Structuring of Methods to Estimate Benefits of Partial Networking
Alexandr Vasenev
ESI, TNO Joint Innovation Centre, Eindhoven, The Netherlands
Keywords: Partial Networking, ECUs, Potential Energy Savings, System Segmentation.
Abstract: Partial Networking, as a mechanism for moving-to-sleep and waking-up embedded systems, is beneficial for
saving energy within a vehicle (or within other complex distributed systems). Even though a number of
models exist which identify benefits of partial networking, they often address rather specific cases. Moreover,
these fragmented efforts do not necessarily make explicit which methodological steps were taken. Explicating
and analysing methodologies of existing research is beneficial to construct an overarching structure how to
estimate potential energy savings for partial networking implementations. This structure can be used to select
which steps to take to investigate the savings, and how to construct an argument for presenting the findings.
This paper describes initial results of such a research. It reviews several models, illuminates their (sometimes
not explicitly documented) methods, and outlines a generalized sequence for estimating partial networking
benefits. Besides, it provides a list of questions to consider when introducing partial networking. The outlined
methods and the analysis can be of interest to other domains interested in energy savings, such as smart grids,
smart cities, and internet of things.
1 INTRODUCTION
Saving energy by reducing consumption of in-vehicle
components, such as ECUs (Electronic Control
Units), is essential for both electric as traditional
vehicles. Specifically, switching ECUs from a fully
active state to a less power demanding state can
provide several advantages (Butzkamm and Bollati,
2012):
- Reducing energy consumption which leads
to less CO2 emission (thus, tax advantage)
and increased range of electric vehicles;
- Optimizing of operational strategy when the
vehicle is at rest can support: new comfort
functionalities (e.g., the sun roof control);
reduced strain on the battery; longer time
periods when the engine can be re-started;
less fuel is needed to charge the battery;
- Shorter operating time of embedded systems
can lead to reduced lifetime requirements for
ECUs.
Partial Networking (PartialNW), as a mechanism
to put to sleep and wake up ECUs, can provide
significant energy savings. It requires models to track
and optimize system-level energy consumptions.
Such models can provide inputs for the following
considerations (Lingadahalli et al., 2016.):
- Feature deployment to ECUs for optimum
power consumption;
- Comparison of various electrical
architecture alternatives;
- Power budgeting for features;
- Defining the fuel economy drive cycle
impact.
This paper reports intermediate outcomes of
research directed at systematizing methodologies and
constructing a framework to estimate benefits of
partial networking. It aims to provide inputs for
designing various PartialNW models best suited at
different product development stages, based on the
available level of details. New models can be
constructed by linking inputs (functions/ECUs) to
resulting values (CO
2
, energy reduction) through the
calculation stage. Some values relevant for models
are mentioned next to references to best practices of
modelling. To note, these values are referred as they
are mentioned in the cited publications and are thus
linked to contexts and assumptions of those articles.
Nevertheless, they can serves as first order estimates
and provide useful insights to the models.
The next sections introduce PartialNW and
overview models to estimating its benefits. A
generalized sequence of estimating benefits and list
some relevant issues is presented afterwards.
Vasenev, A.
Structuring of Methods to Estimate Benefits of Partial Networking.
DOI: 10.5220/0006803205750581
In Proceedings of the 4th Inter national Conference on Vehicle Technology and Intelligent Transport Systems (VEHITS 2018), pages 575-581
ISBN: 978-989-758-293-6
Copyright
c
2019 by SCITEPRESS – Science and Technology Publications, Lda. All rights reserved
575
2 BACKGROUND
This section starts with an overview of the structure
of power savings related to ECUs, continues by
introducing partial networking, and then lists
potential ECU candidates for partial networking.
2.1 Power Saving Techniques
In general, techniques to obtain power savings for
embedded systems include the following (Heinrich
and Prehofer, 2013):
1. Improving consumption of an ECU (i.e. CPU)
by dynamic hardware resource management:
a. Dynamic voltage/frequency scaling;
b. Reducing quality of service to decrease
energy consumption;
c. Energy-efficient task scheduling;
2. Improving system-wide energy consumption:
a. Dynamic power management (DPM),
which deactivates unused components of a
system to save energy;
b. Dynamic voltage scaling and DPM -
considers the energy consumption of
peripherals (e.g., memory) within standby
and activation/deactivation time of the
processor.
