal application areas is the focus of Section 4
, while Section 5
discusses themes that constitute auspicious avenues for further study. Section 6
concludes the paper.
2. Smart Traffic
We are increasingly becoming digital citizens. However, it is implicit even in the term “citizen” that cities and urban areas are our main ecosystem.
Smart traffic and the intelligent applications associated with it represent an opportunity to optimize resources, encourage sustainable practices, fight inequalities, create new opportunities, and, generally, improve the welfare of people, paving the way for an efficient, technologically advanced, green and socially inclusive city [8
]. The challenges of such an endeavour are multifaceted in the same way, and they include transforming the huge volumes of data generated by sensors and citizens into useful information, deploying and managing limited resources, realizing scalable solutions able to support millions of users, stimulate cooperation and coordination between selfish agents, and effectively communications with citizens with diverse conditions, interests, motivations, and culture.
When the humongous amount of data produced by vehicular traffic is profitably analysed and used, both government and residents will reap the benefits. Traffic information and forecasts will help residents better plan their journeys, while authorities can discipline traffic, effectively deploy officers, and also monitor environmental parameters with the help of air quality and acoustic sensors [9
]. Good traffic estimates can be beneficial for many tasks, such as trip planning, traffic management, road engineering, and planning the construction of new roads [10
]. The benefits associated with less congestion and with a real-time knowledge of alternative, less crowded roads also relate to the financial advantages of optimized deliveries and improved shipping movements [11
]. A better flow of traffic is beneficial to public health, with reduced emission of pollutants, and it also alleviates problems related to equity, with respect to the fraction of citizens who do not have access to a car, yet have to bear the consequences (physiological and financial) of car traffic.
Management allows governments, organizations, and decision-makers to make their projects and services more efficient, and coherent with the needs of citizens and customers. The objectives of municipal authorities when embracing smart traffic projects can be various. Some approaches prioritize the anticipation and mitigation emergency situations over an analysis of normal traffic. An example is the city of Rio de Janeiro, which is now equipped with a citywide Emergency Response System [12
]. Besides allowing priority mechanisms for emergency vehicles like ambulances and fire trucks, thus improving the response to accidents and crises, authorities can also gain actionable insights by looking at details of past road mishaps. Knowing the circumstances under which an accident occurred and the speed of the vehicles involved can help reduce the number of accidents. A detailed monitoring of the roads can also be helpful in detecting stolen vehicles. A related area of interest is the increasing willingness of city authorities to incentivize usage of public transportation instead of private cars. Traffic lights may adapt their behaviour in accordance with the presence or absence of public transport vehicles. The home-office commute is an ideal target in this pursuit, keeping into account that in this context cars remain parked 95% of the time [13
2.1. Utopian Visions and Critiques
Infrastructures are critical elements for the everyday lives of people. Endowing infrastructures with sensors and actuators capable of collecting information and act adaptively and timely has a tremendous potential of accommodating the needs, demands, and desires of citizens. In the words of Kitchin, it would be possile to “optimize resources, plan preventive maintenance activities, and monitor security aspects while maximizing services to citizens” [14
]. Townsend goes to greater lengths, affirming that we are facing “an historic shift in how we build and manage cities”, comparable to the “laying of water mains, sewage pipes, subway tracks, telephone lines, and electrical cables” [15
]. However, risks exist, as pointed out in a provocative essay by Greenfield [16
], who argued that the corporate-driven utopian vision of smart city
is too narrow, centred only on the technological perspective, and it can be detrimental and promote, instead, an undesirable vision of future cites with centralized computational surveillance and control, servicing primarily those in power. Specifically focusing on sensors, Greenfield also underlines that beyond possible malfunctions in operation or improper deployment, measurement of a complex system such as a city is surely going to modify it. Finally, Greenfield warns about the use of easily-measured values as proxies for a reality so complex that it seems to escape attempts at quantification. Regarding data as “transcendent, limpid and uncompromised by human frailty” crystals of perfect knowledge ignores choices made in designing, implementing, and deploying sensors, as well as conscious or unconscious attempts at obfuscation performed by those who are subject to observation or measurement [16
Aspects related to privacy, security and control constitute one of the most diffused fears about smart mobility, especially when citizens become active sensing nodes, or citizen-sensors. In a vision where data coming from traffic sensors is to be integrated with information gathered from smartphones, and the latter also includes human-generated feedback, the underlying assumption is that citizens own a smartphone or even a self-driven one, are digitally educated, and aware of their role [17
]. In this rather homogeneous view, there is little room for minorities characterized by different social positioning and cultural habits. In some areas, most people have no access to a car, let alone a smart one [18
]. Less extremely, traffic information should be quickly, effectively, and accurately delivered to drivers, via smartphones or other communication devices. It is conceivable that vendors try to gain a competitive advantage by providing their customers with more timely, relevant, an precise information, widening the gap between those who have access to their solutions and those who do not. While competition among vendors drives technological improvement, public safety should be protected by defining minimum standards of information quality that should be available to everyone.
