Publish/Subscribe-Middleware-Based Intelligent Transportation Systems: Applications and Challenges

Abstract

Countries are embracing intelligent transportation systems (ITSs), the application of information and communication technologies to transportation, to address growing challenges in urban mobility, congestion, safety, and sustainability. Architecture Reference for Cooperative and Intelligent Transportation (ARC-IT) is a notable ITS framework comprising Enterprise, Functional, Physical, and Communications Views (or layers). This review focuses on the Communications View, examining how publish/subscribe middleware enhances ITS through the communication layer. It identified application areas across ITS infrastructure, transportation modes, and communication technologies, and highlights key challenges. In the infrastructure domain, publish/subscribe middleware enhances responsiveness and real-time processing in systems such as traffic surveillance, VANETs, and road sensor networks, especially when replacing legacy infrastructure is cost-prohibitive. Moreover, the middleware supports scalable, low-latency communication in land, air, and marine modes, enabling public transport coordination, cooperative driving, and UAV integration. At the communications layer, publish/subscribe systems facilitate interoperable, delay-tolerant data dissemination over heterogeneous platforms, including 4G/5G, ICN, and peer-to-peer networks. However, integrating publish/subscribe middleware in ITS has several challenges, including privacy risks, real-time data constraints, fault tolerance, bandwidth limitations, and security vulnerabilities. This paper provides a domain-informed foundation for researchers and practitioners developing resilient, scalable, and interoperable communication systems in next-generation ITSs.

1. Introduction

A U.S. Federal Highway Administration report defines an intelligent transportation system (ITS) as the integration of information technologies, including computing, sensing, and communications, into surface transportation systems, enhancing the efficiency, safety, and environmental sustainability of transportation networks [1]. Also, the United Nations (UN) defines an ITS as encompassing a range of information and communication technologies (ICT) applied to road transport (recent research broaden the definition to all travel modes [2,3]) to enhance safety, security, and efficiency [4]. These systems offer a range of applications, including significant reductions in traffic congestion, real-time traffic monitoring, and contributions to environmental sustainability and energy conservation [2]. ITS employs technologies, such as sensors, video surveillance, digital mapping, variable message signs, and communication networks, to enable dynamic traffic management and support seamless data collection, analysis, and dissemination for data-driven decision making that improves travel time [2]. It is also vital in enabling Automated Driving Systems (ADSs) [5] and integrating Advanced Driver Assistance Systems (ADASs) [6] for enhanced safety, while also providing real-time data crucial for efficient emergency response and informing long-term transportation planning [3].

Figure 1 shows the U.S. Department of Transportation’s reference framework for ITS, called the Architecture Reference for Cooperative and Intelligent Transportation (ARC-IT), formerly known as the National ITS Architecture [7]. ARC-IT provides a common framework for planning, defining, and integrating ITS deployments across all modes [8]. The architecture consists of four interconnected views (layers) that capture all aspects of an intelligent transportation system: Enterprise, Functional, Physical, and Communications Views. The Enterprise View defines the organizations, agency roles, and institutional relationships involved in ITS deployment. It focuses on stakeholders and their high-level interactions, such as how a local transportation authority, a transit agency, and state police might collaborate on traffic incident management. The Functional View acts as the application logic layer that outlines abstract functions or processes the system must perform and their logical data flow interactions, like detecting traffic incidents, processing data, and generating traveler alerts. The Physical View translates these functions onto actual physical objects (subsystems and devices) like traffic signal controllers, onboard vehicle units, and control centers, clearly specifying how these components interface and exchange information (center-to-center, center-to-field, vehicle-to-field). Finally, the Communications View details the underlying communication protocols and data standards that support information exchanges among physical entities, ensuring consistent data interpretation across systems through recognized protocol stacks such as IEEE and ISO standards.

Furthermore, the communication view effectively defines the networking solutions needed to ensure interoperability among the Physical View Objects [9], directly influencing the overall system performance and flexibility. This paper focuses on the Communications View, examining how a publish/subscribe middleware can unify heterogeneous communication devices and protocols to enable seamless data exchange across an ITS. Adopting a publish/subscribe paradigm, such as the Data Distribution Service (DDS) standard [10], can enhance system performance by decoupling the senders and receivers. Thereby ensuring efficient, asynchronous message dissemination, improving bandwidth and computing resources utilization while enabling real-time data flow [11]. The review also discusses the challenges of integrating publish/subscribe middleware into a heterogeneous ITS environment. It provides insights for researchers, ITS practitioners, and policymakers who aim to deploy resilient, high-performance communication infrastructures in next-generation ITS.

Middleware is software that facilitates communication between various applications or systems and serves as a bridge for efficient data exchange. Operating as a layer of abstraction, it enables applications to communicate without delving into the intricacies of the underlying network or hardware infrastructure [12]. In transportation, as mentioned, middleware facilitates communication and updates between components like vehicles and control systems, providing vital data such as actual vehicle positions, deviations from the static timetable, crowdedness information, etc. [13]. In addition, middleware finds diverse applications, such as data and information management, functioning as a repository for dynamic vehicle environment data within a network. This data encompasses sensor data, information exchanged between vehicles, and static data like digital maps [14]. Hadiwardoyo et al. [12] emphasized the role of middleware in addressing disparate operating systems and the necessity for standardized interfaces, providing a unified platform for application interaction across different operating systems.

Additionally, middleware can be deployed in transportation systems, including underground railways, for safety monitoring and control, ensuring efficient data transfer to overcome periodic communication disruptions. As highlighted by Gianuzzi et al. [14], middleware is crucial in tackling real-time data processing and analysis challenges. It enables real-time data processing and supports various applications to enhance road safety and efficiency by effectively disseminating and processing real-time data from beacons, supporting flexible queries, and aggregating data flow to provide drivers with timely and accurate information. Utilizing middleware in transportation is considered essential to modern transformations and an important factor in constructing Smart Cities, as it seeks to achieve effective integration between modes of transportation, improve road planning, enhance the traveler experience, improve fleet management, and achieve environmental sustainability.

Gianuzzi et al. mentioned in their study [14] that publish/subscribe middleware is a type of middleware that simplifies communication between publishers and subscribers in a distributed system. It allows publishers to send messages independently from subscribers, who can express interest in receiving specific message types called topics. Some publish/subscribe systems comprise brokers, which receive messages and subscriptions, filter messages based on subscriptions, and deliver the messages to the appropriate subscribers. Publish/subscribe middleware plays an essential role in transportation by supporting diverse applications. It can provide timely and accurate information about road conditions, accidents, and hazards to drivers, thereby supporting safety applications.

Moreover, the asynchronous and one-to-many characteristics of the publish/subscribe system make it more suitable than other communication paradigms [15]. Publish/subscribe middleware is also helpful in managing platoons of vehicles, analyzing traffic patterns and trends, and supporting context-aware applications in transportation. Hadiwardoyo et al. [12] illustrated that publish/subscribe middleware, such as real-time DDS, can be used for on-board applications in the railway domain. It enables disruption-aware communication with the outside network, ensuring efficient real-time data transfer.

In general, publish/subscribe middleware offers a flexible and efficient communication mechanism for exchanging information in transportation systems. Employing the publish/subscribe model through middleware enables scalability, flexibility, and responsiveness, making it suitable for dynamic and distributed environments like ITS. This review aims to explore and synthesize the existing knowledge surrounding the application of publish/subscribe middleware in transportation, offering a solid understanding of their varied implications and contributions.

This review seeks to analyze papers focused on middleware in transportation, exploring its applications, contributions, employed models and techniques, and associated challenges.

Table 1. Research questions.

RQ#	Question	Section	Objective
RQ1	What is Middleware?	Section 3	Define middleware, explain its core functionalities, and highlight its importance in distributed systems.
RQ2	What is a publish/subscribe Middleware?	Section 3.1	Define publish/subscribe middleware and explore the different publish/subscribe middleware types.
RQ3	What are the different applications of middleware in transportation?	Section 4	Explore and understand middleware technology’s positive impacts and advantages on transportation systems and identify the specific benefits, such as improving efficiency, enhancing communication, and saving lives.
RQ4	What communication models and techniques are employed in transportation systems with the integration of middleware?	Section 4.3	Identify and examine communication models used in middleware-based ITS in the literature.
RQ5	What are the challenges that middleware is facing in transportation?	Section 6	Identify and analyze the obstacles, hindrances, and difficulties that middleware encounters when applied to real-world transportation scenarios.

shows the research questions this research aims to answer. The table has four columns: the first column contains the research questions (RQs) ID, RQ1, RQ2, RQ3, RQ4, and RQ5. Thus, heading RQ# means Research Question Number, where # is a serial number identifying each research question. The second column explains the corresponding RQs. The third column shows the section that answers the RQs, and the fourth presents the objectives of the respective research questions.

The rest of the review is organized as follows. Section 2 discusses the methodology for the review. Section 3 discusses middleware background, the basics of publish/subscribe middleware, and how it is applied to ITS. Section 4 reviews the applications of publish/subscribe middleware in ITS. Section 5 gives a comparative analysis of reviewed middleware applications studied in Section 4. Section 6 presents the challenges of applying to publish/subscribe middleware in ITS. Finally, Section 7 concludes the study.

2. Methodology

This systematic literature review (SLR) aims to identify and analyze middleware research in the transportation domain, focusing on publish/subscribe middleware. To define the scope of this study, we conducted a keyword-based search in the Scopus database, retrieving publications where the terms middleware, publish/subscribe systems, and transportation appeared in the title, abstract, or keyword section. The distribution of retrieved publications is visually illustrated in Figure 2, highlighting the intersection of these three key domains. The shaded area in the Venn diagram represents the 87 publications that mention both “publish/subscribe” and “transportation” in their title, abstract, or keyword sections. Of these, 15 publications also explicitly mention “middleware”, placing them at the intersection of all three terms. We consider all 87 publications relevant in this review, regardless of whether the term “middleware” appears explicitly, because publish/subscribe is a type of middleware. The inclusion of the “middleware” circle in the Venn diagram aims to highlight the relative scale of publish/subscribe research within the broader middleware domain: while over 29,000 papers mention middleware, only 3803 mention publish/subscribe, indicating that publish/subscribe represents a specialized subset of middleware research. This visualization underscores the focused and relatively underexplored nature of applying publish/subscribe middleware to ITS, reinforcing the motivation for this systematic review.

A notable distinction in the Venn diagram is the separation of publish/subscribe from Middleware, as some research publications do not explicitly include the term “Middleware” in their metadata. For example, a study on Message Queuing Telemetry Transport (MQTT)-based vehicular communication may focus on the publish/subscribe model without mentioning middleware directly, even though it fundamentally relies on middleware for message exchange. Also, some publications use the word “system” instead of “middleware” [16]. The shaded region in the diagram represents the subset of research publications aligning with this review’s focus—studies integrating publish/subscribe middleware into transportation applications.

This approach ensures a comprehensive yet targeted selection of literature, capturing relevant peer-reviewed publications while avoiding studies that do not contribute meaningfully to the field. This focused review allows us to provide in-depth insights into the role, challenges, and future directions of publish/subscribe middleware in ITS while distinguishing it from other middleware architectures.

The SLR used the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA [17]) framework. PRISMA is a set of evidence-based guidelines designed to improve the transparency and quality of reporting in systematic reviews and meta-analyses [17]. It provides a structured 27-item checklist and a flow diagram to ensure comprehensive reporting of methodology, data selection, and synthesis processes. PRISMA helps researchers enhance reproducibility, minimize bias, and improve the reliability of systematic reviews in various scientific disciplines. Figure 3 shows the PRISMA SLR flowchart for this review.

The identification stage searches for all publications that may help answer the research questions. The flowchart shows that we used the Scopus database [18] to find the papers for this review. We selected the Scopus database for this review due to its extensive coverage of peer-reviewed literature across multiple disciplines relevant to intelligent transportation systems, its support for reproducible search strategies, robust analytics tools, and comprehensive indexing of publications from all major and reputable publishers. We searched for all publications with the keywords “publish/subscribe” and “transportation” within their titles, abstracts, or keywords. The search term did not include middleware because the Venn diagram in Figure 2 shows that the 15 papers it contributed to are at the intersection of all search terms. Thus, they will be included even without it in the search term. The search yielded 87 publications from 2003 to 2024, with no duplicate entries. However, publications from 2003 to 2008 did not meet our inclusion criteria, which we describe shortly. As a result, the final set of reviewed papers spans from 2009 to 2024. The screening stage removed 4 entries because they are conference proceedings titles, not publications, resulting in 83 papers for further analysis.

At the eligibility stage, we applied well-defined inclusion (in

Table 2. Table of attributes and descriptions.

Attribute	Description
Focus	The papers should focus on publish/subscribe-based middleware implementation in transportation, addressing specific application areas, associated challenges, or potential security and privacy concerns.
Type	We consider all scientific publications, including peer-reviewed articles, conference papers, and review papers.
Relevance	The papers must utilize publish/subscribe-based middleware as a communication channel in transportation applications.
Recency	We included all published papers from 2009 to 2024 to capture historical developments and emerging trends.

) and exclusion criteria (in

Table 3. Table of exclusion criteria.

Attribute	Description
Duplicates	We excluded articles with substantial content overlap to maintain diversity and originality in the review.
Sources	We excluded all non-peer-reviewed materials, such as websites, blogs, and opinion pieces, to ensure the academic integrity of the review paper. However, we use middleware or transportation companies’ and governmental websites for the latest examples in the field.
	MDPI Please check if meaning is retained.
Completeness	We eliminated publications that were either retracted, incomplete, not accessible, or invalid.
Language	We excluded all papers published in languages other than English.

) to filter the publications systematically. Included papers must fulfill all inclusion criteria and none of the exclusion criteria.

The inclusion criteria in Table 2 ensured that only relevant and high-quality papers were selected. Specifically, we included papers focusing on publish/subscribe-based middleware implementation in transportation, addressing specific applications, challenges, or security concerns. We considered all scientific publications, including peer-reviewed articles, conference papers, and review papers, ensuring a comprehensive analysis. The selected publications were required to utilize publish/subscribe-based middleware as a communication channel in transportation applications and fall within the publication timeframe of 2009 to 2024 to capture historical developments and emerging trends. We used this time frame because all the publish/subscribe middleware in transportation papers in the Scopus database fall within it.

The exclusion criteria for this research are shown in Table 3. We eliminated duplicate publications to maintain diversity and originality. We also excluded non-peer-reviewed sources like blogs, websites, and opinion pieces. This criterion ensures academic integrity because unreviewed articles may contain biased, wrong, or false information. We also removed retracted, incomplete, inaccessible, or invalid papers. Additionally, this work considers only publications in English, excluding those in other languages. Lastly, this review excludes all articles that discussed middleware without a direct application in transportation.

These criteria excluded 18 papers based on their titles, as they did not align with the research scope. Furthermore, 25 papers were removed after thoroughly reviewing their abstracts or the complete content, as they did not meet the inclusion criteria. Ultimately, 40 papers published between 2009 and 2024 passed for the final review.

We notice a large number of publications were invalid entries. Investigations showed a high false-positive rate because “transportation” could mean the transportation layer in computer networks or the movement of people and goods. Therefore, distinguishing between these contexts proved challenging without comprehensively examining the abstract or the entire paper.

3. Background of Middleware

Researchers and vendors have defined middleware in various ways. Microsoft describes it as a software layer that facilitates communication and data management between an operating system (OS) and applications, acting as a hidden translation layer that enables seamless interaction across distributed components [19]. Bernstein [20] defined middleware as a bridge between applications and system-level services governed by processor architecture. Manewith et al. [21] emphasized its role in managing communication and data exchange across distributed systems, enabling shared access to databases and services. Bishop and Karne [22] offered a broader perspective, describing middleware as software that abstracts the complexities of distributed systems while facilitating interaction among applications, networks, hardware, and operating systems.

Figure 4 illustrates how middleware operates as an intermediary layer between distributed applications and the underlying system software and hardware [23]. Each application interacts with the middleware through its Application Programming Interface (API), abstracting away the complexity of direct communication with system components such as OSs, drivers, and utilities. This decoupling allows applications to operate independently of specific hardware or system configurations, improving portability and scalability across heterogeneous platforms. For instance, in intelligent transportation systems (ITSs), middleware enables different stakeholders and infrastructure components to use independent platforms while interacting seamlessly. It achieves this by managing data exchange, service invocation, and resource access across multiple distributed nodes.

Researchers have demonstrated several applications of middleware in the transportation sector. Delsi and Amutha [24] used middleware to manage diverse communication protocols and data formats, enhancing device compatibility and interoperability within IoT-based ITS in Smart Cities. Jan et al. [25] reported how middleware facilitates the integration of machine learning (ML) and autonomous vehicles in ITS. Almadani et al. [26] studied DDS applications in Smart Traffic Systems (STSs), including real-time data processing, urban traffic management, vehicle-to-everything (V2X) communication, and Cooperative Intelligent Transport Systems (C-ITSs). However, this paper focuses on the publish/subscribe-middleware-based ITS. Section 3.1 provides background on publish/subscribe middleware.

3.1. Publish/Subscribe Middleware

A publish/subscribe middleware is a software that enables asynchronous and decoupled interactions between message producers (publishers) and consumers (subscribers) suitable for distributed systems [27,28]. In this paradigm, publishers generate messages (events) and send them to all subscribers that express interest in them. This loose coupling between components enhances scalability, flexibility, and efficiency in communication, allowing systems to handle dynamic and high-throughput environments more effectively [28].

