A Review and Analysis of IoT Enabled Smart Transportation Using Machine Learning Techniques

This section establishes a fundamental comprehension of the principal technologies that underlie smart transportation enabled by the Internet of Things (IoT), with a focus on clarity and direct relevance to the study. Table 1 compares different Smart transportation system that uses IoT.

Table 1. Smart transportation system using IoT

• Foundations of IoT: Smart transportation relies on the robust infrastructure provided by the Internet of Things (IoT). The components of IoT—sensors, actuators, and network connectivity—are succinctly delineated, emphasizing their respective roles in the real-time collection of data and decision-making within transportation systems.
• Communication Technologies: Crucial for ensuring seamless communication in smart transportation, technologies such as RFID, NFC, and WSNs play a pivotal role. The descriptions of these technologies are refined to underscore their specific functions in facilitating both vehicle-to-infrastructure and vehicle-to-vehicle communications.
• Data Processing and Machine Learning: The utilization of Machine Learning (ML) to process extensive datasets generated by IoT devices is of paramount importance. Key ML techniques pertinent to transportation, including predictive analytics for optimizing traffic flow, are outlined concisely.

Organizations get more data when they add more devices. These data are analyzed by new applications, which draw conclusions and make recommendations. This results in the use of Machine Learning (ML) techniques to create Artificial Intelligence (AI) [9]. Many IoT technologies are described in the literature, including:

2.1 Radio-Frequency Identification (RFID)

Readers and tags work together in RFID, which is a very important part of the IoT system [10]. Giving everything in a company a unique digital name helps with automatic identification. The real world and the internet world are linked by this complicated network. RFID can be used in smart systems, to keep track of things, in healthcare, to manage parking, and to protect valuables. It doesn't just store data that you enter by hand; it's also a good monitor that gathers data about its surroundings.

2.2 Near Field Communication (NFC)

Near Field Communication (NFC), visible in smartphones, offers two-way communication across short distances, notably in smart cities [11]. NFC, which operates within a few millimeters, transforms electronics, notably smartphones, into digital wallets. This revolution lets the smartphone do more than simply authenticate your identification; it now manages financial transactions, regulates access, and trades data. The two-way nature of NFC gives up numerous options. For example, by positioning NFC nodes strategically at home, you can set up coordinated actions like turning on Wi-Fi immediately when you enter, illustrating how NFC may substantially affect how devices interact with one other.

2.3 Low-Rate Wireless Personal Area Network (LWPAN)

Low-Rate Wireless Personal Area Network (LWPAN) is a sort of efficient wireless communication that operates over small distances yet may span a vast range, up to 10-15 km [9]. One of its major benefits is how it saves energy, keeping sensor networks operational for approximately a decade on a single battery. When it's paired with IEEE 802.15.4, it creates a cost-effective means for devices to interact. LWPAN is intended for the physical and medium access levels and effortlessly links with other technologies like 6LoWPAN and ZigBee [12]. This makes it particularly critical for developing powerful and long-lasting sensor networks.

2.4 Wireless Sensor Networks (WSNs)

Wireless Sensor Networks (WSNs) are like helpful assistants utilized in numerous domains including healthcare, governance, and keeping an eye on the environment [13]. When they operate along with RFID systems, they grow better at gathering data by demonstrating where objects are, how they move, and what temperatures are like. Each wireless sensor contains pieces like a communication item, a converter altering information, a lot of sensors, a memory, and a power supply [11]. It's built up in a manner that various sensors may be utilized, transforming the measurements they collect into digital data. To communicate this material, there's a system with memory, a little computer, and a method to speak wirelessly, all powered up to keep operating properly.

2.5 3G and Long-Term Evolution (LTE)

3G and Long-Term Evolution (LTE) are like large waves in wireless communication, enabling robust connectivity for dozens of things [9]. This type of link spreads everywhere, even to locations that aren't highly developed. While 3G and LTE thrive in large-area Wide Area Networks (WANs), their appropriateness reduces in short-range settings. Financial concerns for data supply by service providers, as well as the conceptual complexity of permitting communication across various devices, are challenges that must be carefully considered.

2.6 IoT applications areas

The landscape of Internet of Things (IoT) applications has already left its mark on various aspects of the “smart city” concept. Figure 3 represents the design of a smart city using IoT concepts. These applications may be effectively classified by delineating them into six different categories [11]:

1. Smart Homes: This category includes standard domestic appliances, such as refrigerators and washing machines, as well as light fixtures, that have experienced a revolutionary development to achieve inter-device communication and network connectivity. This capability provides not only enhanced control and administration of these devices, but also results in energy usage optimization. Beyond the confines of traditional domestic utilities, emerging technologies have spread, giving rise to intelligent home aids and inventive inventions such as smart door locks.

2. Health-Care Facilitation: In the field of medicine, new technologies have had a huge influence on aiding patients. For example, there are specific bandages with microscopic sensors that keep an eye on how wounds are doing. These sensors provide critical information with physicians even when they're not right there. There are plenty of sensors, some you can wear and others that are inserted within you, that can monitor things like your pulse, how much oxygen is in your blood, the sugar in your blood, and even your body temperature.

3. Mobility Ecosystems: Sensors used in transportation are highly beneficial. They're installed in automobiles or even on phones and in city objects like streetlights. This helps in lots of ways including giving you the best route to travel, making it simple to reserve a parking place, making streetlights smart, boosting public transit, reducing accidents, and making self-driving vehicles conceivable.

4. Environmental Surveillance: When we deploy lots of wireless sensors in cities, we acquire heaps of knowledge about the environment. These sensors can monitor things like air pressure, humidity, and wind speed. They join together to build weather stations, although these are more sophisticated. Smart sensors also let us evaluate how clean the air is and whether there's pollution in the water in major cities.

5. Streamlined Logistics and Supply Chain Governance: Smart usage of special tags called RFID tags helps manage things from where they're manufactured to where they're sold. This notion offers a lot of benefits, notably in saving time and money. Furthermore, the scope of intelligent packaging reveals a variety of capabilities such as trademark protection, quality assurance, and the tailoring of customer interactions.

6. Improved Security and Surveillance Paradigms: Intelligent camera deployment pervades the urban environment, where the convergence of real-time visual object identification crystallizes into a strategic tier of security systems. These technologies demonstrate their worth by identifying suspects and preventing dangerous situations, so advancing the cause of urban security and preservation.

3.png

Figure 3. IoT-enabled smart city architecture

Smart transport is a smart city application that has piqued the interest of academics because of its ability to drastically revolutionize modes of transportation for both people and commodities. The incorporation of IoT technology into transportation systems gives various benefits for motorists in an intelligent urban environment, including improved traffic management, optimized logistical operations, efficient parking infrastructures, and increased safety measures. Smart transport is the incorporation of all of these benefits into transport system applications. The study reviews current research that is based on these areas and gives a full overview of the advances in the applications described previously. The authors also investigate the communication protocols that enable intelligent transportation, such as Wi-Fi, Bluetooth, and cellular networks, as well as their function in allowing seamless data interchange. The study dives into the various smart transportation systems and frameworks, such as cloud computing, edge computing, and fog computing. Overall, the study presents a thorough examination of the technology and applications employed in smart transportation systems, as well as their potential to increase transportation efficiency and safety in smart cities [14].

A thorough analysis of today’s smart transport networks is provided in the section that follows. As shown in Figure 5, our analysis of these systems divided them into six separate groups, each of which defined a unique functional domain. Figure 6 offers a comparison study, contrasting the application of various Machine Learning algorithms along with IoT in the context of smart transportation applications. These systems’ systematic classification helps provide a structured understanding of the many typologies present in smart transportation domains. Our understanding of the complexities of smart transport systems is improved by this comparative demarcation, which permits a detailed investigation of their distinct features.

5.png

Figure 5. Different applications of smart transportation using IoT

6.png

Figure 6. Performance comparison of different approaches using Machine Learning and IoT for smart transportation

Smart transportation solutions that use IoT and Machine Learning are critical to revolutionizing urban mobility. This section goes into these systems, relating each feature directly to our study aims and identified gaps in the literature.

