You are on page 1of 6

2018 IEEE International Conference on Vehicular Electronics and Safety (ICVES)

September 12-14, 2018, Madrid, Spain

Towards a Generic IoT Platform for Data-driven


Vehicle Services
Efi Papatheocharous∗ , Emmanuel Frécon∗ , Christian Kaiser† , Andreas Festl† , and Alexander Stocker†
∗ Software and Systems Engineering
RISE SICS, Isafjordsgatan 22, Kista, Stockholm, Sweden
Email: fname.lname@ri.se
† Virtual Vehicle Research Center, Graz, Austria

Email: fname.lname@v2c2.at

Abstract—Advances in the field of engineering have resulted in Taking into account the new advances and trends in the
vehicles becoming a digitised source of data from which scenarios research field of vehicle engineering, our work lists a number
of Quantified Vehicles emerge. Even though the benefits and of challenges starting from a trend, coined as the “Quantified-
range of emerging services are ample, several challenges cap the
extent of opportunities, such as determining the business benefits, self” (QS) or “Quantified Vehicles” (QV), which occurs as
as well as constructing and operating an independent, scalable, the world is becoming more and more equipped of smart
and flexible platform ensuring e.g. privacy, accountability. In our connected things with sensors and actuators. Nevertheless,
work in progress paper, we propose a conceptual architecture of the business and technical benefits are not entirely obvious,
a generic IoT platform for enabling such data-driven services for i.e., how can they be reaped in a successful way, and several
the vehicle domain, while considering important characteristics,
such as data security and privacy, improved service operations, contenders (from start-ups to automotive suppliers, manufac-
safety and value creation for end-users. We then describe how turers up to large software companies) are in the search of the
this platform can be demonstrated, including the vehicle gateway optimum platform and position to host and release it. Thus, a
device (Vehicle Data Logger) capturing the vehicle data, to finally popular trend is the search of a universal Internet of Things
enable a set of useful and usable data-driven services for vehicle (IoT) platform from a variety of emerging commercial and
drivers and other stakeholders.
open source solutions. There are specific requirements that
need to be satisfied, including offering functionality that opens
I. I NTRODUCTION , M OTIVATION AND G OALS
up to new opportunities and more rigid challenges, such as
Automotive electronic systems nowadays expand rapidly privacy, safety and complying with regulations, e.g., the EU
and massively in becoming fully connected and capable of regulation of data privacy enforced recently in all businesses
extended functionalities. Modern vehicles typically consist of (General Data Protection Regulation - GDPR) and the General
several dozens Electronic Control Units (ECUs) linked to Safety Regulation (EC) No 661/2009 in the automotive sector.
communication networks processing data to ensure a vehicle’s Our work in progress paper describes the conceptual ar-
functionality. This rapid development is even more visible chitecture of a generic IoT platform to support security,
in the advances in vehicle software which are increasing safety, automation, operations and management, based on
both in complexity and functionality, relying on platforms the concepts mentioned above and the new advances and
developed entirely for the automotive domain (e.g., the stan- trends outlined, enabling data-driven services in the vehicle
dard Automotive Open System Architecture - AUTOSAR). domain. As a preliminary proof-of-concept, we have developed
These denote extended software programming development a demonstrator which supports two-way connectivity needs of
environments with underlying subsystems, including language, third-party cloud services for quantified connected vehicles.
runtime, components and other associated libraries. Our proposed solution accepts, stores and processes vehicle
These recent developments offer a range of possible service data obtained from a Vehicle Data Logger connected to the
types classified in [1] as: (i) product services, to extend a OBD-II interface of the vehicle. Raw vehicle data is then
product with new or improved functionality (usually targeting analysed on our proposed cloud-based platform with a focus
an end-user group), (ii) process services, to improve the oper- on extracting driving styles and behaviour. The paper shows
ation of a product as part of a larger process (often oriented to how our solution enables the operation and management of
the product owners and having limited control over the actual third-party access to collect, store, transform, process and
product development), (iii) lifecycle services, to use data in visualise data, and then feeding back the result to human
ways that improve the associated lifecycle processes, including drivers or other stakeholders.
predictive and preventive maintenance (usually targeting the The main contribution of our paper is the presentation of
manufacturer of the product), and, (iv) extended services, use a generic IoT platform, building on open source technologies
data from products to improve the operation of decoupled to and containers, for seamlessly enabling novel data-driven con-
the product products or services (targeting administrators and nected vehicle services. We then validate our platform through
governments). a demonstrator, that is feeding interesting information, related

978-1-5386-3543-8/18/$31.00 ©2018 IEEE


to trips and driving behaviour, gained from processing real
data from vehicles to drivers and other stakeholders through a
mobile and a web application.
The paper is organised as follows: Section II covers the
background work and challenges in the area, Section III
analyses our proposed conceptual architecture and solution
walk-through, Section IV is about our proof-of-concept im-
plementation through a demonstrator, Section V provides a
discussion of the main business and technical benefits of the
solution, and, finally, Section VI describes future research
steps and conclusions.
II. BACKGROUND AND CHALLENGES
In this section, we introduce the notion of Quantified Fig. 1. Generic IoT platform design. MQTT is preferred when communicating
back to devices is necessary.
Vehicles, as an extension of the Quantified-self concept, and
we also describe our journey towards realising a generic IoT
platform that proposes a technical solution to support security, companies, looking to reap the business opportunities created
safety, automation, flexibility and scalability. by data.
A. From Quantified-self to Quantified Vehicles The increasing prevalence of several software platforms
The trends of digitising real world events and self- today calls for the ability to augment solutions and support an
digitisation have become increasingly important and have emerging portfolio of leading technology solutions and trends.
transformed humans to data generators [2]. The “Quantified- It is unthinkable to design or use any software technology
self” (QS) is a term coined to describe the quantitative assess- without standing on the shoulders of a multitude of layers
ment of measurable characteristics about a person, including of existing platforms. Popular contenders include: Google
biological, physical, behavioural, and/or environmental aspects Cloud, Microsoft Azure, Amazon Web Services, IBM Cloud
[3]. QS has become a major creator of value with the increase and many more. Industrial vendors ride the Everything-as-a-
number of applications on smartphones and other consumer Service (XaaS) [10] wave, through the provision of computing
devices. E.g., Strava and Runtastic are popular applications resources and platform services for the realisation of “Internet-
to analyse and compare sports behaviour (running, biking, first” applications, offering different possibilities for manage-
etc.) and therefore possess detailed knowledge on what, how, ment, operation and orchestration of the infrastructure.
where and when their users perform sports, which is highly Especially relevant to this paper is a wide range of cloud
privacy relevant. Runtastic was acquired by Adidas in 2015 for platforms tuned to the specific needs of IoT applications.
about 220 million EUR, which underpins the high potential of A documentation analysis of the commonalities between the
exploitable volumes of QS data [4]. architectures of three IoT platforms from Google, Amazon
However, carrying out complex computing tasks with QS and Microsoft and, to a lower extent, from Ericsson AppIoT
data on consumer devices, as well as sending it to a cloud was carried out by a subset of the authors of this paper,
platform can be challenging in terms of bandwidth and latency with the overall aim to realise a generic platform with as few
due to the amount of data produced (autonomous vehicles restrictions as possible. The motivation is wrapped around the
are expected to produce up to 4TB per day [5]). Modern emerging opportunities of: (i) lower entry barriers (i.e., allow
vehicles are equipped with many sensors already and even for new entrants to easily use and contribute the platform),
more will be added for enabling assistive driving function- (ii) increased shared responsibility (i.e., provide joint respon-
alities (i.e., LIDAR/RADAR - LIght/RAdio Detection And sibility and to not just one party), (iii) increased framework
Ranging, video). Several start-ups (including otonomo.io and support, and, (iv) lower technological lock-in.
moj.io), and research initiatives (including automat-project.eu Our analysis resulted in the generic design of an IoT
and aegis-bigdata.eu) have identified and targeted this promis- platform encompassing the common features supported by
ing field, aiming to create value out of QS data of drivers and the platforms analysed, shown in Figure 1. The analysis has
vehicles, termed as “Quantified Vehicles” (QV) [6], [7], [8]. revealed a number of common functionalities and usage of
However, vehicle manufacturers ensure that data, generated for same high-level concepts, even though different APIs and
and used for vehicle functionality, will not reveal any internal implementations occur. One of the contributions of this paper
knowledge or access or intellectual property for safety reasons, is the design of this generic IoT platform and its instantiation
e.g. to not provide to outsiders with the information on how in a demonstrator consisting of data-driven vehicle services
data is actually processed and used within a vehicle [9]. for drivers and other stakeholders.
In the generic IoT platform design Figure 1, devices (any
B. The Outlook for a Universal IoT Platform smart device with connectivity) go through the gateway (a
The development of cloud platforms to enable QV applica- device or software) to accomplish exchange of data with the
tions is a major trend in current technology start-ups and large platform. Some devices can also bypass the gateway. The
exchange is carried out through MQTT [11] or HTTP connec- The platform supports exchange of data with devices (either
tions, even though there is a plethora of other similarly pur- directly or via a gateway) and can accommodate cloud services
posed standards. The platform offers: (i) telemetry ingestion deployed to ingest data, store, process and manage data in
(accepts data), (ii) stream processing (data flows are processed ways to create end-user value. The end-users are broadly
and converted to unified formats), (iii) storage (data is stored illustrated in Figure 2 as connected vehicles, user-groups,
in one or several databases), analytics (data is statistically and operators (e.g., data operators, fleet managers), and, external
semantically analysed to extract information), (iv) machine partners (organisations or governments).
learning (data is processed with machine learning algorithms Potential applications for the end-users support among
to extract knowledge and intelligence), (v) visualisation (data other: (i) driver scores (calculations of risk, financial/cost, effi-
is depicted in meaningful charts and graphs to extract sum- ciency, or safety based on amplitudes of braking, accelerating,
marised information, generalisations, locate anomalies, etc.), cornering, distraction, etc.); (ii) driver tutoring (suggestions,
(vi) lifecycle management (consists of supporting functions recommendations, warnings to improve e.g. safety); (iii) driv-
for the management of devices, such as software updates or ing route recommendations (alternative and recommended
(re)configuration), (vii) state (consists of storing the state of driving paths based on other drivers behaviours, e.g. usual
devices at all given times), and, (viii) apps (consist of extended “stand-still” times, diversions); (iv) entertainment (games: earn
applications and services that can extend the platform and points for driving safe, or become the major of a street)
some offer additional functionality or end-user value). (v) benchmarking (compare with other drivers and award
“safest drivers”); (vi) city planning (suggest alternative routes
C. Provisions for Privacy and Regulations based on heat map overlays and identify potholes or obstacles).
Another challenge encountered is performing secure data
IV. E VALUATION THROUGH A D EMONSTRATOR
operations in cloud platforms provisioning for security, pri-
vacy, safety, trust, etc. which is challenging for the following As a proof-of-concept we have developed a demonstrator