Such opportunities benefit from three degenerated
states of ECUs (Schmutzler et al., 2010:
1. ECU degradation decreasing the clock rate or
feature set;
2. Stop mode the microprocessor is in a
powered stop mode, while all subsystems
except the communication subsystem are
deactivated;
3. Sleep mode.
This paper concentrates on the latter state, i.e., sleep
mode of ECUs. Specifically, the paper concerns
methods of estimating benefits of Partial Networking
(PartialNW) in connection to energy savings. Other
drivers, such as wiring rationalization, system
segmentation, and potential impact on security (e.g.,
in connection to e.g. the extended vehicle concept)
are not covered here in detail.
2.2 Introduction to Partial Networking
PartialNW, is a coordinated go-to-sleep and wake-up
protocol. As part of a network management
middleware it aims to maximize the time ECUs sleep
to save energy. Its use can realize an additional
number of viable sleep scenarios and thus be
beneficial for the current consumption of functions
when the vehicle is at rest or moving.
PartialNW is a further step in granularity of
selective sleep of ECUs, compared to the bus-wide
sleep. This bus-level sleep is directly linked to ECUs
located on a specific bus, and thus to reasons why and
how the bus was constructed. This can constrain the
network architecture, because of a bus (as a sub-
network) purpose to: (1) correspond to a functional
domain, (2) help to restrict the impact of bus or node
malfunctions, and (3) decrease the wire length of
individual bus arms. For instance, residential
functions can be linked to a specific CAN. PartialNW
relaxes assumptions behind such bus composition
logics. In PartialNW, nodes form clusters which wake
up and respond on demand. If a particular cluster is
not requested, its participants can move to sleep. This
mechanism allows ECU nodes to enter sleep mode
even if the bus is active.
Introduction and elaboration of selective sleep
affects network communications and the process of
its design due to challenging the following
paradigms:
- Cyclic communication: network management
policies allow nodes to move to sleep and wake
up again. Therefore, some nodes do not always
participate in communication;
- Deployment (Heinrich et al., 2016): In 'Energy-
focused allocation', software (SW) components
are places to reduce the energy demand of the
system. For instance, some SW components can
be grouped. Within 'Function based allocation'
contrarywise suppliers provide hardware
(HW) and SW components as an integrated
system, which can reduce number of active
ECUs needed. Modern SW architectures, such
as AUTOSAR, introduce independence of SW
components from HW components and help to
move from the function-based to energy-
focused allocations.
While participating in PartialNW, a
microcontroller may switch into an unpowered mode.
The sleep to active transition time can be up to 100
ms. Sleep mode switches off the power supply of
peripherals and microcontroller. Starting can be
compared to a cold boot of an ECU (in comparison,
Pretended Networking can bring the ECU into the
stop mode, where the current consumption would be
higher). After moving from sleep to active, the node
has an obsolete representation of the system, which
might need to be identified again. Potentially, that
status can be stored in a persistent memory, but it may
require bigger memories.
In case of Automotive Ethernet, additional wake
up specifics relevant to PartialNW apply due to the
need to establish communication links. Three wake-
VEHITS 2018 - 4th International Conference on Vehicle Technology and Intelligent Transport Systems
576
up mechanisms can be differentiated as follows
(Suermann and Müller, 2014):
- Selective wake-up. For instance, the time taken
to establish a link according to the BroadR-
Reach specification this can be up to 200 ms. If
four consequent links shall be established it
might sum up to 800 ms, which is unacceptably
long for many in-vehicle functions;
- Global wake-up via a separate wake-up line
(therefore, the need for a wake-up line arises);
- Global wake-up via the Ethernet network, when
transceivers re-send the wake-up signal. The
wake-up message spreads swiftly throughout
the network. Unneeded clusters are then
switched off again.
PartialNW could employ a local decision unit
which analyses bus traffic and decides if a s single
node is needed (thus, generate a local wake up event).
In addition, other nodes must be informed about
ECUs which are about to sleep, because a sleeping
ECU could generate errors in other ECUs. Equipped
with the state of other ECUs, nodes they can decide
whether a signal is missing because the responsible
cluster or ECU is sleeping, or if an error occurred.
2.3 Candidates for Partial Networking
To benefit from PartialNW, the distributed system
properties should include (Huber, 2010):
1. Functions need to have large execution periods,
so ECUs can go to sleep;
2. Functions should be distributed to ECUs
adequately (in the view of a distributed vs
centralized ECU architecture). In other words,
sleep modes cannot be applied, if a single unit
performs many essential vehicle functions.
