Takamasa Higuchi

Geo-spatial resource allocation for heterogeneous vehicular communications: poster

Vehicular networks need to make optimal use of their limited radio resources to achieve sufficient performance and reliability. However, the scale and dynamicity make it very challenging to optimize this use and to meet the strict requirements. In this paper, we propose a centralized scheme that allocates resources to geographic locations to address this problem. The scheme can improve the overall network performance by leveraging wide-scale information instead of the restricted local views used in distributed approaches.

Parked Cars as Virtual Network Infrastructure: Enabling Stable V2I Access for Long-Lasting Data Flows

Vehicular networks are envisioned to cover various use cases from safety related applications to infotainment. While there exist standardized solutions for exchanging single messages in a geographical area, coping with longer lasting flows of messages is more difficult. Developed solutions may require high node densities for fast delivery or even require pre-installed infrastructure (e.g., road side units). In this paper, we present a concept where we exploit parked cars to form a virtual network infrastructure. In particular, we form clusters representing a virtual road side unit spanning a relatively large geographical area. To reduce the channel load, we select a subset of the parked cars as active gateways to the cluster. This is done in a way to ensure that the connectivity to the cluster is not impaired. We evaluate the proposed algorithm using the Car4ICT architecture, a concept that enables cars in smart cities to discover and use various kinds of services.

Vehicular Micro Clouds as Virtual Edge Servers for Efficient Data Collection

Automakers have already started rolling out cars with communication capabilities. They become more than simple means of transportation - their mobility combined with networking lets them interconnect people and machines. Looking at the network part, the most important aspect is to make the underlying network scaleable. Mobile edge computing has been introduced in 4G networks to achieve such scalability using caches and processing capabilities at eNodeBs, i.e., at the edge of the network. We introduce the concept of vehicular micro clouds as virtual edge servers. Micro clouds are conceptually clusters of cars, which help aggregating collected data that is transferred to some backend. In contrast to previous work on dynamic clustering in vehicular networks, we investigate the construction of such micro clouds using a map based approach.

Poster: Using clusters of parked cars as virtual vehicular network infrastructure

In future smart cities, cars will be equipped with multiple communication technologies. The Car4ICT architecture aims to exploit such cars for providing services to users. Cars take the role of service hubs and support users in discovering services and utilizing them. So far, Car4ICT has been investigated in urban and rural scenarios, but parked cars have not been considered part of it. As such parked cars are ubiquitous in cities, they help improve the architecture even further by reducing the need to rely on Store-Carry-Forward (SCF) and adding more services. In this paper, we outline our integration of parked cars into the Car4ICT architecture. By combining them into clusters, we are able to add additional network nodes, thus improving the stability of the network topology. Furthermore, members of these clusters are then connected to the Car4ICT network and are able to provide/consume extra services. While there exist solutions for clustering, there are several research questions when integrating such virtual vehicular network infrastructure. We describe our approach and discuss some of the interesting research questions and the problems that have to be solved.

Vehicular micro cloud in action: On gateway selection and gateway handovers

Abstract Recently, the concept of the vehicular micro cloud has been established. The core idea is to use cars as a main ICT resource in modern smart cities being able to store, to forward, and to process data. Relying on the communication capabilities of cars, particularly their ability to use short range communication technologies such as IEEE 802.11p, Wi-Fi, or LTE-D2D, virtual micro clouds can be established similar to the concept of the mobile edge in 5G networks, yet, without any infrastructure support. However, there is a key disadvantage: the network will very likely become fragmented due to the mobility of the cars within the city. Furthermore, low penetration rates in early deployments may amplify this fragmentation. In recent work, we proposed the use of clusters of parked cars to overcome such limitations, i.e., to provide a virtual infrastructure in form of a virtual Roadside Unit (RSU) being able to fulfill all the mentioned actions. In this paper, we investigate two core problems of this system, namely the selection of appropriate gateway nodes in the virtual cluster as well as seamless handovers among such gateways required due to very limited contact times of moving cars to a single gateway. Our simulation results confirm the strength of the proposed gateway selection and handover mechanisms.

Emerging Trends in Vehicular Communication Networks

The potential of connected and autonomous vehicles can be greatly magnified by the synergistic exploitation of a variety of upcoming communication technologies that may be embedded in next-generation vehicles, and by the adoption of context-aware approaches at both the communication and the application levels. In this chapter, we discuss the emerging trends, potential issues, and most promising research directions in the area of intelligent vehicular communication networks, with special attention to the use of different types of data for multi-objective optimizations, including extremely large capacity and reliable information dissemination among automotive nodes.

Cars as the base for service discovery and provision in highly dynamic networks.

In the smart cities of tomorrow huge amounts of data have to be processed. Cellular networks alone are unlikely to be able to cope with the amount of data and would introduce a critical dependence. Cars, on the other hand, are ubiquitous - even after massive disasters - and will be equipped with abundant communication technologies. This makes them a salient basis for connecting users and machines of smart cities. Discovery and providing services in such a highly dynamic network is a challenging task. We tackle this problem in our Car4ICT architecture which could be shown to work very well in simulations. The next logical step is to investigate the performance in experiments. To perform experiments we built a prototype which is able to emulate multiple vehicles on just a few machines. We show the feasibility of the proposed service discovery protocol and our emulation approach.

Regional InfoHubs by vehicles: balancing spatio-temporal coverage and network load

The growing ubiquity of on-board wireless communication devices in vehicles is steadily opening up new possibilities of vehicular networks. In this paper, we design a framework to utilize radio-equipped cars as regional information hubs, which periodically disseminate public regional information (e.g., advertisements, travel tips, emergency info, etc.) over a designated geographical area in a city. The cars driving in the designated area repeatedly broadcast the data messages, allowing other drivers and pedestrians to passively obtain the information with their on-board wireless devices and/or smartphones. Although the idea follows a natural extension of the current capability of vehicular networks, vehicle densities in urban areas are typically highly non-uniform both in time and space, making it difficult to robustly disseminate data messages with minimal network load. We tackle this issue by introducing a probabilistic framework to intelligently adapt message dissemination strategy. Analyzing location broadcast messages, which are periodically exchanged among vehicles for safety applications, our protocol predicts future locations of neighboring radio-equipped vehicles and dynamically adjusts probability of message transmissions at each moment to suppress redundant network traffic. Simulation results show that our protocol can reduce the total amount of redundant message transmissions by up to 78%, while maintaining the spatio-temporal coverage of message dissemination.