After several years on standardization efforts, the Wireless Access in Vehicular Environments (WAVE) has been accepted as the system architecture for vehicular communication, and several standards have been released for short-range communication in VANETs. These standards are described in Table 8.1 [19–25]. IEEE 802.11p (IEEE Std 802.11p-2010) [18] is the physical layer standard including a set of extensions to the IEEE 802.11 standard. Upper layers include the family of IEEE 1609 standards, which relies on IEEE 802.11p (Fig. 8.2). The family of 1609 standards defines the architecture, communications model, protocols, security mechanisms, network services, multichannel operation and the use of Provider Service Identifiers in the vehicular environment. It is aimed to support high-speed (up to 27 Mb/s) short-range (up to 1000 m) low-latency wireless communications [26].

VANET applications are classified in a variety of ways in the literature. Willke et al. [31] define four types of applications according to the aim of the application: (1) general information services, (2) vehicle safety information services, (3) individual motion control, and (4) group motion control. Karagiannis et al. [32] categorize applications in three categories: (1) active road safety applications, (2) traffic efficiency and management applications, and (3) infotainment applications. One classification has been defined in Ref. [33] as shown in Fig. 8.4. In Ref. [33], applications in VANETs are generally classified as safety or non-safety applications. Safety applications are subcategorized as situation awareness applications and safety messaging applications. Non-safety applications include the applications for comfort driving, enhancing the driving process and traffic information systems, which do not present any safety or life-critical requirements.

The major goals of enabling ITS based smart cities are to increase traffic safety and to increase energy efficiency. Bifulco et al. [45] have addressed the 2020 ambition of Europe with efficient and sustainable energy utilization. According to the study of Kley et al. [46] one solution to achieve the goal will be introducing electric vehicles with modern ITS. In addition, Bifulco et al. mentioned the concept of smart driving which has been explored in various research projects including projects from Microsoft in San Francisco, Accenture in Amsterdam and IBM in Singapore to build smart cities. For smart transportation, Microsoft launched smart parking in 2004 in Bay Area of San Francisco. With combination of geo-referencing and Windows Azure; Microsoft research utilizes real-time data on transportation system and traffic flow [45,47].

Traffic law violations have a major impact on road safety of smart cities. Specially, speed, tailgate and fitness issues are primarily being monitored to ensure road safety standards. Moreover, unregistered and law violating vehicles can create security threats.

There are several clustering algorithms proposed to address the challenges of VANET clouds. Main challenge in clustering algorithms is forming more stable clusters in terms of cluster size, member node exchange, cluster head election and long duration of cluster heads. One good clustering algorithm for VANETs should also form fewer clusters to ease maintenance and stability. A Fuzzy Clustering-based Vehicular Cloud Architecture (FCVCA) has been proposed in Ref. [57] with a new clustering technique to group vehicles. In this paper, authors propose a new 3-step approach for estimation of traffic volume in a particular road segment. Initially, traffic information has been collected from different clusters with the help of the proposed clustering algorithm. With a virtual chain among clusters this information is transmitted toward the road-side cloud. The virtual chain is used to meet the connectivity demand between clusters with RSU as RSUs have limited transmission range. Later, the total traffic volume has been calculated with a generalization method from the collected data. In the simulation, the proposal has been tested with performance metrics of inter-vehicle distance, density and flow rate. Flow rate is the amount of nodes crossing particular road segment within specific time. The metrics used in the simulation environment are duration of cluster head and amount of clusters. Within the similar environmental scope, the proposed algorithm is compared with Lowest-ID [58] and MCMF [59] techniques. According to comparison results, it has been reported that the proposed approach can construct a higher number of stable clusters. Authors also measured the quality of volume estimation for the evaluation of the traffic volume estimation accuracy in their approach. While the scheme is compared with Online Learning Weighted Support-Vector Regression (OLWSVR), the proposed method performs better in terms of low, mid or high flow rates. Thus this proposed scheme has been illustrated with reduced amount of cluster formation in comparison with Lowest-ID and MCMF approaches.