main reasons: (a) the various system types composing plat- comprising of an own-designed Vehicle Data Logger mounted
forms and their purposes are becoming more and more variable on real vehicles, a cloud platform, collecting data and which
in time, (b) infrastructures span over diverse geographical in turn enables two data-driven services: one for individual
locations, (c) rapid distribution in type as well as in growing drivers and one for external stakeholders (e.g., companies and
size and complexity of components and technologies, and, (d) governments). These are realised through a mobile application
there is a growing need in elasticity (resilience) and dynamic and a web application.
in the system’s structure, behaviour as well as interactions. 1) Vehicle Data Logger: Our own-designed and imple-
Thus, developing and maintaining cloud platforms with end- mented Vehicle Data Logger solution (shown in Figure 3),
to-end security and privacy is challenging but at the same is used in multiple vehicles and by different drivers.
time mandatory and a requirement from legislations and The Vehicle Data Logger implementation is based on a Bea-
governments (i.e., GDPR, eCall - the European initiative for gleBone Black single-board computer, a low-cost, community-
emergency rapid assistance to motorists involved in a collision supported development platform. The hardware runs Debian
anywhere in the EU). To cover this gap, cloud platforms need Linux and is extended by a cape, an own-developed Printed
to provision for security, privacy, safety and control over data Circuit Board (PCB) equipped with a CAN chip, and with
in an efficient way. The solution we propose in this work GPS and IMU sensory, switches and buttons. The cape is
mitigates some of these challenges, as explained next. mounted on the BeagleBone’s plug connector. It provides a
software capable of reading telemetry data from the vehicles’
III. C ONCEPTUAL A RCHITECTURE AND S OLUTION OBD-II interface (or directly from the CAN bus system of the
WALK - THROUGH vehicle) when connected to it. Table I provides a list of the
In this section the conceptual architecture of the solution main components used in this low-cost device.
followed by a solution walk-through is described. The archi-
tecture of the proposed solution is composed of a generic IoT TABLE I
B ILL OF M ATERIAL FOR THE M AIN C OMPONENTS
platform to support data-driven services for the Quantified Ve-
hicles scenario and is illustrated in Figure 2. The architecture Type Actual Component Used
Base System Hardware BeagleBone Black
is designed in three layers providing: 1) a cloud platform to Acceleration Sensor ADXL345 (via SparkFun 6 DoF IMU)
support novel QV services, 2) operations, and, 3) management Rotation rate Sensor ITG-3200 (via SparkFun 6 DoF IMU)
provisions. GPS positions Adafruit Ultimate GPS Breakout
CAN/OBD-II data Chip MCP2551 CAN chip
A vehicle logger acts as the vehicle gateway device, and is
designed as a generic data logger to capture OBD-II [12] (or
CAN-bus [13]) data from vehicles as well as GPS, rotation, The sensor data collected includes about five measurements
and acceleration data. The data is then stored directly on the per second for: i) vehicle data, e.g. available OBD-II data like
logger waiting for favourable mobile connection conditions. vehicle speed, ii) acceleration and gyroscope measurements
When connection is established via the mobile network, the (in x, y, and z directions) from the IMU sensor, and iii) the
logger sends the data to the cloud platform. GPS position including the current time-stamp. The data is
Fig. 2. Conceptual architecture of the solution deployed for the Quantified Vehicle scenario. Infrastructure is managed and controlled using Docker (rightmost)
and, supervised and operated in the cloud. The current architecture maps onto the generic IoT platform, using, for example an MQTT broker for telemetry
ingestion, several databases for storage and Grafana as one of the visualisation solutions.

