A Network Stack for Computation-Centric VehicularNetworking

Dennis Grewe*, Claudio Marxer‡, Christopher Scherb‡, Marco Wagner*, Christian Tschudin‡* Robert Bosch GmbH, Corporate Sector Research and Advance Engineering, Renningen, Germany

{dennis.grewe,marco.wagner3}@de.bosch.com‡University of Basel, Department of Mathematics and Computer Science, Basel, Switzerland

{claudio.marxer,christopher.scherb,christian.tschudin}@unibas.ch

ABSTRACTRecently, vehicle-to-vehicle (V2V) and vehicle-to-infrastructure(V2I) connectivity transitioned from a vision of the future to reality.Applications in such environments vary from local propagation ofroad conditions to large-scale traffic flow control systems. In thisdemo, we present a network stack for the data exchange in the auto-motive IoT, based on the Named Function Networking (NFN) prin-ciples. In NFN, the communication model is not restricted to propa-gation of static data but natively supports computation-offloadingto other nodes. We present solutions and report on experimentswith real cars on a test course.

CCS CONCEPTS• Networks→ In-network processing; Naming and addressing;Network protocol design;

KEYWORDSConnected Vehicles, Named Function Networking, Mobile EdgeComputing, Information-Centric Networking

ACM Reference Format:Dennis Grewe*, Claudio Marxer‡, Christopher Scherb‡, Marco Wagner*,Christian Tschudin‡. 2018. ANetwork Stack for Computation-Centric Vehic-ular Networking. In 5th ACM Conference on Information-Centric Networking(ICN ’18), September 21–23, 2018, Boston, MA, USA. ACM, New York, NY,USA, 2 pages. https://doi.org/10.1145/3267955.3269015

1 INTRODUCTIONContent-centric networks (CCN) offer the service to deliver infor-mation based on the content’s name [2]. This means, clients arereleased from holding server locations and request content in ahop-by-hop fashion. As an extension, Named Function Networking(NFN) allows users to request for derived, i.e. dynamically produced,data [7]. In NFN, named functions are published in the network,which can be applied to any available content on demand. For thispurpose, a user composes a data name which encodes the appli-cation of named function(s) to certain data. It is the task of thenetwork to compute and deliver the result. Behind the scenes, the

Permission to make digital or hard copies of part or all of this work for personal orclassroom use is granted without fee provided that copies are not made or distributedfor profit or commercial advantage and that copies bear this notice and the full citationon the first page. Copyrights for third-party components of this work must be honored.For all other uses, contact the owner/author(s).ICN ’18, September 21–23, 2018, Boston, MA, USA© 2018 Copyright held by the owner/author(s).ACM ISBN 978-1-4503-5959-7/18/09.https://doi.org/10.1145/3267955.3269015

Figure 1: NFN on ETSI ITS-G5 protocol stack.

network implements a forwarding strategy which decides where acomputation is executed [6].

Mobile Edge Computing with NFN. In challenged networkssuch as connected vehicle environments, Edge Computing (EC) hasbecome popular to bring computational resources to the edge ofthe network to produce and manage data closer to the consumers.By combining NFN and EC, the computation of data is performedonly few hops away from the consumer. In NFN, the identificationof the computation location is realized by using so-called resolu-tion strategies. As part of this work, the Find and Execute (FaX)forwarding strategy is used, originally designed for mobile IoT en-vironments [5]. An exemplary presentation of the strategy can bedescribed as follows: it is expected, that a result of a computationintensive operation can not always be shipped directly to a mobileconsumer. In FaX, the closest computation-capable node starts thecomputation immediately, and checks adjacent computation nodesfor a already computed result in parallel. As a result, FaX reducesthe resolution time and increases data delivery success rate.

This work contributes a full stack implementation of NFN forthe 5th Generation of Intelligent Transportation Systems of theEuropean Telecommunications Standards Institute (ETSI ITS-G5)connected vehicle protocol stack. Furthermore, the stack has beentested including real cars and a infrastructural deployment showingthe offloading of computations from vehicles to the infrastructure.

2 INTEGRATION OF NFN INTO ETSI ITS-G5SYSTEMS

The stack developed in this work integrates the NFN implementa-tion PiCN and enables vehicular units to exchange ITS-G5 as wellas CCN messages between other vehicles and an infrastructure inparallel. Figure 1 illustrates the process of integrating NFN into

ICN ’18, September 21–23, 2018, Boston, MA, USA D.Grewe, C.Marxer, C.Scherb, M.Wagner, C.Tschudin

Interest: I1 Interest: I1

Car Movement

Computation

Search for cached Result

Result

Result 1

3

4

5 2

6

Computation5

7

RSU1RSU2

Figure 2: Scenario of field experiments.

the stack. Existing safety-critical applications are still supportedusing the ETSI ITS-G5 stack, while computations can be offloadedto infrastructural components using the NFN layer.