The breadth of the “smart” in the notion in smart traffic was also pinpointed by Aguilera et al. [19
], who defined a smart city as a “very broad concept, which includes not only physical infrastructure but also human and social factors”. The concept of smart city is also connected to the notions of smart buildings [20
]. A rapid technological advancement provokes the insurgence of a push to continually release new products into the market, with a fierce competition among technology vendors—and consultancies, too—to secure a niche in a rich and expanding business. Particularly in times of recession, this push can represent an essential opportunity for growth for ICT companies [21
]. As the push is solely based on supply, it contrast with the demand pull, i.e., the solutions researched, devised, and commercialized as a response to needs eplicitly expressed by society [12
]. The two impulses should, as far as possible, reconciled. One effort in this direction is the definition of smart urban mobility
as “connectivity in towns and cities that is affordable, effective, attractive and sustainable” offered by Lyons [22
2.2. Smart Traffic Lights
Smart traffic lights enable a good degree of control of traffic flows within a city or area. This will enhance commutes, reduce congestion, and improve transport systems. Smart traffic lights and signals are important tiles in the overall picture. Usually, in large cities traffic is controlled by traffic lights that use timers while in the suburbs, small towns and county roads, traffic signal sensors are preferred because they not only manage the unsteady traffic flow effectively, but also detect when cars arrive at intersections, when several cars are stacked at an intersection and when cars have entered turn lanes. Smart traffic lights might be triggered by sensors of different technologies like microwave radar, cameras, induction loops, or lasers that detect motion. Smart traffic lights can switch the light signal depending on the workload of the roads, thus eliminating needless delays at intersections where load is unevenly distributed across the roads. Another switching parameter of the traffic light can be the measured level of pollutants—carbon dioxide, nitrogen dioxide, particulate matter—thus improving air quality [23
]. More than a quarter of CO
emissions are due to transportation, and road transport contributes around 65% of it [24
The chart in Figure 2
is based on information from the European Automobile Manufacturers’ Association (ACEA) and the International Monetary Fund (IMF). It depicts the number of passenger cars from each European country versus the gross domestic product per capita (GDP). Almost all points lie in the upper left part of the chart, showing a trend to own more cars in proportion to what could be expected in proportion to the GDP. The total number of vehicles in use in the European Union grew by over 6.28% between 2012 and 2016, and by more than 2.1% from 2015 to 2016 [25
]. The global number of cars on the roads is estimated to nearly double by 2040 [26
]. Table 1
, compiled with data from ACEA, the International Organization of Motor Vehicle Manufacturers (OICA, http://www.oica.net/category/vehicles-in-use/
), and [26
], contains the overall number of vehicles, including passenger cars, trucks, and buses, in Europe and all over the world.
Using the Greenhouse Gas Equivalencies Calculator https://www.epa.gov/energy/greenhouse-gas-equivalencies-calculator
from the U.S. Environmental Protection Agency, the emissions produced yearly by vehicles from entire world for the years 2015, 2025 and 2040 could be converted for comparison in other units, such as the annual emissions by households. The pollution produced by cars is equivalent to the yearly energy consumed by 620.4 million houses (2015), 846 million houses (2025) and 1128 million houses (2040). All the above data highlight the danger created by emissions and the necessity to devise smart solutions for traffic congestion.