• Traffic Control Systems: IoT-enhanced modern traffic management provides real-time data analysis for effective traffic flow. The research fills a knowledge gap on how ML algorithms might further enhance these systems, particularly under dynamic traffic conditions that are frequently disregarded in previous studies.
• Intelligent Parking Solutions: Smart parking facilitated by IoT minimizes congestion and pollutants. Unique ML-based methods to parking management are investigated, a topic that is frequently treated in isolation in the literature, thereby offering a holistic viewpoint critical for complete urban mobility solutions.
• Transportation Smart Lighting Systems: The coupling of smart lighting with transportation networks is a novel and uncharted topic. This study analyzes how IoT and Machine Learning may improve this junction, filling a major gap in existing smart city programs.
• Optimization of Public Transportation: Most research have focussed on IoT applications for individual automobiles. This study, however, widens its attention to include public transit networks. By examining how Machine Learning may optimize the timetables and routes of buses and trains, the research tackles a critical gap in enhancing the efficiency of mass transportation.

4.1 Traffic management and optimization

4.1.1 Route optimization

The issue of traffic bottlenecks is a serious worry in cities, particularly with more automobiles on the roadways. To solve this problem and enhance travel efficiency while decreasing emissions, there's a heavy emphasis on route optimization. This entails selecting the optimal route to a location with the aim of saving time and cutting down on emissions [48]. Scholars have dug into the complexity of route optimization, particularly within the Internet of Things (IoT) framework. Google has innovatively exploited crowdsourcing, a vital part of current service innovation, via Google Maps. This popular software for mobile devices incorporates real traffic information, providing consumers a handy solution since 2009. This includes employing navigational sensors like GPS, accelerometer, and gyroscope in mobile devices. This novel method, unlike traditional traffic monitoring systems, does not rely on fixed sensors [49]. Instead, it uses the aggregate mobility data created by users when their devices anonymously ping position and velocity insights into the digital ether. This wealth of information enables Google Maps to foresee alternate routes that will avoid congested roads.

Afidis at al. [50] delve deeper into the world of vehicular quandaries, conducting a comprehensive investigation into the interconnection of congested traffic, environmental emissions, and user-generated data from Google’s “peak hours” characteristic. The study gear consists of a combination of on-road cameras and automobiles equipped with a Global Navigation Satellite System (GNSS) data logger. The Vehicle Specific Power (VSP) model is used to calculate emissions, which combines with Google Maps’ popular time intelligence. The study corpus reveals a strong association between the crowdsourced “popular times” dataset and emissions, while the future route calls for more data refining, calibrated using an adaptive learning schema to provide fine-grained correlations.

In keeping with the crowdsourcing spirit, the study [51] guides the path towards optimizing routes by utilizing a swarm intelligence algorithm—an antithesis of individualism that celebrates the collective. This proposal embodies a Modified Crowd-Sensing form of the Ant Colony Optimization algorithm, an endeavor resembling ants travelling in unison, pushed by chemical traces. In this case, users’ mobile interactions serve as chemical trails, channeling communication and assisting fellow travelers in wandering towards less busy routes. Further examining the various junctures of route planning, the study [52] sets out on a mission to improve navigation precision, particularly over the final mile to a destination—a juncture frequently fraught with sophisticated maneuvering. Common techniques, such picking the shortest route or the fastest time, don't properly represent the problems of this last section of the voyage. The authors offer a novel approach considering the limits of existing algorithms.

CrowdNavi, a genuine app, illustrates how crowdsourced navigation systems operate. It solves challenges identified in existing navigation systems, notably in the latter half of the route near the destination. While standard navigation delivers basic directions, the last mile sometimes creates obstacles, such locating a drive-through coffee shop near a freeway. CrowdNavi handles this by aggregating actual driving data to analyze local driving trends. Users then obtain crowd-sourced ideas, enhancing their navigation in tough last-mile scenarios.

4.1.2 Smart traffic light control

Controlling traffic at junctions is done using traffic lights to regulate vehicle entrance effectively, particularly in crowded situations. The efficacy of these crossings generally depends on employing sensors to modify the traffic signal sequences. These sensors act as discriminating sentinels, noting areas where traffic concentration occurs as the intersection approaches. When a green light is detected, the length of the green light is dynamically extended, allowing for a larger vehicle passing opportunity. Furthermore, transponders placed inside intersections can be used to provide priority to particular entries, most notably emergency vehicles and public transport, ensuring quick passage. Intersection control techniques strive to maximize throughput while minimizing instances of pausing by arranging the precise synchronicity of traffic signal timings and approaching vehicle velocity. Figure 7 demonstrates an intelligent traffic signal control system using the Internet of Things (IoT).