As a result, PartialNW candidates are ECUs that
do not require cyclic CAN communications, e.g.:
- parking assistance, lane departure, or other
systems to be used at specific speeds;
- door, mirror, roof, and seat control modules;
- some elements from the infotainment segment or
special accessories (e. g. trailer control units).
3 OVERVIEW OF MODELS TO
ESTIMATE BENEFITS OF
PARTIAL NETWORKING
This section overviews state of the art solutions to
estimate PartialNW benefits. It explicitly outlines the
methods behind such calculations. The section starts
with first order estimates, introduces the logic of
grouping ECUs based on power consumption, and
finally overviews several models.
3.1 First Order Estimations of CO2
and Fuel Savings
To calculate fuel savings, one can employ the
heuristic that an increase of 100 Watt means that fuel
consumption rises by 0.1 litre per 100 km. This in turn
leads to an increase in CO2 emissions of 2.5 gram per
km (Monetti et al., 2012). Alternatively, energy saved
by ECUs can be linked to savings in fuel using caloric
value of petrol (Schmutzler et al., 2010):
- Assume a caloric value of 10 kWh per liter of
petrol and energy efficiency of a car with a
combustion engine as 0.225 (efficiency of
engine 0.25 * efficiency of a generator 0.9);
- With 3.84 W*t energy savings of sleeping
ECUs, Fuel consumption = 3.84/(0.225*10) =
1.71 ml/h * t;
- Fuel reduction is 0.56ml (With Motor Vehicle
Emissions Group A driving cycle t=1180s and
11km);
- Reduction of CO
2
is 0.12 g/km (Assuming
burning 1L of petrol to produce 2.33 kg CO2).
Also, PartialNW can be linked to the amount of
money saved by avoiding CO
2
penalties (Huber,
2010)):
- Energy saved per ECU is found by multiplying
differences in current and voltage
consumptions.
- Assuming that N network nodes can be put into
selective sleep, total energy saved per vehicle
can be calculated by multiplying the previous
value by this number N;
- Assuming 0.0265 grams of CO
2
emissions per
kmW, CO
2
saved per km can be determined;
- The resultant CO
2
savings are multiplied by 95
Euro penalties for exceeded emission (CO
2
per
km) to find savings per vehicle;
- Cost reduction per ECU can be found by
dividing the savings by N nodes.
NB: the values of 0.0265 grams of CO
2
emissions
per kmW and 95 Euro penalties (as intended for the
EU starting in 2015) are mentioned in (Huber, 2010)
without further details. These values depend on the
car model used for calculation and the regulatory
context of the car use. While the values can change,
the outlined method might still hold.
3.2 Inputs: Classes of ECUs
For modelling purposes, power requirements of
ECUs can be grouped into classes. For instance, such
Structuring of Methods to Estimate Benefits of Partial Networking
577
classes can be linked to how, e.g., NXP
Semiconductors differentiates between several types
of applications: advance driver assistance systems, in-
vehicle networking, body, chassis, powertrain and
safety (NXP, 2018). In another example (Schmutzler
et al., 2010), four classes of ECUs are distinguished
in connection to average power consumptions, i.e.,
high-end power train, high-end, mid-end, and low-
end (average current consumptions in the mentioned
article are 400mA, 70 mA, 20 mA, and n/a).
3.3 Estimations using High Level
Scenarios
Scenarios may be uses to determine energy savings
(such as driving and hybrid charging). (Schmutzler et
al., 2010). The logic behind this method is as follows:
1. The power usage of automotive controllers is
assumed to correlate with their feature sets and
processing power;
2. Several power classes are distinguished: high-
end power train, high-end, mid-end, and low-
end based on their consumption (Section 3.2);
3. The supply voltage is used for calculations, as
most ECUs use linear regulators. Assumptions
are: average supply voltage level (12.6V) and
an average 1mA quiescent current per ECU
while sleeping;
4. Two scenarios are outlined (driving scenario A
and a stationary charging scenario B);
5. By estimating which ECUs can sleep, power
categories per scenario are outlined. In scenario
A 2 high-end and 13 mid-end ECUs are
required only 20% of time.
6. Energy savings are estimated as follows:
- Energy savings in Scenario A are calculated as
(320mA-15 (sleeping ECUs) *1mA) *12.6V *
t =3.84W * t;
- Energy savings in Scenario B are calculated
24.26W * t (with 35 ECU sleeping);
- Yearly savings=60.64 kWh (charging cycle 10
hours, 250 times a year) = 24.26*10*250.