The layers used in the stack include:• IEEE 802.11p access layer in the 5.9 GHz band through aLinux kernel modification [4]. As part of the prototype, theITS-G5D band (5905MHz - 5925MHz frequency with 10MHzchannel spacing) and the service channel G5-SCH5 are usedwhich are reserved for non-safety future applications [1].

• ETSI ITS-G5 Network & Transport layer (including the De-centralized Congestion Ccontrol (DCC), Geo-Networking(GeoNet), CAM/DENM) based on OpenC2X [3].

• OpenC2X NFN module based on the NFN implementationPiCN 1. The module implements the required CCN datastructures as well as the FaX resolution strategy.

3 SYSTEM DEMONSTRATIONThis work demonstrates two contributions: (i) computation offload-ing supporting mobility using the novel network stack and (ii) ademonstration of the stack in a real world automotive deploymentincluding road-site units (RSU) and vehicular units (VU) using ETSIITS-G5.

The feature set of the stack will be demonstrated on a RSU anda VU using IPC boards (NF36-2600 - 1GHz CPU, 1GB RAM). Aspart of the demonstration, the communication and the computationoffloading capability are presented using NFN computations ina simple two node setup. Furthermore, we will present our fieldexperiments used to validate the system (cf. Figure 3) The vehiclesends out NFN requests for a computation intensive operationduring the journey (e.g., augmentation of a map) (Figure 2, part1). A RSU (here: RSU1) starts the computation (Figure 2, part 2),while the car moves out of the communication range of the RSU1(Figure 2, part 3). As soon as the car connects to RSU2, it sendout the NFN request again (Figure 2, part 4). Based on the FaXstrategy, RSU2 starts the computation and searches for a cachedresult in the network (Figure 2, part 5). Since the previous RSU1 hasfinished the computation, the result can be directly transported toRSU2 (Figure 2, part 6). Finally, the result can be directly shippedto the car (Figure 2, part 7). This field experiment has shown the1https://github.com/cn-uofbasel/PiCN

Figure 3: Vehicular unit hardware on the field test.

offloading of computations to the infrastructure and demonstrateshow a result can be fetched, in case the computation time is longerthan the time the car is connected to a single RSU.

For the demonstration of the automotive scenario, mobility is akey aspect and the FaX forwarding strategy is designed for highlymobile use cases. Since it is difficult to demonstrate mobility withmoving vehicles due to space constraints in the demo session, in thesecond part this aspect will be presented with a video, a simulationand a poster. We report on experiments involving real cars on atest course.

REFERENCES[1] European Telecommunications Standards Institute. [n. d.]. Intelligent Transport

Systems (ITS); Access layer specification for Intelligent Transport Systems operatingin the 5GHz frequency band (ETSI EN 302 663 V1.2.1), journal = European Standard(Telecommunications series), year = 2013. Technical Report.

[2] Van Jacobson, Diana K. Smetters, James D. Thornton, Michael F. Plass, Nicholas H.Briggs, and Rebecca L. Braynard. 2009. Networking Named Content. In 5th Proc.Int. Conf. on Emerging Networking Experiments and Technologies (CoNEXT). ACM,New York, NY, USA, 1–12.

[3] S. Laux, G. S. Pannu, S. Schneider, J. Tiemann, F. Klingler, C. Sommer, and F.Dressler. 2016. Demo: OpenC2X x2014; An open source experimental and pro-totyping platform supporting ETSI ITS-G5. In 2016 IEEE Vehicular NetworkingConference (VNC). 1–2.

[4] R Lisovy, M Sojka, and Z Hanzálek. 2014. IEEE 802.11p Linux Kernel Implemen-tation. (2014).

[5] C. Scherb, D. Grewe, M. Wagner, and C. F. Tschudin. 2018. Resolution Strategiesfor Networking the IoT at the Edge via Named Functions. In Proceedings of theIEEE Consumer Communications & Networking Conference (CCNC). Las Vegas,USA.

[6] Christopher Scherb and Christian Tschudin. 2018. Smart Execution StrategySelection for Multi Tier Execution in Named Function Networking. In 2018 IEEEInternational Conference on Communications Workshops (ICC Workshops). IEEE.

[7] Manolis Sifalakis, Basil Kohler, Christopher Scherb, and Christian Tschudin. 2014.An Information Centric Network for Computing the Distribution of Computa-tions. In 1st Proc. Int. Conf. on Information-centric Networking (ACM ICN). ACM,New York, NY, USA, 137–146.