VANETs have an important role for message propagation during emergency situations and disasters such as cyclones, earthquakes, fires, volcanos, etc to reduce loss of life and resources. It has been noted that message dissemination during a crisis is one of the key challenges for future research in this area. Alazawi et al. [63] have proposed a system using VANETs, cloud computing, and mobile technologies which is based on the transportation in a real setting. The system is able to retrieve real-time data and utilize the connectivity in between mobile and social networks with VANETs. In emergency situation, an efficient system is expected to control road traffic accordingly and able to propagate the information with available resources. Through data analysis the system can design a strategy to optimize the impact of disaster, for example, public transportation for evacuation from the point of incident. Authors have evaluated the proposed solution during disaster scenario and found the proposed system better than conventional approach for evacuation. Their future work aims to work on real-time cases and enlighten the scopes of intelligent disaster management system.

Security challenges in VANETs and cloud computing are inherited by vehicular clouds as mentioned in Ref. [43]. These challenges have been studied in detail in Refs. [68–70]. These studies can be improved by taking the specific conditions and requirements of vehicular clouds into consideration. In Ref. [66], the authors summarize these challenges as spoofed identities, nonrepudiation, Denial of Service (DoS) attacks, and mobile authentication. Yan et al. have proposed a security framework that addresses most of these challenges [68]. The main targets for an adversary are reported as confidentiality of VMs in the vehicular cloud, integrity of the content that is stored in a distributed manner, and the availability of physical machines, resources, services and applications in the vehicular cloud. Based on these three targets, a typical attack scenario has been defined as follows: (1) the geographic location of the victim vehicle is identified and possible physical hosts (ie, vehicular nodes) in the vicinity are discovered where the victim vehicle is possibly being served, (2) the vehicular node that serves the victim vehicle is discovered upon submitting several service requests to the cloud, (3) Once the VM that is allocated to the victim is identified, services are requested on the same host, (4) Finally higher privilege is aimed to be obtained to collect assets via system leakage. This type of attack is inspired from the attacks that explore information leakage in the clouds [71]. While authenticating the nodes with high mobility, it is not viable to use well-known metrics such as ownership, knowledge and biometrics. Furthermore, Sybil-like attacks are always possible in such an information network [72].

Privacy is a major concern for cloud users due to virtualization-based vulnerabilities where sensitive information can be revealed to adversaries [75,76]. The authors in Refs. [77,78] present the benefits of mobile cloud computing in a vehicular cloud environment while presenting the privacy and security challenges of a vehicular cloud environment in detail. Existing privacy considerations for cloud systems and VANETs [69,70,73] can be adopted by vehicular clouds however special requirements of vehicular clouds have to be taken into consideration. Anastasopoulou et al. [79] use game theory-based solutions to improve privacy of cloud-based mobile apps. The proposed methodology analyzes the user interactions, and makes a compromise between the QoS in the cloud and user privacy.

As mentioned before, a vehicular cloud can be considered as a data center with unstable physical hosts. Therefore, virtual machine management appears to be a challenging issue in vehicular clouds. VM migration may occur when a vehicle is about to go off the grid and cannot be in range of any RSUs, or when a handover occurs between two RSUs that are in range of a vehicle. To cope with this challenge, Refaat et al. have proposed a VM migration scheme in Ref. [37]. The VM migration algorithm works as follows: Out of the nearest vehicles, the source vehicular node selects a destination vehicular node based on the search criteria. If the destination vehicular node does not have sufficient available capacity to host the VM or if the VM cannot be migrated to the destination node in a predefined time window, migration is re-attempted by excluding the corresponding destination. Otherwise, the VM is migrated to the destination vehicular node. A migration attempt is marked as unsuccessful if a certain number of migration attempts fail. An unsuccessful migration requires intervention of the RSU. Thus, the VM is directed to the RSU if the migration cannot be completed.

Context-awareness in a vehicular cloud is emergent for smart city applications for various reasons. In Ref. [81], the authors propose a behavior pattern recognition methodology to detect anomalies in driver behaviors, and inform the other drivers in the vicinity for their safety. On the other hand, Santa and Gmez-Skarmeta propose a context-aware information-provisioning scheme for V2V and V2I communications for road safety [82]. The vehicles are identified by the RFID technology, and by keeping track of vehicles, current traffic and road condition information is obtained as provided to all drivers in the vehicular network. Wan et al. have extended these ideas to propose a cloud-assisted context-aware architecture [65]. A multilayer architecture is proposed for context-aware vehicular cloud implementation. The three layers are summarized subsequently: