Elsevier

Vehicular Communications

Volume 5, July 2016, Pages 18-34
Vehicular Communications

TraC: A Trajectory-aware Content distribution strategy for vehicular networks

https://doi.org/10.1016/j.vehcom.2016.09.005Get rights and content

Abstract

In vehicular networking, contacts have short duration and may seldom occur. Hence, maximizing the amount of data transferred per contact opportunity is of utmost importance. In this work, we propose TraC which combines network nodes caching, typically used in Content-Centric Networks (CCNs), and users' trajectory knowledge. The idea is to improve data delivery by proactively caching the content requested by users over the access points along their trajectories to destination. To accomplish this, we use the network formed by access points to forward individual interests containing information regarding users' destination. In addition, we develop two forwarding strategies consistent with the content oriented paradigm, and a neighborhood discovery protocol, required for the operation of the proposed forwarding strategies. The results obtained through simulations in highway, urban, and rush-hour scenarios show that TraC can increase the fraction of interests satisfied as well as accelerates content delivery compared with a typical CCN implementations for wireless environments.

Introduction

Today, permanent and ubiquitous connectivity is a fundamental requirement for most users. Mobile communications can typically fulfill such needs at the cost of a monthly subscription fee for telecommunication services. Nevertheless, depending on the number of users and the amount of data transferred, mobile operators struggle to avoid network congestion. In this case, a well-known strategy to alleviate the mobile network is the utilization of traffic offloading. Indeed, statistics from 2014 confirm this trend, showing that 46% of the global mobile data traffic was offloaded, and that 54% is expected to be offloaded in 2019 [1].

A preferable technology for traffic offloading is IEEE 802.11, given the network availability and low cost. IEEE 802.11 networks, however, must deal with the limited range of access points, which leads the network to islands of connectivity [2]. In mobile scenarios, e.g., vehicular networks, these coverage gaps represent an obstacle for nodes moving at high speeds since contacts become shorter and less often. Even though mission-critical and security applications are not suitable for such conditions, infotainment applications can still have a high profit from contact opportunities. The issue then becomes how could it be possible to maximize the amount of data transferred via multiple access points?

Taking into account the current TCP/IP model, improving data transfers per contact opportunity opens venue for new Internet architectures since the original one was not designed for mobility. Content-Centric Networks (CCN) [3], [4] are an alternative architecture for the Internet which can also be used for content retrieval in vehicular networks. Users request a given content and the CCN takes care of finding and sending it back to the requesting node [5], [3]. The objective of CCN is to abstract the notion of IP addressing, since users are more and more interested on content, no matter where it comes from [5], [6], [7]. To avoid network flooding per content request, one key feature of the CCN architecture is content persistence in all network nodes and not only at the network edges. Such persistence can leverage data offloading in vehicular networks as a possibility to increase the number of content sources [8]. For instance, combining vehicular and content centric networks, we can think of storing content at each access point along a vehicle trajectory. This approach can potentially improve the efficiency of content retrieval at every contact opportunity between a vehicle and an AP, further maximizing users' interest satisfaction.

The main idea of TraC is to use CCN persistence to build proactive caches in all vehicular nodes based on users' trajectory. This can, at the same time, improve content delivery and circumvent vehicular mobility issues as CCN does not rely on host-oriented approaches. Previous CCN-based strategies neither rely on content caching in APs nor on users' trajectory information. Hence, the whole strategy has not been explored yet in the literature to the best of our knowledge. We rewrite the fourth paragraph of the Introduction to emphasize our paper contribution.

In this work, we propose TraC, a Trajectory-aware Content distribution strategy, which couples vehicular networking to the content-centric paradigm. Our main idea is to use CCN persistence to build proactive caches in all vehicular nodes based on users' trajectory. This can, at the same time, improve content delivery and circumvent vehicular mobility issues as CCN is not host oriented [2]. Previous CCN-based strategies, to the best of our knowledge, neither rely on content caching in APs nor on users' trajectory information. In TraC, we assume previous knowledge of users' geographical destination to forward interests toward APs that will probably be crossed along their trajectories to destination. These APs can, as a consequence, proactively download the content requested from the Internet even before the vehicle arrives. As a result, when the user associates to an AP along her trajectory, the content requested will be already available. The time needed then to request and transfer the content from the Internet to the connected AP is saved and the content can be immediately retrieved. To permit such proactive caching, we propose two strategies for vehicular interest forwarding between APs, Triangular Area Forwarding (TAF) and Distance Minimization Forwarding (DMF), and a neighborhood discovery protocol. Using TAF and DMF, received interests are only forwarded to APs along users' trajectories. To this end, APs need to be aware of users and neighbor APs position, which are obtained, respectively, with modifications to interest packets and with the proposed neighborhood discovery protocol. The performance of TraC is evaluated via simulations in three vehicular scenarios: urban, highway, and a realistic rush-hour scenario using the Cologne dataset [9]. In all experiments, we compute the fraction of users' interests satisfied, the content delivery ratio, and the network responsiveness in terms of how fast users' interests are satisfied. Compared with a typical implementation of CCN for wireless networks, results show that TraC satisfies more interests more quickly, reaching gains up to 50% in the fraction of interests satisfied.

This work is organized as follows. Section 2 introduces the CCN architecture and its utilization in the vehicular scenario. Section 3 proposes the Trajectory-aware Content Distribution (TraC) strategy. Section 4 describes TraC operation, providing more details concerning the proposed trajectory-aware strategies for content request and the Neighborhood Discovery protocol. In Section 5, we describe the simulation scenarios, parameters, and configurations. The obtained results are shown in Section 6. Section 7 lists related work. Finally, Section 8 concludes this work and discusses future directions.

Section snippets

Content-Centric Networks (CCN)

In this section, we overview the traditional CCN architecture and, in the following, we draw arguments for CCN deployment in vehicular networking. At the end, we briefly compare CCNs to Delay-Tolerant and Disruption Networks (DTNs).

The trajectory-aware content distribution (TraC) strategy

In this work, we propose the use of proactive caching in access points (APs) to improve content distribution in vehicular networks. The key idea consists in proactively download the content requested by a user to the APs along her trajectory toward destination. The goal is to make the content available in the cache of each AP even before the requesting user arrives. This strategy contrasts to the host-oriented approach since users do not have to wait the content to be downloaded every time she

TraC operation

This section shows in more details how users inside vehicles request and receive content, and how APs operate to deliver Internet content to users.

Simulation settings

This section describes the different scenarios and traffic patterns used in our simulations (Section 5.1). In addition, we also introduce the network parameters and the data structures used (Section 5.2).

Results

In our results, DMF refers to TraC using the distance minimization forwarding, TAF refers to TraC with the triangular area forwarding, and CCN is the adapted version of CCN to the vehicular scenario. Note that in the adapted CCN version, APs neither perform proactive caching nor forward V-INTs to neighbors. Also, in the adapted CCN, APs use a single FIFO structure to handle all incoming content requests. Whenever applicable, results show a vertical error bar representing a confidence interval

Related work

Works that employ CCN in vehicular scenarios usually take care of the medium access to avoid multiple transmissions of the same interest in a short period of time [6], [10]. Therefore, nodes schedule interest transmissions and listen to the medium during a random amount of time. If a node overhears another node transmitting the same scheduled interest, it simply cancels the transmission. The same holds for content chunks, which have also to avoid duplicated retransmissions.

Using the CCN

Conclusions and future work

This paper proposed TraC, a Trajectory-aware Content Distribution strategy, which uses proactive caching in Access Points (APs) to increase the probability of content delivery in vehicular scenarios. We have proposed two geographical strategies for vehicular interest forwarding between APs, and an additional neighborhood discovery protocol for content-oriented networks. The forwarding strategies were evaluated in three vehicular scenarios: highway, urban, and a rush-hour segment of the Cologne

Acknowledgements

The authors would like to thank FAPERJ, CNPq, and CAPES Brazilian research agencies for partially funding this work.