Figure 5 shows a typical publish/subscribe-middleware-based network. The figure shows devices communicating with one another through the middleware. These devices can be vehicles, traffic lights, RSUs, environmental sensors, or the command and control center of the traffic department. It also shows that the devices use a publisher to send data and a subscriber to collect data from the middleware. This architecture avoids direct communication between the devices, which decouples them temporally and spatially.

3.1.1. Types of Publish/Subscribe Middleware

Researchers classify publish/subscribe middleware based on how they route, process, and deliver messages. This section explores the various publish/subscribe middleware types. We develop the taxonomy in Figure 6 from existing taxonomies the literature [16,28,29,30]. Each of these classifications impacts performance, scalability, and use-case suitability. The figure classifies publish/subscribe middleware according to the subscription model, architecture, middleware location, and event messaging technique.

In terms of the subscription model, middleware can be push-based, delivering messages as soon as they are published, ideal for real-time applications, or pull-based, where subscribers fetch data on demand, which is better for bandwidth efficiency and historical data access [16,30]. On the system architecture category, middleware can be either brokered, where a central or distributed broker manages message routing and enhances reliability and QoS [28,29], or brokerless, where devices communicate peer-to-peer (P2P) for low latency and improved fault tolerance, especially suitable for resource-constrained environments like VANETs and IoT networks [16]. Classification of middleware by deployment location also matters: on-premise middleware offers greater control and reduced latency, while cloud-based solutions enable scalability and accessibility, especially in large-scale or IoT-heavy systems [30,31]. Lastly, middleware classification by event messaging techniques includes subject-based (by channel, topic, or group), content-based (based on data attributes), and type-based (based on message structure). Each technique affects how subscribers filter and receive information, with content-based middleware offering high relevance but higher processing overhead [28,29,32].

3.1.2. Publish/Subscribe-Middleware-Based ITS

Middleware plays a crucial role in ITS by enabling seamless communication, real-time data exchange, and efficient event-driven processing across various transportation infrastructures. Integrating middleware solutions to ITS becomes essential for ensuring interoperability among heterogeneous systems, optimizing decision making, and enhancing overall system performance. Middleware facilitates data distribution in traffic surveillance, vehicle-to-everything (V2X) communication, Smart Infrastructure, and transportation logistics, providing a scalable and adaptable framework for intelligent mobility solutions.

Figure 7 illustrates how publish/subscribe middleware connects the different views in the ARC-IT framework shown in Figure 1. Each node (Node 1, Node 2, up to Node n) represents a distinct infrastructure element within the ITS, such as sensors, vehicles, traffic lights, Roadside Units (RSUs), or surveillance cameras. These nodes have individual Physical and Communications Views, which may vary based on vendor implementations. The middleware facilitates inter-node communication by interfacing with the transport layer of the Communications View, thereby enabling interoperability across heterogeneous Physical Views. Also, the Enterprise and Functional Views interact with the Physical and Communications Views through the middleware, enabling unified data exchange and service coordination. This architecture simplifies system deployment, as developers do not need to ensure direct compatibility between new and existing components. The middleware also embeds essential services such as QoS and security mechanisms to support reliable, scalable, and secure communication within ITS.

3.1.3. Example of Publish/Subscribe Middleware

Table 4. Prominent publish/subscribe middleware solutions.

Ref.	Middleware Name	FOSS	Developer-Stated Application Domain
[33]	MQTT (Protocol)	✔	Lightweight IoT messaging, embedded systems, telemetry
[34]	Apache Kafka	✔	Distributed event streaming, big data pipelines, analytics
[35]	NATS	✔	Cloud-native messaging, microservices, control systems
[36]	RabbitMQ	✔	General-purpose messaging, enterprise integration, background jobs
[37]	ZeroMQ	✔	Low-latency messaging for embedded, edge computing, and HPC
[38]	Apache Pulsar	✔	Cloud-native streaming and messaging for data analytics
[31]	Google Cloud Pub/Sub	✖	Cloud-native messaging and event ingestion
[39]	RTI Connext DDS	✖	Safety-critical real-time systems: automotive, aerospace, robotics
[40]	OpenDDS	✔	Real-time distributed systems: defense, automation
[41]	Fast DDS	✔	Real-time robotics, autonomous vehicles, industrial systems
[42]	CoreDX DDS	✖	Embedded real-time systems: avionics, defense, automotive
[43]	Vortex OpenSplice	✔	High-reliability embedded systems: marine, aerospace, SCADA
[44]	GurumDDS	✖	Robotics, defense systems, autonomous platforms
[45]	Cyclone DDS	✔	ROS2-based robotics, automation, distributed real-time applications

shows some prominent publish/subscribe middleware with possible applications in ITS. The first column is the source of the middleware, the second is its name, the third shows whether the middleware is Free and Open Source Software (FOSS), and the fourth describes the possible application of the middleware according to the developer. The table also includes the MQTT protocol because it is common in ITS and other embedded systems. Middleware is a complete messaging infrastructure, while a protocol is a communication standard that defines how messages are formatted and exchanged between publishers and subscribers [16]. Thus, middleware solutions often implement one or more protocols.

4. Application of Publish/Subscribe Middleware in ITS

This section explores middleware applications across different ITS domains, including infrastructure, where publish/subscribe architectures and distributed systems improve traffic monitoring; transportation modes, where middleware enhances maritime, land, and air transportation; and communication technologies, where advancements in mobile networks (4G/5G), P2P systems, and Information-Centric Networking (ICN) drive ITS efficiency. ITS can leverage middleware technologies to support real-time data dissemination, improve safety measures, and optimize transportation management, paving the way for next-generation smart mobility solutions.

Figure 8 shows a functional taxonomy of publish/subscribe middleware applications in ITS that we extracted from the publication pool. The diagram categorizes middleware usage in ITS based on the transportation and computer networks paradigm. Our study found middleware applications in all modes of transportation. We also categorize the computer networking aspect of the applications based on the networking infrastructure types and communication technologies. The red boxes represent the main subsections of this section. The green boxes correspond to their respective subtopics, discussing each identified mode, infrastructure type, and communication technology in detail in the following subsections.

4.1. Infrastructure

Middleware is essential for ITS, enabling real-time traffic monitoring, vehicle communication, and smart mobility applications. This section explores key ITS infrastructure components, including IoT-based traffic surveillance, Vehicular Ad hoc Networks (VANETs), and road sensor networks (RSNs), and how middleware enhances them. IoT infrastructure utilizes smart cameras and sensors for event-driven traffic management, while VANETs support Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I) communication to improve road safety. Also, RSNs enhance ITS by providing real-time road condition data, supporting autonomous vehicles, and Smart City integration.

4.1.1. Internet of Things (IoT) Infrastructure

A key finding in our literature review is that all traffic surveillance infrastructure utilizes smart cameras (IoT devices) connected to a network for real-time data collection, processing, and transmission. Middleware solutions facilitate real-time data exchange, event-driven processing, and intelligent decision making, enabling advanced applications such as traffic monitoring, vehicle tracking, and incident detection. Researchers have proposed various middleware-based approaches integrating publish/subscribe messaging, distributed systems, and machine learning techniques to improve visibility, surveillance accuracy, and automated traffic management. This section explores how IoT-driven middleware has been applied to real-time traffic monitoring and ITS, highlighting its benefits, limitations, and evolving capabilities.

Zhang et al. [46] designed a system to improve highway visibility, monitoring, and warnings utilizing video camera facilities in China. It provides real-time visibility data through image analysis and data fusion processes, which are transmitted and shared between the lower-end and upper-end subsystems using communication networks such as optical fiber and the internet. The lower-end subsystems capture and analyze video data for visibility information, and the upper-end was built on the J2EE platform to provide essential services such as data transmission, data fusion, monitoring, alert, and data publish/subscribe mechanism.

Cao et al. [47] utilized real-time publish/subscribe middleware for wireless video surveillance of ITS. The authors designed it for traffic jam monitoring and vehicle tracking in transportation. The system uses wireless sensor nodes for event detection and a Wireless Mesh Network (WMN) for transmitting video streams. In terms of strengths, firstly, it reduces deployment expenses due to the wireless infrastructure, supports mobile users, and provides flexibility in defining and subscribing to events. Secondly, real-time data exchange is critical in ITS as it enables quick decision making, which improves traffic management, safety, and efficiency.

Furthermore, Kul et al. [48,49] used Apache Kafka version 2.2, a distributed publish/subscribe messaging system, for real-time vehicle tracking using traffic monitoring cameras. In [48], it uses an Event-Based Microservice framework with synchronous and asynchronous communication. The microservices detect vehicles, extract their type, color, and speed features, and store them in a metadata repository. After that, the system publishes these features as events while other microservices subscribe to them. This results in a combination of all possibilities (type-color, type-speed, color-speed, type-color-speed) for vehicle identification. The system can return the best-matched vehicles based on user-selected features without re-processing the videos. Real-time information filtering is supported, and new microservices integration with the current system is possible. The system’s strength comes from utilizing traffic cameras to optimize vehicle tracking and monitoring and hybrid communication using both synchronous and asynchronous methods for versatility. Its weaknesses come from requiring constant data from traffic cameras, which means dependence on continuous data flow. Also, a large amount of vehicle data might lead to processing challenges.

Recently, the system was improved using a docker-container-based framework of the Apache Kafka node ecosystem for vehicle tracking [49]. The system follows three steps: (1) use You Only Look Once version 3 (YOLOv3) [50] to localize the license plate, (2) use Tesseract OCR character recognition with Optical Character Recognition (OCR), and (3) transmit the detected license plate using Kapka. This process increases scalability and improves user experience by allowing users to textually search the license plate and start and end time while Kapka retrieves the visual data and the system OpenCV library to compile and produce a video of the event. The system enables the users to easily play videos of events involving a car with a given license plate during a precise time interval. However, the system will have a high initial investment cost due to its high processing power requirements.

Researchers also investigated the use of fog computing to improve ITS-based services. Fog computing places computing devices near end-user devices to process data locally, thereby minimizing latency compared to cloud-based processing [51]. Liu et al. [52] discussed a novel joint design of a publish/subscribe (pub/sub) communication model based on vehicular fog computing (VFC) architecture. It utilizes fog nodes as the data platform for message aggregation and employs a message dissemination approach based on the communication model. Considers the interplay between the pub/sub system and fog computing, combining these two research fields to support efficient and reliable communication in the Internet of Vehicles (IoV).

Middleware in traffic surveillance has demonstrated its significant impact on real-time data processing, event-driven decision making, and system scalability in Smart Traffic Infrastructure. After reviewing the literature, Apache Kafka became the authors’ favorite because it is well-suited for handling high-throughput, real-time event streaming, which is critical for vehicle tracking and traffic monitoring systems. We also found that machine learning (ML) is an integral part of the Smart Traffic surveillance system, helping in image detection, identification, or converting to text for easy transmission. However, we also found that challenges such as high computational demands, reliance on continuous data streams, and potential processing bottlenecks must be addressed to ensure system efficiency and scalability.

4.1.2. Vehicular Ad Hoc Networks (VANETs) Infrastructure

VANETs are composed of mobile nodes (vehicles) equipped with on-board units (OBUs) and often fixed nodes in the form of Roadside Units (RSUs) [53,54]. Figure 9 shows a typical VANET system. They facilitate both Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I) communications, commonly using protocols such as IEEE 802.11p to enhance traffic safety and efficiency [53]. The network topology is highly dynamic, with vehicles forming a mesh topology that continuously reconfigures itself as they move without the help of a central coordinator. Also, the nodes relay messages hop-by-hop to enable multihop reach beyond direct wireless range [54]. RSUs help bridge vehicles to broader networks, traffic management centers, or the Internet [54]. VANETs are integral to ITS, supporting applications like passenger safety, traffic management, and road condition monitoring. They also leverage IoT technologies to improve transportation efficiency, safety, and comfort, ultimately playing a key role in developing autonomous vehicles [53].

Researchers stress the importance of VANET and middleware in improving ITS performance. Patil et al. [55] explored reliable communications within VANETs, a key component of ITS. Its strengths lie in improving real-time safety alerts, traffic management efficiency, flexible communication modes, and adaptability to various traffic conditions. They highlight challenges in VANET communication, emphasizing high mobility, unstable network conditions, and heterogeneous communication technologies that affect reliability and performance. To address these issues, the study suggests multipronged communication models that integrate Vehicle-to-Vehicle (V2V), Vehicle-to-Infrastructure (V2I), and Vehicle-to-Broadband Cloud (V2B) communication, ensuring seamless data exchange between vehicles, RSUs, and cloud-based services. The authors argue that middleware solutions are critical in managing heterogeneous networks, facilitating adaptive switching between Wi-Fi, cellular, and dedicated vehicular networks (802.11p and WAVE) to maintain stable and low-latency communication. However, it faces challenges such as the complexity, cost of implementation, and potential security and privacy issues.

Chen et al. [56] proposed V-ICE, a publish/subscribe-based communication model for timely black ice warnings in ITS. It leverages Information-Centric Networking (ICN) and a hierarchical namespace architecture. It also uses Roadside Units (RSUs) as intermediaries to ensure vehicles subscribe to relevant road segments and receive updates when black ice is detected. The study introduces V-ICE-R (an extended version of V-ICE) that integrates the Content-Oriented Publish/Subscribe System (COPSS) [57] middleware to enhance reliability and timeliness, ensuring notifications are received even when vehicles temporarily lose network connectivity. Simulation results demonstrate that V-ICE-R significantly outperforms server-based and broadcast-based approaches, reducing the percentage of cars encountering black ice without prior notification to 0.32% in a 30-min event scenario and 1.45% in a high-frequency 6-min scenario, compared to 19.30% and 66.04% for server-based solutions, respectively. V-ICE-R also minimizes network congestion, generating 145,758 messages in the 30-min scenario. However, the server-based polling produces 222,241 messages, and the broadcast-based flooding results in 5.99 million messages, demonstrating the efficiency of event-driven messaging in reducing redundant data transmission. The system also employs message filtering mechanisms at RSUs to suppress duplicate reports from multiple vehicles, optimizing network traffic and system performance. Through SUMO-based simulations in Luxembourg, the study validates that publish/subscribe middleware provides a scalable, low-latency, and reliable solution for vehicular safety applications, significantly improving network efficiency compared to traditional ITS communication models.

However, VANETs have one challenge that middleware alone cannot solve: the high infrastructure cost of deploying RSUs across large geographic areas to ensure reliable communication. For example, the simulation setup for VANET in Luxembourg deployed 2247 RSUs at junctions, with an additional 191 edge RSUs placed at 500-m intervals on highways to ensure complete coverage [56]. Such a dense RSU deployment translates to substantial initial investment costs, including installation, maintenance, and connectivity expenses. A cost-effective alternative is leveraging existing infrastructure (e.g., traffic lights and streetlights) already strategically placed at intersections by enhancing them with embedded systems and lightweight publish/subscribe middleware. The Intelligent Traffic Light (ITL) event based publish/subscribe system demonstrates how traffic signals equipped with wireless communication and a real-time publish/subscribe model can relay safety messages to vehicles, reducing the need for additional RSUs [58]. By enabling vehicles to subscribe to traffic light alerts, ITL ensures real-time dissemination of critical warnings, such as red light running alerts and required deceleration notifications. Also, separating publishers from subscribers allows for greater system scalability and a more dynamic network topology. This approach minimizes infrastructure expansion costs while enhancing vehicular communication and road safety.

Alternatively, the vehicles can cooperate and provide service to one another using V2V communication. Mousannif et al. [59,60] proposed a brokerless service-oriented middleware solution called Cooperation as a Service (CaaS) designed for VANET and Vehicular Sensor Networks (VSNs) using content-based routing (CBR) for intra-cluster communications and Delay- and disruption-Tolerant Network (DTN) routing for inter-cluster communications. By leveraging V2V communication, it aims to provide free services, increase the quality of communication, and significantly enhance the driving experience by facilitating more efficient and service-oriented communication between vehicles without adding extra infrastructure. At its core, CaaS enables vehicles to cooperate, sharing valuable information for all road users. The architecture utilizes a hybrid publish/subscribe mechanism, where drivers express their interests in specific services, and vehicles that have subscribed to the same service cooperate in providing the necessary information to one another. By leveraging V2V communication protocols, CaaS ensures a seamless exchange of information. Also, the architecture boasts high delivery ratios. However, the system has a high complexity of implementation and maintenance. Also, the system requires at least two cars for communication, which is only possible in cities but seldom happens on highways.

Researchers use publish/subscribe middleware in VANET to improve the reliability, scalability, and responsiveness of the ITS. The solutions reviewed in this section use COPSS for content-oriented safety alerts [56], CaaS for decentralized V2V service exchange [59,60], and the ITL system for infrastructure-integrated messaging, demonstrating how custom middleware can address specific challenges such as network congestion, infrastructure cost, and real-time coordination. Notably, all the middleware implementations in this section are custom-developed to meet the unique demands of their respective use cases rather than relying on standard platforms like DDS or MQTT. This section emphasizes the need for domain-specific middleware architectures in VANETs, where factors such as mobility, intermittent connectivity, and deployment constraints require specialized design.

4.1.3. Road Sensor Network (RSN) Infrastructure

A Wireless Sensor Network (WSN) consists of resource-constrained devices with limited computing power, memory, and energy designed to perform fine-grained sensing. Figure 10 shows a typical WSN. Each WSN node typically comprises a microcontroller, a power source, and a radio transceiver for communication [61]. These nodes are generally classified into four types to facilitate efficient resource management: sensor nodes (SNs) for environmental monitoring, actuator nodes (ACNs) for altering the monitored environment, relay nodes (RNs) for forwarding data, and a base station (BS) for linking the WSN to external networks. The collective name of SNs and ACNs is end nodes (ENs) [61]. The RNs transfer data between ENs and the BS, while the BS collects data from the SNs and relays instructions from the network administrator to the SNs or ACNs.

Road sensor networks (RSNs), a specialized subset of WSNs designed for ITS, consist of stationary sensor nodes embedded in or alongside roadway infrastructure. These networks form a multihop architecture to facilitate the real-time acquisition, transmission, and processing of traffic-related data, including vehicle movements, road conditions, and environmental parameters, which it subsequently relays to a central management system through the BS [62,63]. These networks play a crucial role in traffic management and Smart City applications, utilizing various sensor types such as in-pavement sensors, fiber optic sensors, and WSNs to collect data on traffic volume, vehicle speed, and road roughness [63,64]. Due to the impracticality of deploying wired links to every roadside sensor, most on-road sensors communicate wirelessly with designated gateway nodes, often employing a cluster-based architecture where local cluster heads aggregate and transmit data via long-range communication methods such as cellular 4G/5G or optical fiber [62]. The topology of these networks is typically hybrid, combining a linear multihop structure along roadways with a star topology between cluster heads and the central system [62]. Beyond traffic monitoring, RSNs contribute to the advancement of autonomous vehicles, providing essential real-time information on road hazards and conditions to enhance safety and operational efficiency [63]. These networks can dynamically adapt to changing traffic patterns, enabling real-time updates and intelligent decision making for effective road traffic management. Their integration into Smart City infrastructure enhances urban mobility, resource allocation, and overall road safety [63].

Regarding the utilization of sensors, Li et al. [65] presented a parameterized spatio-textual publish/subscribe system for RSNs, focusing on ITS where large volumes of geo-textual data arrive continuously (e.g., sensor readings, events, or alerts). Unlike typical ITS middleware solutions that only support basic keyword or Euclidean distance filtering, this system includes actual road distances. It also allows subscribers to specify their interests with weighted keywords, specific locations, and thresholds for textual match and spatial proximity. The authors developed and compared two indexing architectures: Basic Indexing Architecture (BIA) and Enhanced Indexing Architecture (EIA). The BIA employs a quadtree for spatially organizing road segments, augmented with inverted lists of subscription keywords and tables storing minimum/maximum distance bounds. The EIA improves BIA by adding “early pruning” techniques and calculating tighter distance thresholds based on partial keyword matches, aggressively ruling out subscriptions or road networks, thereby reducing the time spent on false matches. The authors simulated the two indexing methods on the San Francisco Bay road network. They found that EIA runs significantly faster, sometimes as low as 17.6% of BIA’s computation time. However, the filtering process is computationally intensive, especially as the system is scaling up.

Thus, RSNs are crucial in ITS by enabling real-time traffic monitoring, road condition analysis, and Smart City integration. As a specialized subset of WSNs, RSNs leverage multihop communication architectures to efficiently collect and transmit traffic-related data using a combination of sensor nodes, relay nodes, and base stations. Their hybrid topology ensures scalability and adaptability to dynamic traffic patterns, enhancing urban mobility and road safety. The use of publish/subscribe middleware further optimizes data processing and event-driven communication, as demonstrated by Li et al.’s [65] parameterized spatio-textual publish/subscribe system, which significantly improves filtering efficiency for high-volume, geo-textual sensor data. However, network infrastructure constraints, energy efficiency, and real-time data processing challenges remain key areas for further optimization. Integrating RSN with edge computing, 5G connectivity, and AI-driven analytics will be essential for advancing autonomous vehicle technologies and intelligent traffic management solutions.

4.2. Transportation Mode

Researchers used publish/subscribe middleware to improve ITS across different transportation modes: aviation, land, and marine. This section discusses various systems that use middleware to achieve two main goals: ensuring passenger safety and providing the best travel experience, which includes improved comfort and convenience, faster and more reliable travel times, and real-time updates.

4.2.1. Marine Transportation

Maritime intelligent transportation systems (MITS), or e-navigation systems, apply real-time data collection, communication, and automation to enhance vessel traffic management, port efficiency, and environmental sustainability [66]. These systems improve navigation safety through Vessel Traffic Services (VTSs), collision avoidance, and route optimization while optimizing port logistics with IoT-enabled smart ports and automated terminal systems. For example, in [67], the authors developed Marine Sensor Networks (MSNs) for monitoring ocean conditions. In marine transportation, using middleware is essential for enhancing operational efficiency and safety, which evolves to meet modern maritime challenges and improve navigational precision, thereby becoming more reliable for global marine operations.

A notable contribution in this area is the work of Bingham et al. [68]. They proposed a supervisory control framework and modular software architecture built around the Lightweight Communication and Marshaling (LCM) publish/subscribe message passing system for marine robotic applications. They used LCM to develop an unmanned port security vehicle from a commercial off-the-shelf autonomous underwater vehicle for visual feature-based navigation. The system is designed for easy adaptation to various marine robotic applications, enhancing its usability across different scenarios and ensuring low-latency, real-time performance, which is crucial for the effective operation of marine robots. However, the system has some weaknesses, such as complex implementation and limited environmental testing.

4.2.2. Land Transportation

Land transportation is the most widely used mode of transportation. Its popularity is due to its accessibility, flexibility, affordability, and the developed infrastructure of roads and highways. This peculiar characteristic makes ubiquitous ITSs challenging. Figure 11 is the distribution of vehicles in the United States [69]. It shows that 91% of vehicles are cars, light trucks, and motorcycles. Heavy-duty vehicles and boats make up less than 10%, and trains and planes account for under 1% of the total. Therefore, ITSs with real-time functionality are necessary for smart land transportation due to the diversity and size of this transportation category.

Bikes and Bicycles

Alexey Kashevnik et al. [70] developed an intelligent support system for cyclists that is context-based and makes use of information collected from various on-board bicycle sensors as well as a smartphone that is in front of the cyclist. It creates a real-time model of the physical space to determine the cyclist’s capabilities and the surrounding context of the e-bike. The system makes recommendations to the cyclist and controls the level of electrical assistance given based on the information provided. It uses a publish/subscribe mechanism for information sharing in the Smart Space, allowing for efficient communication and collaboration between different system components. Also, the system monitors the e-bike’s engine’s electrical assistance level, which assists in lower power usage and preserves battery life. The system combines a smartphone, sensors, and e-bike data for intelligent cyclist support, enhanced cyclist experience, and improved safety. However, some users may find that using this system is complicated to adapt, and the effectiveness of the system depends mainly on the capabilities and reliability of the smartphone.

Heavy and Light Vehicles

Another application of ITS in land transportation is event monitoring. Deshanta et al. [71] proposed an Event Data Recorder (EDR) for both heavy and light land vehicles. The system records visual and textual data from the car and uploads it to a server. The authors used the MQTT protocol to manage the EDR configurations, such as time intervals for data recording. The server is a broker and publisher, and the EDR subscribes to synchronization information. However, the architecture assumes reliable connectivity between vehicles and the central server, which may not hold in remote or low-coverage areas. Additionally, relying solely on MQTT without integrated security or fault-tolerance mechanisms poses risks to data integrity. Also, using a single centralized broker further introduces a critical single point of failure, which could disrupt the entire system in the event of a crash or attack.

Szabó et al. [72] proposed a participatory sensing system for a dynamic public transport feed using an Extensible Messaging and Presence Protocol (XMPP)-based publish/subscribe middleware. Its primary objective is to provide precise arrival times for public vehicles at specified stops. It combines two different kinds of data to achieve this accuracy. (1) Static data includes fixed schedules and routes of public transportation. It is the predetermined information about when vehicles are supposed to arrive at stops according to their regular schedules. (2) Real-time crowd-sensed data gathered from the passengers using their mobile devices. It provides real-time updates on vehicle arrival times, accounting for delays or schedule changes. By combining these two, the system ensures that it publishes reliable and up-to-date information to the customers subscribing to the feed and reflects the real-time situation of the public transportation system. Similarly, Farkas et al. [73] developed TrafficInfo, a participatory sensing-based live public transport information service using XMPP-based publish/subscribe middleware. It is a smartphone application that utilizes the power of the crowd to gather data, share information, and send feedback from many passengers to tell everyone where the buses and trains are and if there are any problems. Its strength is the real-time updates on the precise locations of public vehicles. The weakness of these systems is that they depend on user interaction to detect events, potentially inaccurate information, and reliability issues.

Autonomous Vehicles

Autonomous vehicles are changing the landscape of both public and private land transportation. However, they face challenges in handling complex sensor data fusion, managing interconnected functions in intelligent vehicles, ensuring scalability and interoperability, achieving deterministic high-speed communication, and coping with dynamic communication patterns, all of which require advanced technologies and adaptive software platforms for reliable and real-time data exchange. As a solution to these challenges, Arestova et al. [74] proposed a system that integrates Open Platform Communications Unified Architecture (OPC UA) [75], Time-Sensitive Networking (TSN) [76] with Automotive Open System Architecture (AUTOSAR) Adaptive [77,78] to provide deterministic high-speed communication. AUTOSAR Adaptive Platform is a service-oriented communication platform for standardized data exchange. OPC UA is a well-established communication standard [79] for modern industrial automation that supports publish/subscribe architecture, enhancing scalability. Publish/subscribe is designed to work with TSN, a collection of standards that adds real-time aspects to standard Ethernet [80] networks. TSN allows services with different requirements to share a single physical network.

Publish/subscribe middleware also allows communication between different types of autonomous vehicles. In [81], the authors explored new messaging mechanisms for two-way communication between autonomous vehicles, focusing on scenarios where a rover (using Real-time Operating System 2 (ROS2)) and a drone (using PX4) either work independently or coordinate tasks. The PX4-FastRTPS Bridge connects the PX4 and the ROS2, enabling them to exchange messages between their components through the publish/subscribe model. The authors found that the microRTPS bridge facilitates communication and control between the rover and drone, effectively enabling seamless interaction and bidirectional communication. However, the system’s reliance on protocol bridges introduces additional latency and potential points of failure, particularly under network congestion or limited bandwidth [82].

Trains

Rother et al. [83] investigated how safety-critical railway applications can achieve reliable and secure communication within on-premise SIL4 cloud environments. An on-premise SIL4 cloud refers to a private cloud environment that is located and managed by an organization, specifically designed to host safety-critical applications and meets the highest safety integrity level (SIL4) as defined by CENELEC (European Committee for Electrotechnical Standardization) functional safety standards [84,85]. The study emphasizes the importance of adhering to the EN 50159 safety communication standard [86]. It also evaluates the middleware Rail Safe Transport Application (RaSTA), DDS, and OPC UA protocols for compatibility with safety-critical railway requirements. The authors discovered a system architecture where DDS and OPC UA operate in a safe partition alongside safety components, separated from the non-safety-critical components partition. This design ensures safe communication over unreliable networks (black channels) by implementing a safety layer that handles errors such as delays, sequence faults, and unauthorized access. Their evaluation confirms that DDS and OPC UA Safety fulfill key safety requirements, support publish/subscribe models, and meet criteria such as authenticity, integrity, and timeliness, making them viable for scalable and secure railway communication systems. However, they found that RaSTA has limited security features. They also found that it is limited to point-to-point on-premise SIL4 cloud applications.

Hadiwardoyo et al. [12,87] developed a DDS-middleware-based safety system for underground railway environments. It uses DDS middleware (OpenSplice version 5.5) to ensure the sharing of crucial information throughout the network, even in the event of a disconnect in communication. The system intends to facilitate the exchange of data obtained from video files between train wagons and train stations to monitor passenger movements and traffic. As a result, passenger movements are constantly tracked and monitored. In [87], the authors leveraged QoS settings in DDS middleware to avoid regular communication interruptions in underground environments as a train moves from one station to another. The system uses the middleware for data exchange between trains and stations during stops despite being disconnected from the global networks. However, finding the optimal QoS settings is non-trivial.

4.2.3. Air Transportation

Air transportation is a complex system and a mode of transportation that moves passengers and goods by aircraft, such as airplanes [88]. It is considered one of the most crucial components of the global transportation system, which aids in traveling long distances for different countries in the shortest time possible and with a high safety level and convenience. It is worth mentioning that this type of transportation is essential for international commerce or trade, business, and tourism. Liu et al. [89] emphasized that air transportation struggles with fragmented data sharing, inconsistent interfaces, and delayed or inefficient data exchange across diverse stakeholders and heterogeneous systems. A publish/subscribe middleware can perfectly address these problems by providing a decoupled, event-driven framework that ensures real-time, secure, and scalable distribution of critical information among all relevant parties.

Chambers et al. [90] proposed an MQTT-based publish/subscribe middleware to manage multiple drones and services that operate with a high level of autonomy yet must collectively share congested airspace. The authors proposed Pop-up Drone Zones (PuDZs), i.e., temporary operational environments where small Uncrewed Aerial Systems (sUAS) can be rapidly deployed for tasks like emergency response. Traditional air traffic control methods are ill-suited for these transient, fast-changing scenarios since each sUAS and related service (e.g., weather updates, no-fly zones) has its local objectives and might not directly coordinate with other actors. Without a cohesive communication fabric, there is a real risk of conflicting airspace allocations or inconsistent responses to critical events such as erratic drone flights or sudden adverse weather. The authors developed the MQTT-based publish/subscribe architecture in which each system (e.g., airspace “air leaser”, digital shadow, drones) runs a local self-adaptive loop and shares events (faults, no-fly updates, weather changes) in real time. This method ensures loosely coupled but coordinated behavior. A Policy Manager enforces global rules, such as keeping only one active air traffic control service. Experiments show that the UAVs safely shared the airspace, avoided collisions, and responded consistently to unexpected events such as off-course flight, loss of ground control, or new no-fly zones. However, MQTT is not optimized for time-sensitive or safety-critical applications, and reliance on this lightweight protocol may limit system robustness in high-density or mission-critical drone operations [26].

4.3. Communication Technologies

Effective communication in ITS is crucial for coordinating network operations, ensuring data flow accuracy, and enhancing overall system efficiency and reliability. We found that selected papers use WSN/Vehicular Sensor Networks (VSNs), VANET, fog computing/vehicular fog computing (VFC), P2P communication, Information Centric Network (ICN), and mobile networks (e.g., 4G and 5G).

4.3.1. Peer-to-Peer (P2P) Communication

P2P has many applications in land transportation. Most protocols standardized under the European standard EN 50159 [91] are P2P protocols that involve two communication partners [83]. The standard specifies the requirements for safety-related communication in railway applications, focusing on the safety measures necessary for communication systems that are essential for the operation and safety of railway systems.

In road transportation, Ma et al. [15] proposed the ITS Information Platform (IIP), an event-based platform for ITS. The platform comprises two main components: a cloud component that provides canonical publish/subscribe services to traffic management facilities and mobile nodes and a client component that enables pub/sub functionality among mobile nodes. To address situations where cloud communication is infeasible due to latency constraints, the authors introduced a mobile P2P interface within the client component. This interface facilitates direct, infrastructure-independent communication between mobile nodes, such as vehicles, pedestrians, and cyclists, via short-range, high-bandwidth wireless channels. A motivating use case is at unsignalized intersections, where entities must coordinate right-of-way without traffic signals or centralized control. However, a key limitation noted in the paper is that the mobile P2P event broker operates as a best-effort mechanism, lacking guarantees for complete event-subscription matching in highly dynamic environments such as vehicular networks. This constraint poses challenges for safety-critical ITS applications where reliable and timely communication is essential.

Similarly, Garcia et al. [92] proposed an IP Multimedia Subsystem (IMS)-based platform for ITS that supports publish/subscribe communication through standardized IMS services. The system also has IMS Presence Service, which enables vehicles (as presentities) to publish real-time status information, such as location, speed, and operational state, while allowing authorized watchers (e.g., traffic control centers) to subscribe to this data. The architecture integrates supporting components like the Resource List Server (RLS) and XML Document Management Server (XDMS) to manage contact lists and filtering rules, enabling scalable and privacy-aware dissemination of presence information. This publish/subscribe infrastructure supports services such as fleet monitoring, access control to restricted areas, and emergency coordination, with the underlying IMS layer providing robust session control, authentication, and network abstraction across multiple communication technologies. However, the authors acknowledged a potential performance bottleneck: Application Servers (ASs) may become overwhelmed by the volume of notifications with heavy utilization of presence-based services. Although the authors proposed mitigations such as load balancing and multiple IMS domains, the solution’s scalability under large-scale real-world deployments remains an open challenge.