stored on a local buffer and sent to a MQTT broker via a


4G cellular network modem, thus enabling applications for
various stakeholders. An example of visualisation of telemetry
data for one driver is shown in Figure 4. Using a standard
single-board computer running Linux is a clear benefit, while
software and building instructions for the own-developed cape
are planned to be published in the future as well to enable
scientists globally to participate in this research. A public
dataset produced by the Vehicle Data Logger, demonstrating
data quality, is already published at Zenodo [14].

2) Cloud Services: The cloud services architecture de-


Fig. 3. The Vehicle Data Logger mounted in a vehicle (connected to OBD-II, ployed for our demonstrator is already described in Figure 2.
left) and the cape with GPS / IMU sensory. LEDs indicate the software status,
switches / buttons are used for user interaction, e.g. start recording.
Data ingestion from remote vehicles is made primarily through
a MQTT broker, and formatted as JSON[15]. Use of the broker
and the publish-subscribe pattern [16] makes it possible for
remote and external trusted partners to receive raw data if
necessary. Data is transformed and pushed into two separated
time-series databases: InfluxDB and Timescale, a module of
PostgreSQL, two competing solutions that we are still evalu-
ating. Upon trip detection, raw data is analysed and features
are extracted in order to be inserted into the PostgreSQL
database so they can serve as a base for the extended services
described in the next subsection. Raw data can be visualised
through Grafana, even though this is best achieved through
application-specific user interfaces. The cluster provides web
APIs for accessing analysed data, for example from remote
smartphone applications. All access to the services provided by
Fig. 4. Telemetry data visualisation from OBD-II interface. Only a few of the
metrics that are collected in real-time are shown, i.e., engine RPM, vehicle the cluster, including the MQTT broker, is forcefully encrypted
speed, engine load, fuel/air mixture (intake manifold) pressure, etc. with Transport Layer Security (TLS) and certificates from
topic, whereas the lucrative role of the platform provider is
already identified as a key element in the landscape. In times
of increasing awareness and governance on privacy (GDPR),
the drivers producing data may benefit from the innovation
perspective, e.g. by innovative driver assistance systems, but
they will require to have a strong incentive to share their data.
The proposed architecture based on open source tech-
nologies empowers IT companies and freelancers to set up
individual platforms and to develop their own services in
order to create an even more competitive business landscape.
The need for innovation, productivity and reduced time-to-
market in products may dramatically increase and create a
difficult collaborative environment for key players of the
automotive industry striving for competitive advantage. Many
risks still need to be mitigated, e.g., related to licensing, rights,
Fig. 5. End-user application: Analysis of trip data and visualisation data ownership, accountability, negotiations, agreements and
business models.
B. Technical Benefits
Let’s Encrypt1 .
Through a substantial use of the various Docker [17] tools 1) Security: The proposed architecture takes a security-
(Engine, Compose, Swarm, Machine, etc. - see the Manage- by-design approach through a widespread use of network
ment layer in Figure 2) and of Machinery2 , the architecture encryption techniques and careful management of sensitive
can easily be deployed and redeployed at any of the available data within the platform. For example, data is ingested directly,
cloud providers. It also encompasses a number of containers through secure APIs and using renewable TLS certificates.
and solutions for daily operations of the applications, including Within the cloud cluster, network communication occurs
data backups or application supervisions. within encrypted overlay networks and keys are regularly and
3) End-User Applications: For a proof-of-concept, we have automatically rotated. Secrets that need to be shared between
used the data provided by the Vehicle Data Logger and the architectural components are stored in an encrypted and repli-
Cloud Services to enable two end-user applications, one for cated key-value store, and made available as memory mapped
the driver and another one for a city planner. While the files to relevant processes only. Finally, containerisation tech-
first application provides interesting driving statistics including niques provide for a high level of encapsulation between the
safety-relevant events as an overlay on a geographic map in a components, making connections and dependencies explicit