Provided that air quality measurements of high granularity are available, focused restrictions for local situations can be envisaged—e.g., forbid road traffic only in a specific area instead of the whole city center. In addition, if drivers could be notified dynamically about restrictions—via an app, but ideally through direct communication with the onboard GPS navigator—vehicular flow could be acted upon not only based on traffic conditions but also on environmental variables [27
Interconnected across a city, sensors detect different parameters relative to the traffic flow, for example waiting time at the lights, density, or speed. Based on the readings, a system can make informed decisions and instruct the traffic lights and signals appropriately. Given that the more data are available to such a system, the more complete and integrated view it will have, it would be preferable to collect data from all traffic lights across the city, requiring the use of big data analytics [11
]. Real-time data are useful to calibrate models to predict traffic patterns. In fact, random variations in traffic due to the stochastic variability are not reflected in one-day traffic counts. The effectiveness and potential of decentralised, adaptive traffic signal control systems has been proved by a pilot implementation for a nine-intersection road network in Pittsburgh, Pennsylvania (USA) [28
2.3. Intelligent Transportation Systems
An Intelligent Transportation System (ITS) is able to improve the efficiency and safety of transportation, at the same time lightening the environmental impact [29
]. ITS, comprehensive systems augmenting conventional transportation infrastructures with technology, are believed to have a great potential in transportation management [30
]. ITS are cyber–physical systems (CPS), with a high level of integration between the cyber components and the physical components. The cyber part consists of communication, information collection, control mode, collaborative algorithms, whereas the physical part includes different kinds of sensors, basic infrastructures, and on-board computers and controllers. A multimodal ITS can integrate data from cellular networks and GPS probes to estimate vehicle speed, space occupancy, and congestion [31
]. A smart traffic infrastructure can also support autonomous and semiautonomous vehicles, whose predictive trajectory guidance systems must be able to withstand dynamic environments [32
]. Visible light communication technology can also be the basis for an ITS, as shown in [33
In an ITS , the control strategy to alleviate traffic congestion is essential. Huang et al. presented a specific policy, with simulations results showing its effectiveness [34
]. An ITS can be used to support the planning for urban evacuation under emergency conditions, also providing the basis to a quantify risk reduction in the transportation system [35
]. It was pointed out that the information trasnferred in ITS may be used to extract individual moving preferences, violating privacy, but this aspect has been seldom taken into consideration by designers of the CPS for ITS [36
In 2016, the British research firm Juniper Research established a hierarchy regarding smart cities in the world starting from different criteria like adoption of smart grid technologies, intelligent lighting, using of intelligent tools to improve traffic, Wi-Fi access points, smartphone penetration, and the app landscape for smarter travel (https://www.iotworldtoday.com/2016/05/18/world-s-5-smartest-cities/
). Table 2
summarizes some points.
The wide thematic area of smart traffic has generated several fertile research lines, each focused on a specific aspect, and the interchange between areas has been limited. Works focused on the technological issues have paid little attention to managerial and sociological considerations. However, a purely technological approach is arguably inadequate to monitor and control an extremely complex system as vehicular traffic, whose dynamics are determined by an intricate reticule of interactions involving multiple actors at different levels. A survey aiming at collecting the fundamental ideas and the associated challenges from the array of works in the literature has been provided in the previous sections, in an attempt to keep a multidisciplinary perspective facilitating a global interpretative reading of the changes ahead. On the other hand, this mindset inherently implied that comprehensiveness of coverage and thoroughness of technical detail in each single topic had been forcefully sacrificed, and several interesting and significant papers have surely been left out. Several open problems that deserve research and analysis have been outlined, embracing sustainability, co-opetition, forensics, and insurance. Directions to be explored for further study include technical and organizational challenges, and future research will arguably attain better outcomes if both aspects will be kept in mind jointly, rather than considering them in isolation.