4.3.2. Mobile Telecommunication

In 2010, 4G was first launched by Telia Sonera in Finland, offering significant improvements over 2G and 3G through higher data transfer rates, better security, and advanced technologies like LTE-Advanced and Wi-Max [93]. It uses OFDMA for efficient spectrum use, MIMO for enhanced signal performance, and MU-MIMO to improve multiuser connectivity. However, LTE supports both FDD and TDD for seamless broadband wireless service. Also, 4G enables high-speed internet access, multimedia streaming, and mobile network interoperability, albeit introducing security risks such as IP vulnerabilities, VoIP spam, and potential network-wide failures due to shared core infrastructure [93]. It revolutionized ITS by enabling real-time vehicle communication, Smart Infrastructure, enhanced traffic management, and improved safety measures.

Dasanayaka et al. [94] indicated a 4G-based V2I road data communication mechanism that supports a publish/subscribe paradigm to multicast road safety messages according to the user’s location. It introduces an advanced filtering method based on the traveling road and the user’s heading direction to filter the most relevant road events. The authors used Global System for Mobile Communications (GSM) to transfer EDR sensor data to a server and subscribe to configuration settings from an MQTT broker version 3.1. The system backs up sensor data and allows users to access them when the EDR breaks during an accident. However, 5G is better for ITS. Experiments in [95] demonstrated that 5G outperforms 4G in speed, latency, capacity, and efficiency, making it ideal for real-time applications like autonomous driving, remote surgery, and IoT. Also, 5G achieves up to 20 Gbps, compared to 150 Mbps in 4G, and reduces latency from 30 to 50 ms to just 1ms, enabling Ultra-Reliable Low-Latency Communication (URLLC) applications. Additionally, it supports up to 1 million devices per square kilometer, making it far superior in high-density environments. Technologies like massive MIMO, beamforming, and millimeter-wave frequencies further improve spectral efficiency, network slicing, and reliability, whereas 4G struggles with congestion and scalability.

4.3.3. Information-Centric Network (ICN)

Information-Centric Networking (ICN) is an innovative technology that facilitates content retrieval through a name-based communications protocol, detaching content from original servers and creating new access points using in-network caches on intermediate routers or network nodes [96]. This method reduces congestion and enhances delivery performance by embedding multicasting, multipath communications, and security at the protocol level. Researchers have shown that ICN improves ITS through [97]: content-based routing, which addresses mobility challenges, a Dynamic Naming approach for efficient network sensing, and hierarchical naming for flexible and precise object tracking. Furthermore, ICN optimizes routing space utilization with a block centralized storage structure, improving overall sensing efficiency within the ITS network. However, the challenges of scaling the architecture, deploying it on today’s Internet, and securing it against new threats are non-trivial.

Losciale et al. [98] developed the BONVOYAGE Communication System, an innovative communication system developed within the European H2020 BONVOYAGE project, focusing on optimizing multimodal and door-to-door transport for passengers and goods. The system uses the Internames principle of the ICN communication paradigm, DATEX II standard [99,100], request–response, and publish/subscribe schema to disseminate static, real-time, and travel-centric content. The network consists of data producers (public or private transport operators publishing travel-centric information) and data consumers (end users who collect it). Experimental tests have shown that BONVOYAGE significantly reduces data processing for Data Consumers, resulting in a substantial decrease in bandwidth consumption and achieving an 80% reduction in latency and a 60% reduction in bandwidth consumption compared to baseline mechanisms. Although the proposed system is robust and comprehensive, its architectural complexity may pose configuration, integration, and long-term maintenance challenges. Moreover, as acknowledged by the authors, the system’s routing, security, and QoS mechanisms require further enhancement to ensure reliability and scalability in real-world deployments.

5. Comparative Analysis

Table 5. Comparative analysis of middleware solutions in ITS.

Refs.	Middleware	Use Case	Real-Time	Scalability	Limitation
[48,49]	Apache Kafka	IoT traffic surveillance	Yes	High	High processing demands; dependent on continuous camera data
[56,57]	COPSS *	VANET black ice warnings	Yes	High	High RSU infrastructure cost
[59,60]	CaaS *	V2V cooperation	Yes	Moderate	High complexity; requires dense vehicular presence
[58]	ITL System *	VANET traffic light alerts	Yes	High	Limited to infrastructure-based intersections
[65]	Spatio-Textual Pub/Sub *	RSN geo-textual filtering	Partial	Moderate	High computational cost for filtering at scale
[68]	LCM Pub/Sub *	Marine robotics	Yes	Moderate	Complex implementation; limited testing scope
[71]	MQTT	Land vehicles event monitoring	No	Low–Moderate	Requires reliable connectivity; single point of failure
[90]	MQTT	UAV swarm coordination	Partial	Moderate	MQTT unsuitable for mission-critical timing
[72,73]	XMPP-based Pub/Sub *	Public transit crowdsensing	Yes	Moderate	Dependent on user input; potential data inaccuracy
[74]	OPC UA + DDS + TSN	AVs in-vehicle communication	Yes	High	Complex integration; QoS tuning is difficult
[81]	ROS + microRTPS Bridge	Drone–rover coordination	Yes	Moderate	Added latency and fragility under congestion
[12,87]	OpenSplice DDS	Train–station communication	Yes	High	Non-trivial QoS configuration
[15]	IIP Platform *	Road ITS client-cloud-P2P	Yes	Moderate	Mobile P2P matching is best-effort only
[92]	IMS Presence Service *	Road ITS fleet/status monitoring	Yes	Moderate	Server bottlenecks under heavy subscriptions
[98]	BONVOYAGE System *	Multimodal ITS using ICN	Yes	High	Complex architecture; QoS and security need refinement

* Indicates custom-developed middleware tailored to a specific ITS use case.

compares the middleware solutions reviewed in Section 4 by use case, real-time capability, scalability, and key limitations. The middleware with “*” is built based on a standard. It includes standardized platforms (e.g., Apache Kafka, DDS, MQTT) and custom-developed solutions (e.g., COPSS, CaaS, ITL), offering a functional overview of middleware performance across various ITS domains. The table reveals trade-offs in middleware design, such as between real-time responsiveness and scalability or between implementation complexity and adaptability to constrained environments. The table shows that custom middleware dominates in specialized contexts like VANETs, where domain-specific communication models are necessary to address safety and coordination challenges.

However, standard middleware platforms like Apache Kafka and DDS demonstrate high scalability and find applications in IoT and railway systems, where throughput and reliability are a priority. However, these often come with increased computational or integration overheads. Lightweight protocols such as MQTT and XMPP are favored in low-bandwidth or user-contributed environments such as EDR systems and public transit apps, even though they may lack the robustness or security features needed for safety-critical ITS applications.

Finally, the table highlights real-time capability as a differentiator: while most middleware reviewed supports some form of real-time data exchange, only a few, such as DDS, COPSS, and OPC UA with TSN, offer deterministic or bounded-latency performance suitable for safety-critical operations like autonomous driving and train control. Conversely, scalability is limited in middleware solutions that depend on central brokers (e.g., MQTT in the EDR system) or rely heavily on human input (e.g., XMPP-based crowd-sensing).

6. Challenges

The selected papers show that publish/subscribe middleware can tackle a variety of ITS problems: they can improve connectivity, enable infrastructure-less communication using P2P technologies, reduce latency and increase reliability using QoS settings, and support real-time systems using real-time publish/subscribe middleware. However, many challenges affect the deployment of publish/subscribe-middleware-based ITS. This section discusses some of them publish/subscribe-middleware-based ITS face.

6.1. Privacy Challenges

According to the UN [101], privacy is a fundamental human right recognized in numerous international treaties, including the right to avoid observation and to be left alone and the ability to keep one’s thoughts, beliefs, identity, and actions confidential. Privacy also entails the right to decide and manage why, how, where, what, to whom, and when to disclose personal information. This perspective on privacy closely connects it to information (or data) protection.

Privacy is also one of the potential challenges of publish/subscribe middleware in ITS. Garcia et al. [92] illustrated an example of this, where presence-based services (PSs) are services that act as “watchers” of a resource list. These services monitor the status of entities, known as Presentities, which are vehicle terminals that send status information. Vehicle terminals connect to only one PS, receiving all status information in the same notification. This situation may lead to watchers receiving unnecessary data about other services.

The deployment and maintenance of sensors for ITS is expensive. Thus, researchers proposed using crowdsensing to collect data, actual vehicle positions, deviation from the static timetable, crowdedness information, vehicle condition, and road conditions [72,73,102]. Crowdsensing helps provide real-time data and connectivity in locations without network or ITS infrastructures. Commuters either input data manually or their smartphone sends its sensors’ data to the ITS, where the service provider collects, aggregates, cleans, and analyzes the data to infer the public vehicles’ status [73,102]. However, participants send sensitive data like their location and preferred route, which poses a significant privacy challenge. Researchers must provide techniques to protect participants’ privacy while ensuring the real-time performance of the system is not affected.