mobile application, the latter application produces a heat map and traceable.
of safety-relevant hotspots in a city to help decision-makers 2) Automation: The cloud infrastructure and application
improving the transport flows in a web dashboard. Figure 5 are deployed and managed by Machinery, a high-level infras-
shows a snapshot of the driver end-user application highlight- tructure management tool at the top of the Docker tooling
ing the events occurring during the trip and providing summary pyramid. Machinery takes a declarative approach to cluster
statistics for the trip (duration, speed, fuel consumption, etc.). management, resulting in describing the entire architecture
(machines, services, networks, etc.) as configuration and code
V. B ENEFITS AND D ISCUSSION that can be managed as a set of project files placed under a
The following benefits, divided to business and technical, version control system - whereas keeps secrets apart.
resulting from the taken approach are worth to discuss. 3) Open APIs: For the realisation of smartphone appli-
cations, or other external services, the platform provides a
A. Business Benefits REST interface3 to the trip data. This interface is application-
Quantified Vehicles excavate a hidden business treasure independent and generic, and it facilitates access to the under-
by simply harvesting sensor data already produced for other lying database from modern back-end services or devices. All
purposes. The dreamlike scenario having live data streaming other components of the architecture have well-documented
from every single vehicle in operation, attracts many data APIs to support communication between the internal services.
scientists and creative minds, not only from the automotive 4) Standardised Data Formats: While MQTT provides a
industry. Hence, several stakeholders (e.g. start-ups, vehicle standard for loosely decoupled communication between inter-
manufacturers, automotive suppliers, IT companies, service ested parties, it leaves several open design decisions, such as
company providers) battle for obtaining the supremacy of this topics organisation and data formats. The proposed architec-
ture uses SenML [15] serialised to JSON as a common data
1 Let’s Encrypt is a certificate authority that provides free certificates for streaming format. JSON was selected because of its ubiquitous
TLS encryption via an automated process (https://letsencrypt.org/).
2 Machinery is an open source software architecture management tool 3 PostgREST (https://postgrest.com/) turns any PostgreSQL database di-
available at: https://github.com/efrecon/machinery rectly into a RESTful API.
implementations across platforms, toolkits and languages, but ACKNOWLEDGMENT
SenML can also be serialised to, e.g. CBOR for improved Part of the work is supported by the Electronic Component
compactness. Integrating further formats, such as the CoRE Systems for European Leadership Joint Undertaking under
link format [18], could provide an embryo to a standardised grant agreement No 737422. This Joint Undertaking receives
management of devices and their capabilities. support from the European Unions Horizon 2020 research and
5) Scalability: Scalability of the solution is principally innovation programme and Austria, Spain, Finland, Ireland,
achieved through the careful selection of architectural com- Sweden, Germany, Poland, Portugal, Netherlands, Belgium,
ponents that have a known migration path to horizontal Norway.
scalability and high availability. In most cases, this means
choosing and configuring components and technologies so that R EFERENCES
they will be able to scale out when necessary. Containers are [1] J. Axelsson, E. Papatheocharous, and J. Andersson, “Characteristics of
software ecosystems for federated embedded systems: A case study,”
orchestrated through Docker Swarm, thus providing a route to Information and Software Technology, vol. 56, no. 11, pp. 1457–1475,
increasing replicas under load. 2014.
6) Flexibility: The major benefit of the declarative ap- [2] P. McFedries, “Tracking the quantified self,” IEEE Spectrum, vol. 8,
no. 50, 2013.
proach is to facilitate migration between cloud vendors through [3] M. Swan, “Emerging patient-driven health care models: An examination
redeployments in other premises, provided that the appropriate of health social networks, consumer personalized medicine and quanti-
credentials are available. By being built on top of Docker fied self-tracking,” International Journal of Environmental Research and
Public Health, vol. 6, no. 2, pp. 492–525, 2009.
Machine, Machinery is not only able to interface with all [4] C. Kaiser, A. Stocker, G. Viscusi, A. Festl, P. Moertl, and M. Glitzner,
major cloud vendors for the creation of virtual machines, but “Quantified cars: An exploration of the position of ICT start-ups vs.
also to facilitate onboarding of existing bare-metal servers, car manufacturers towards digital car services and sustainable business