Tubaishat et al. [53] suggest a novel decentralized architecture for traffic light control that takes advantage of the capabilities of wireless sensor networks. Within the system architecture, this paradigm emerges across three hierarchical strata: the wireless sensor network, the localized traffic flow model policy, and the elevated coordination mediated by the traffic light agents. The Nearest Intersection Control Agent (ICA), which operates at the nucleus, interacts with data gathered from wireless sensors. This data repository contains characteristics such as vehicle count, velocity, and other relevant properties, allowing the flow model of the junction to be derived. Real-time adaptive control of traffic lights, resulting in an enhanced vehicular mobility experience, is a significant achievement. The Intersection Control Agent is responsible for managing greater regions by talking with other agents. Its job entails approved control of traffic lights in this attempt.

7.png

Figure 7. Smart traffic light control system using IoT

The study underscores the necessity for flexible routing approaches and resilient infrastructure to handle the expanding data from traffic management devices. It brings out issues in vehicle-to-vehicle (V2V) communication, asking for additional security measures to protect the safety of communication networks. With the integration of multiple sensors in automobiles, security risks develop, including the potential of illegal control and data manipulation, which may lead to accidents. To handle these issues in the developing area of traffic management and V2V communication, inventive solutions are necessary.

4.2 Safety management

4.2.1 Road anomalies detection

Recognizing road concerns is vital for smart mobility, particularly in spotting problems like potholes and uneven surfaces. This system tries to notify drivers to possible dangers posed by bad road conditions, which may lead to traffic bottlenecks, vehicle damage, and accidents. Addressing this, revolutionary research by Cha et al. [54] established a Convolutional Neural Network (CNN) to recognize concrete fractures in photographs acquired with handheld cameras under varied lighting situations. The CNN, trained on a dataset of 40,000 photos, attained an astonishing 98% accuracy. Notably, it excelled at identifying tiny fractures in low-light circumstances, an area where standard approaches like Canny and Sobel edge detection fail.

As explained in study [55], the strategic combination of knowledge transfer and pre-existing deep learning models appears in the domain of crack damage diagnosis inside Unmanned Aerial Vehicle (UAV) photographs recording civil infrastructure, including a subset of road surface portrayals. The effectiveness of this method was amply supported by results demonstrating a remarkable accuracy threshold of up to 90% for crack identification under real-world situations without the requirement for augmentation or complex preprocessing methods.

4.2.2 Accident detection

Accident detection and prevention are critical objectives in the field of intelligent transportation, given their ability to avoid fatalities. Driver attention on the road may be a powerful preventative measure against accidents. An effective accident prevention device can alert drivers of approaching risky situations, allowing for quick and decisive action. Furthermore, proactive detection of accident-prone areas or real-time problems inside the traffic network adds to a reduction in the number of accidents and traffic congestion. Figure 8 represents an accident detection system using IoT as an application of smart transportation. The benefit of Machine Learning is most seen in the detection of traffic events, outlining patterns that indicate the possibility of future mishaps, and then notifying drivers, allowing for preventative maneuvers.

8.png

Figure 8. Accident detection system using IoT

Kokilavani and Malathi [56] propose a complete IoT cloud infrastructure designed to provide real-time traffic visualization and early alarms, preventing unplanned slowdowns that might lead to accidents. The proposed architecture combines Infrastructure as a Service (IaaS), Platform as a Service (PaaS), and Software as a Service (SaaS), as well as the novel addition of IoT as a Service (IoTaaS). Volunteer automobiles will be supplied with GPS data gathering equipment, which will then be sent to a cloud-based server over a 4G network conduit. The cloud server orchestrates and manages GPS data expertly thanks to the combination of the OpenGTS platform and OpenStreetMaps. Data is designed in SQL format and consolidated within a Distributed MongoDB database for faster examination. The use of Docker containers improves backend scalability, which aligns with the need for quick response times. The temporal metric of system response is critical to the effective realization of this implementation since the suggested approach broadcasts an alarm traversing a distance of 1 kilometre in a mere 120 ms.

4.3 Energy efficiency and environment

4.3.1 Smart lights

The domain of Smart Street Lights (SSLs) gently weaves a vital component of the new smart city scene, congruent with the wide tapestry of smart mobility services. These dazzling sentinels herald a new age of intelligent urban lighting solutions, situated at the crossroads of energy efficiency and dynamic operability. This study [57] is a powerful example of SSL implementation in the account of IoT-driven urban innovation.

The combination of a constellation of sensors—illumination sensor, infrared (IR) sensor, global positioning system (GPS), and wireless connectivity module — allows this orchestration to infuse streetlights with a tapestry of cognitive capacities. The augmentation of streetlights’ reactivity to their surroundings is the litmus test for this integration. Their natural ability to detect heavily inhabited areas results in dynamic regulation of luminance, making high-density areas safer while also conserving energy. The GPS provides a centralised orchestrator with the ability to scan the geographical and operational states of each luminaire, enabling a more rapid restoration regime in the event of a malfunction. The symbiotic relationship between SSLs and the management nexus is maintained by the umbilical thread of the Narrowband IoT (NB-IoT) network, which is tethered to this paradigm. The management fortress, fortified by a slew of Fog nodes, serves as an agora for data aggregation and state validation, orchestrating the harmonic symphony of streetlight coordination. The capacity to remotely manage SSLs under the auspices of the installed administrative dais is an expressive witness to these infrastructural symphonic. The Raspberry Pi, an enigmatic orchestrator, wears the role of microcontroller [56], a modern-day maestro managing the interplay between luminaire and sensors. Light sensors transcend their usual role in this context, signaling the beginning of a new epoch—a record in which solar conjunctions announce luminous onsets and descents. The light repertoire of these lights attunes to the circumambulating presence of passing automobiles and people in a dynamic crescendo, producing an energy-efficient performance.

Through smart street lighting (SSL) [57], the article demonstrates a realistic example of IoT-driven urban innovation. This entails combining numerous sensors onto streetlights, including as lighting and infrared sensors, GPS, and wireless communication modules, to improve their cognitive capabilities. Dynamic brightness adjustment based on detected population density, GPS-enabled centralized orchestration for speedy fault response, and a symbiotic interaction with management systems are notable characteristics. This SSL solution demonstrates how new technology for responsive and energy-efficient urban lighting may be effectively integrated. The study [57] identifies a limitation in the SSL system wherein, in certain instances of a broken bulb, the server erroneously indicates an offline status instead of accurately reporting the broken bulb condition. This highlights a challenge in precisely detecting and communicating the status of individual street lamps within the SSL system.

Tripathy at al. [58] focus their gaze towards a bright nexus intertwined with Wi-Fi hotspots as the scene grows. Each lamp post transforms into a connectivity emissary, a portal for broadcasting various insights to a central web server. The luminous tapestry orchestrates an elaborate ballet, launching light, regulating brightness, or embracing darkness in response to the flux of the surroundings. These light-bearing sentinels’ tendency for transformation extends to their enhanced vestments, which are swathed with cameras and environmental probes. They reign over regions of safety during emergencies and monitor environmental conditions through these sentient extensions, converting formerly commonplace light posts into stewards of urban vitality.

4.3.2 Smart parking

Smart parking system deployment allows for continuous monitoring of arrival and departure times for various cars across the metropolitan environment. The study [59] investigates the deployment of smart parking systems for effective vehicle arrival and departure management in urban environments. The system continually monitors parking facilities by utilizing wireless sensors, considerably contributing to smart parking. Notable contributions include the development of a data-driven parking management system that makes use of vehicle count and velocity data. This data repository is critical for understanding and controlling traffic movement inside parking lots. Real-time adaptive control based on data obtained improves vehicular mobility and maximizes parking space use. Furthermore, data from the smart parking system help to design and build additional facilities in heavily populated locations. This holistic strategy enhances accessibility, shortens the time it takes to find parking, and helps to overall urban mobility.

As a result, rigorous planning is required for the creation of parking facilities capable of successfully accommodating the traffic inflow within various zones [60]. Furthermore, locations with high vehicle density need the construction of unique parking lots to meet the increased demand [61]. As a result, the repository of data created by smart parking systems provides benefits that pervade both customers’ everyday lives and the business interests of vendors inside smart cities. This service’s operating paradigm is based on the synergy of road sensors and intelligent displays, which adeptly leads cars into ideal parking paths inside the metropolis [44]. Figure 9 compares different approaches used for Smart Car parking system using IoT. This service provides a slew of benefits, including faster parking spot finding, which correlates to lower carbon monoxide emissions and less traffic jams, fostering a sense of happiness among city people. Notably, this service is easily integrated into urban IoT frameworks.

Furthermore, the use of short-range communication technologies such as RFID and NFC enables electronic validation of parking authorizations, ushering in a practical mechanism for the granting of parking privileges. This new solution, supported by secure and technologically sophisticated protocols, enables rapid and convenient access to parking facilities.

9.png

Figure 9. Error rate of different approaches used for smart parking management

In spite of the fact that shrewdly transportation frameworks have picked up favour for their capacity to offer quick, user-friendly, and compelling administrations, they are not void of flaws. An essential concern lies within the blockage of remote systems, stemming from the rising utilization of gadgets for the observation and organization of activity, which presents an eminent obstacle to communication. With the acceleration within the number of gadgets, there emerges a comparing request for an adaptable steering convention that enables the assignment and positioning of assets, at the side a foundation able of overseeing and put away broad street activity information [75-77].

10.png

Figure 10. Major challenges faced in IoT for smart transportation

Figure 10 outlines key impediments existing inside this space to think about. Additional problems are around vehicle-to-vehicle (V2V) communication. To protect the integrity, safety, and secrecy of infrastructure communication networks, new certificate management frameworks must be developed [78, 79]. V2V communications are still vulnerable to transmission interferences and other intrusions, for instance, expecting control of vehicle operations or changing transmitted data to associated versatile peers, possibly coming about in collisions. The incorporation of a plethora of sensors and actuators within vehicles has introduced the difficulty of harmonizing data collecting via a uniform procedure. Furthermore, data collection and transmission to network access points (usually Road Side Units) are intertwined. Investigating this necessitates the acquisition of sensor descriptions and settings [80-82].

Although progress in smart transportation systems has been notable, difficulties remain, needing further study and novel solutions. The next sections detail possible future research directions to overcome these difficulties.

• Improved IoT Communication Protocols: As the number of IoT devices in transportation grows, more adaptive and resilient communication protocols are required. Future research might concentrate on establishing smart routing algorithms that adaptively manage and utilize network resources effectively.
• Elevated V2V Communication Security: Keeping vehicle-to-vehicle (V2V) communication secure is critical for the safety and efficacy of smart mobility. Even though it's crucial, V2V still suffers security difficulties and interference with transmissions. To make V2V communication safer, future research should focus on building better systems for handling certificates and enhancing encryption technologies.
• Machine Learning for Predictive Maintenance: Exploring how Machine Learning may anticipate when transportation infrastructure requires repair is a new research subject. Future research might focus on establishing algorithms to forecast maintenance needs, decrease downtime, and increase the dependability of the system.
• Autonomous Vehicle Integration: Research in the future might focus on creating standardized protocols and implementing the required infrastructure upgrades to facilitate the widespread usage of autonomous automobiles in smart transportation networks.
• Data Privacy and Ethical Considerations: Future research should concentrate on building rules and frameworks to address concerns about data privacy and ethical data usage as data collection grows in smart transportation systems.
• Compatibility Among Systems: A significant difficulty is ensuring compatibility among various smart transportation technologies and systems. Future work might focus on developing universal standards and protocols that allow for the smooth integration of various technologies and systems within the smart transportation ecosystem.
• Smart Transportation Sustainability: Future research should concentrate on making smart transportation systems more environmentally friendly. This entails figuring out how to reduce emissions, optimize energy consumption, and incorporate renewable energy sources into transportation infrastructure.

The Smart mobility provides rich data insights, it has substantial hurdles in protecting data privacy and security. Cyberattacks represent a significant danger, putting critical information and even people's lives at risk. Data aggregation challenges arise when integrating smart devices into transportation owing to diverse formats and a lack of specified techniques. Interoperability is difficult to achieve, particularly in places with minimal technical competence [83, 84]. Furthermore, high implementation costs and technical complications are impediments to acceptance, especially in smaller cities and rural areas. Keeping smart transportation running smoothly requires competence in data analytics, artificial intelligence, and Internet of Things (IoT) technologies.