6.2. Data Challenges

Transportation systems produce massive quantities of data through different sources, such as sensors, GPS devices, and connected vehicles. Effective use of this data is vital for enhancing safety, efficiency, and overall system performance. Nevertheless, one of the main challenges is dealing with real-time data. Shariat et al. [103] pointed out that it is crucial to minimize the time gap between the occurrence of an event (such as a traffic issue) and its detection and analysis to improve the performance of ITSs. However, delays in this process can negatively impact the system’s effectiveness. Traditional sensors like CCTV cameras and loop detectors have a fixed range of coverage that limits their ability to capture real-time data comprehensively. In addition, existing traffic management systems often do not share data outside their agency’s domain, leading to inaccuracy in the real-time data analysis and identifying the underlying causes of traffic issues.

Furthermore, Liu et al. [104] mentioned that multihop connections are necessary for V2V communication in VFC. However, an excessive hop count on a network might result in a low message delivery ratio, high latency, and data redundancy. The authors in [105] argued that in situations with many subscribers (i.e., receivers like cars, motorcycles, and bikes) needing real-time information. Then, each subscriber can take part in delivering the content to make the distribution process more efficient and reduce redundancy. This process does not put much strain on the subscribers because the real-time information is usually transmitted in a single packet, incentivizing them to forward the content to others. This collaborative approach helps to improve the overall efficiency of real-time content distribution, especially in scenarios with a high density of interested recipients.

Fault tolerance is also crucial for some critical situations in transportation. Bender et al. [106] demonstrated a publish/subscribe model that enables a group of loosely connected objects to exchange data within a single process. However, in a complex system, relying on a single process to handle all the functionalities is not ideal, as it creates a single point of failure that can be hard to identify and fix.

Data structure and format heterogeneity is an issue in Smart Cities. Raghavan et al. [107] pointed out that the data collected from sensors in Smart Cities have diverse formats, and the systems currently in use have varying structures. This diversity makes Smart City system integration or expansion with third-party components or services challenging.

6.3. Connectivity Challenges

Bandwidth limitations pose a critical challenge to ITS. Kuhn et al. [108] noted that in ITS scenarios, mobile peers (vehicles) have only a few seconds of connectivity and a limited bandwidth. Reconnecting peers face challenges receiving stored events due to limited bandwidth and a small connectivity window. Therefore, this may create backpressure in the system and pose a risk that essential safety-critical information may not reach the peer in time. This risk increases by delivering irrelevant or incorrect information instead of life-saving information. Moreover, in some ITSs, it is necessary to ensure an exactly-once delivery of subscribed messages, as it is a requirement for some reasons, such as jurisdictional reasons [108]. Durable notification services must store all events a peer has subscribed for while the subscriber is offline. However, meeting this condition may be challenging for publishers with limited resources and lacking broker infrastructures to store the messages.

Bouloukakis et al. [109] mentioned the lack of portability across systems. The current platforms are designed for specific agencies and require the agency’s application to access its services. Developing applications that can work across multiple transportation systems is non-trivial. For example, political, legal, and technical challenges impede the development of a single bus application that commuters can use across cities, let alone different countries.

Efficient data delivery for mobile nodes is also a significant challenge. Ventrella et al. [110] demonstrated that maintaining a stable and efficient communication path for mobile vehicles connected to a core network through wireless communication technology is challenging. Each network attachment point (e.g., RSU) serves as a node in the network. The vehicle’s dynamic movement results in frequent changes in network attachment points, requiring adaptation of the communication path.

6.4. Security Challenges

Security is the most challenging part of computer network systems. Research must tackle two security challenges in publish/subscribe-middleware-based ITS [103]: (1) The system must ensure that it accepts data only from verified publishers, preventing unauthorized data injection; (2) It must also restrict access to the published data to authorized individuals only.

Researchers can leverage the broker in a centralized publish/subscribe middleware to accomplish these requirements. In the Connected Vehicles and Smart Transportation (CVST) platform [103], the broker functions as a reliable key management system by issuing certificates to publishers and subscribers, thereby validating the data’s provenance and integrity. The system uses these certificates to sign interest and data packets in Named Data Networking (NDN), and a shared key encryption algorithm facilitates secure data exchange between publishers, the broker, and subscribers. The broker provides shared keys to authorized users for encrypting and decrypting data, ensuring confidentiality and integrity within the system. Similarly, Deshanta et al. [71] used Secure Copy Protocol (SCP) to secure EDR data they upload to a central server that doubles as an MQTT broker. However, a broker is a single point of failure, which puts it at risk of cybersecurity and cyber-physical attacks. The pay-off to the attackers is even higher if the broker has multiple functions, like in the case of [71]. Also, the security of publicly available brokers or physically located in remote areas of the ITS presents tremendous cyber-physical challenges.

For distributed publish/subscribe middleware, aspect-oriented space containers (ASCs) are applicable data in publisher/subscribe ITS protection. A space container is a customizable storage component designed for efficient storage and retrieval of structured, spatial-temporal data distributed in a fault-tolerant manner through Distributed Hash Tables (DHT) [108]. Developers can use aspects in ASC to realize security measures such as authorization and authentication, ensuring secure communication within the system. The integration of aspect-oriented space containers in publish/subscribe scenarios can enhance security by allowing the implementation of security mechanisms through aspects. However, adding an ASC may lead to memory and latency overheads on the ITS.

Gianuzzi & Merlo discussed some cybersecurity concerns of beaconing in publish/subscribe middleware for VANET [14]. Beaconing is the periodic sending of beacon signals containing a node’s ID and movement details to advertise the node’s presence to nearby neighbors. It helps manage network congestion, support safety, and emerging applications in vehicular networks. The authors argued that beacons in VANET applications raise security concerns regarding the accuracy and reliability of the information exchanged, especially in safety applications like collision warnings and traffic management. Thus, a publish/subscribe middleware should provide intrusion detection systems that evaluate the variation of beacon parameters between consecutive beacons to prevent malicious activities like Sybil attacks [111] or malfunctioning nodes sending incorrect beacons [14]. This threat is particularly significant in large-scale P2P networks due to its application in many railway ITS protocols [83] and other P2P-based land transportation ITS [15,92].

Other security concerns of using publish/subscribe middleware include vulnerabilities in rules and protocols that can lead to network misconfigurations, which account for a significant portion of network failures [112]. Thus, model-checking techniques are essential to identify vulnerabilities in design rules and protocols of communication systems, addressing security issues before any losses occur. Researchers should develop easy-to-configure publish/subscribe ITS, reducing system installation and maintenance costs. Also, stakeholders must focus on staff training to ensure system security.

7. Conclusions

ITS applies advanced information and communication technologies to transportation networks to improve safety, efficiency, and sustainability. The ARC-IT reference architecture provides a structured framework for ITS design, comprising the Enterprise, Functional, Physical, and Communications Views. The Communications View defines the protocols and standards that enable data exchange between system components.

This review focused on the Communications View of ARC-IT v9.3, investigating the role of publish/subscribe middleware in improving ITS performance. Given the limited number of studies addressing this intersection explicitly, our goal was to consolidate knowledge on middleware-based communication models, identify application domains across computer network infrastructure, transportation modes, and communication technologies, and highlight key challenges. The findings aim to guide researchers and practitioners toward building scalable, interoperable, and real-time communication solutions for next-generation ITS.

This study found that publish/subscribe middleware provides clear benefits in several ITS application areas, especially those requiring real-time communication and dynamic scalability. For example, COPSS demonstrated high value in VANET scenarios involving safety-critical alerts, outperforming traditional broadcast and polling methods in efficiency and reliability. Apache Kafka is highly scalable and effective for high-throughput event streaming in IoT-based traffic surveillance, though it requires significant computational resources. DDS middleware has shown strong real-time and reliability capabilities in structured environments like railway systems and autonomous vehicle communication, but its configuration can be complex. Conversely, protocols such as MQTT are easy to deploy but are less suitable for mission-critical or high-density scenarios. XMPP-based participatory sensing and IMS Presence Services offer practical public transportation and fleet monitoring solutions but face scalability and data accuracy challenges. These findings indicate that the choice of middleware for ITS applications is influenced by system requirements, the availability of infrastructure, and performance limitations. Therefore, there is a strong need for hybrid approaches that combine the adaptability of custom middleware with the robustness of standardized platforms to ensure cost-effective, secure, and scalable ITS deployments.

Additional challenges in integrating publish/subscribe middleware in ITS include privacy risks from crowdsensing and data sharing, real-time data processing constraints, fault tolerance, and system heterogeneity, connectivity issues due to bandwidth limits and mobility, and security vulnerabilities such as unauthorized data access, broker attacks, and protocol misconfigurations. Overcoming these barriers is crucial for developing secure, scalable, and interoperable ITS solutions.