models,” in Proceedings of 2nd International Conference on New
if necessary. This increases the robustness of the solution Business Models, 2017, pp. 336–350.
by making it independent of any cloud provider. Continuous [5] B. Krzanich. (2016) Data is the new oil in the future of automated driv-
integration techniques permit to constantly improve and update ing. [Online]. Available: https://newsroom.intel.com/editorials/krzanich-
the-future-of-automated-driving/
application-specific containers, and redeploy as necessary. [6] A. Stocker, C. Kaiser, and M. Fellmann, “Quantified vehicles - novel
services for vehicle lifecycle data,” Business & Information Systems
Engineering, vol. 59, no. 2, pp. 125–130, 2017. [Online]. Available:
VI. C ONCLUSION AND F UTURE W ORK https://doi.org/10.1007/s12599-017-0465-5
[7] M. Swan, “Connected car: Quantified self becomes quantified car,”
Journal of Sensor and Actuator Networks, vol. 4, no. 1, pp. 2–29,
This work in progress paper has presented a generic IoT 2015. [Online]. Available: http://www.mdpi.com/2224-2708/4/1/2
architecture for enabling vehicle services and a demonstrator [8] C. Kaiser, A. Stocker, A. Festl, G. Lechner, and M. Fellmann, “A
with end-user applications as a proof-of-concept. The results research agenda for vehicle information systems.” in ECIS, 2018.
[9] AutoMat, “Deliverable D2.5 Cyber Security Framework: 4.1 Security
demonstrated, enable useful and usable set of data-driven ser- and Privacy Engineering Objectives,” RFC6690, 2018. [Online].
vices for vehicle drivers and other stakeholders. The business Available: https://goo.gl/a51DAs
and technical benefits are discussed. [10] Y. Duan, G. Fu, N. Zhou, X. Sun, N. C. Narendra, and B. Hu, “Ev-
erything as a Service (XaaS) on the Cloud: Origins, current and future
Next, as future work, we plan to investigate improve- trends,” in Cloud Computing (CLOUD), 2015 IEEE 8th International
ments and extensions to the proposed conceptual architecture Conference on. IEEE, 2015, pp. 621–628.
[11] A. Banks and R. Gupta, “MQTT Version 3.1.1 Plus Errata 01,”
presented. An example would be the offer to host private OASIS standard, Dec. 2015. [Online]. Available: http://docs.oasis-
containerised Docker repositories to facilitate security and open.org/mqtt/mqtt/v3.1.1/mqtt-v3.1.1.html
privacy configurations for the realisation and integration of [12] “Road vehicles diagnostic communication over controller area
network (DoCAN) Part 4: Requirements for emissions-related
end-user applications. Moreover, we will investigate how to systems,” ISO Standard 15765-4:2016, Apr. 2016. [Online]. Available:
elastically raise up or destroy machines and containers to https://www.iso.org/standard/67245.html
adapt to the demand of vehicle fleets. Similarly, we plan to [13] “Road vehicles – controller area network (CAN) – Part 1: Data link
layer and physical signalling,” ISO Standard 11898-1:2015, Dec. 2015.
extend our implementations to facilitate continuous deploy- [Online]. Available: https://www.iso.org/standard/63648.html
ments of applications or other back-end services. We would [14] A. Stocker, C. Kaiser, and A. Festl, “Automotive Sensor Data.
also like to integrate deeper privacy concerns and regulations, An Example Dataset from the AEGIS Big Data Project,” Jun.
2017, research dataset of the AEGIS project. [Online]. Available:
so as to enable the automatic erase of user-related historical https://doi.org/10.5281/zenodo.820576
content from all media and/or increase anonymisation and [15] C. Jennings, Z. Shelby, J. Arkko, A. Keranen, and C. Bormann,
minimise data inter-relations. Also, our vision is to further “Sensor measurement lists (SenML),” memo 15, May 2018. [Online].
Available: https://tools.ietf.org/html/draft-ietf-core-senml-15
develop cooperative vehicle-infrastructure applications, eco- [16] K. Birman and T. Joseph, “Exploiting virtual synchrony in distributed
driving and assisted driving coaches and propose novel human- systems,” SIGOPS Oper. Syst. Rev., vol. 21, no. 5, pp. 123–138, Nov.
centric interfaces and displays for these applications. Such 1987. [Online]. Available: http://doi.acm.org/10.1145/37499.37515
[17] D. Merkel, “Docker: Lightweight linux containers for consistent
implementations will be further investigated in future work development and deployment,” Linux J., vol. 2014, no. 239, Mar. 2014.
and compared to existing implementations which enable data- [Online]. Available: http://dl.acm.org/citation.cfm?id=2600239.2600241
driven vehicle services, in order to identify actors and their [18] Z. Shelby, “Constrained RESTful Environments (CoRE)
Link Format,” RFC6690, Aug. 2012. [Online]. Available:
application domains which participate in such platforms, and https://tools.ietf.org/html/rfc6690
their value models.

You might also like