References (24)

  • M. Amadeo et al.

    Multi-source data retrieval in IoT via named data networking

  • P. Deshpande et al.

    Predictive methods for improved vehicular WiFi access

  • Cited by (10)

    • QoCoVi: QoE- and cost-aware adaptive video streaming for the Internet of Vehicles

      2022, Computer Communications
      Citation Excerpt :

      Rezende et al. [17] proposed a Bayesian model to predict vehicles’ locations and a geographic backbone-based routing approach for video streaming. In [18], the content-centric networking (CCN) concept was employed to deliver content in a VANET. The authors also utilized the vehicle’s trajectory to proactively perform caching of user-requested content over the access points along the path to a destination.

    • A survey of data dissemination schemes in vehicular named data networking

      2021, Vehicular Communications
      Citation Excerpt :

      Regarding the V-NDN data dissemination scheme, data discovery is about transmitting the interest packets via V2V, V2I, or/and hybrid communications in urban or/and high-impact scenarios as a means of discovering the data provider with a named-matched data; meanwhile, data delivery is about returning the data packets back from discovered provider to the data consumer over the interest exact-revers path, with the purpose of improving road safety and travelers' comfort [14]. Despite the inherent features of support mobility, as explained in the previous section, the NDN architecture is not primarily designed for VANET or considered for harsh data diffusion conditions; thus, an adapted NDN architecture that can fit data dissemination in VANET is required [27][122]. Therefore, the existence of a reliable, scalable, efficient, and secure V-NDN data dissemination scheme that can seamlessly integrate VANET and NDN represents a key factor in the success and commercialization of future V-NDN [69].

    • Secure authentication and privacy-preserving techniques in Vehicular Ad-hoc NETworks (VANETs)

      2020, Vehicular Communications
      Citation Excerpt :

      However, when the number of co-operating vehicles increases, this may result in overspending on incentives and hence this approach may not be profitable to advertisers. The Trajectory-aware Content distribution strategy (TraC), proposed by Silva et al. [36], uses Content-Centric Networks (CCN) [37,38] to build persistent proactive caches in RSUs. TraC is based on users' trajectory to increase the probability of content delivery, which was not previously taken into account in the CCN based research works in VANET scenarios.

    • A diversity-based search-and-routing approach for named-data networking

      2019, Computer Networks
      Citation Excerpt :

      These protocols only exploit en-route cached contents in a happen-to-meet fashion disregarding all cache spatial diversity [12]. To fully explore in-network caching, it must exist a best-effort complementary mechanism to the end-to-end routing algorithm in order to retrieve content volatile replicas [13,14]. With that in mind, we introduce the Diversity-based Search-and-Routing (DIVER) mechanism [15].

    • GeoZone: An interest-packet forwarding mechanism based on dissemination zone for content-centric vehicular networks

      2019, Computers and Electrical Engineering
      Citation Excerpt :

      In contrast, proactive approaches delivered all requested contents with an average delay 22 times lower. These results corroborate that opportunistic caching and proactive relaying mechanisms are beneficial for vehicular networks [25]. In Fig. 5(b), the denser scenario, we observe that proactive mechanisms reduce by approximately 50% the delay required to retrieve contents.

    • Vehicle-to-Infrastructure (V2I) offloading from cellular network to 802.11p Wi-Fi network based on the Software-Defined Network (SDN) architecture

      2017, Vehicular Communications
      Citation Excerpt :

      In [24], the authors used scalability and privacy as two important issues to devise a distributed real-time re-routing algorithm for vehicles to reach the congestion avoidance. In [33], the authors combine the users' trajectory and Content-Centric Networks (CCNs) knowledge to proactively cache the content and make the data delivery more efficiently. The above method forwards the individual interests containing information under the consideration of users' destination.

    View all citing articles on Scopus
    View full text