3.4 Detailing Scenarios as a Set of
Functions
To detail a scenario, a sequence of function can be
linked to a timeline. For instance, w.r.t. scenario A
from the previous subsection (i.e., driving), functions
linked to ECUs can be as shown in Table 1 (example
from Yi and Jeon, 2015).
Table 1: Sequence of functions within a scenario.
Operating
sequence
Operating
Detail
Operating
ECU
Related
ECU(s)
1
Ignition on
All ECUs
All
2
Adjust Seat
ECU 1
2,3,4,
10
Operating Rear
View Camera
ECU 9
10,11,12
11
Ignition off
-
3.5 Steps to Estimate Impact of
Networking Mechanisms on
Measured Data
Steps to estimate impact of sleep mechanisms using
measured data (e.g., on prototype vehicles) can be
structured as implied in (Hong et al., 2016):
1. Select ECUs that can be assessed:
- Obtain measurement data;
- Identify if an ECU can sleep (with not too
much centralized architecture) or move to
stop. Consider whether an ECU does not need
to operate permanently (e.g., other ECUs
don’t rely on the messages) and is not safety-
critical, e.g. powertrain.
2. Perform test drives to identify ECUs that don’t
sleep (in the mentioned paper Vacuum Pump
ECU and Body Control Module ECU (as a
combination of two ECUs: 1.with accessories
ON and 2.Accessories OFF)):
- Document behavior at startup;
- Analyze when and how the ECU starts to work
(switch-on requirements);
- Document applicable scenarios (e.g., 1. Urban
Test Drive and 2. Freeway Test Drive). Note
relevant events and context (e.g., rain
conditions are relevant to wipers; speed can be
derived from the driving cycle);
3. Estimate energy savings if the ECU could sleep
or move to stand-by in such scenarios.
3.6 Estimations in the Design Process
Different product development stages can provide
inputs for estimating PartialNW savings as follows
(Heinrich and Prehofer, 2013b):
- At the System requirements stage designers
can envision Hardware (number of ECUs,
ECU Energy Classes (see Section 3.2) and
software (Number of features, feature
computation class) elements;
- System Architecture stage adds details on HW
(Topology, Network Bandwidth) and SW
VEHITS 2018 - 4th International Conference on Vehicle Technology and Intelligent Transport Systems
578
(architecture as black box, tasks, task
computation efforts);
- System design adds to HW network-specific
energy consumption and elaborates, e.g.,
using function point analysis, SW
computation efforts of tasks;
- HW/SW Development elaborates ECU Active
and sleep times and Task details.
3.7 Function Deployment
Lingadahalli et al., 2016 outline the following way to
detail function deployment and alternatives (using
Simulink SimEvents):
- Inputs to the model: (1) Feature activation
times (as time_deactivated-time_triggered);
(2) Feature-to-ECU deployments (with
alternatives), e.g. a feature deployment with
participating ECUs; and active and inactive
power consumption values (Table 2).
- Output: Power partial NW=sum(consumption
of all active ECUs)+sum(consumption of all
inactive ECUs)+x (extra power due to NW
delays and internal times).
Table 2: Consumptions of ECUs w.r.t. deployments.
ECU
Active
State
Current
Inactive
State
Current
CAN
Bus
Deploy-
ment 2
(alter-
native)
ECU1
300 mA
100 uA
Bus1
Active
ECU2
200 mA
50 uA
Bus3
Active
ECU3
500 mA
200 uA
Bus2
Inactive
ECU4
800 mA
100 uA
Bus1
Active
ECU5
50 mA
5 uA
Bus1
Active
As a next step, mapping functional chains (Walla
et al., 2012) (which are active in specific conditions)
to different ECUs can help to investigate potential
power savings in more detail. For instance, if a
functional chain is not needed for a vehicle speed
greater than 20 km/h, corresponding ECUs can
degrade their performance (a technique mentioned in
Section 2.1).
3.8 Detailed Communications Model
A detailed model of power savings can consider
activation/deactivation of components (including SW
components within an ECU) and communication
power demands (Heinrich et al., 2016). It can include
CAN/Ethernet communications (CAN/Ethernet
communication controllers, transceivers, power
needed to send a message between networks, time and
power for activate/de-activate an ECU and SW
components). The corresponding steps could be as
follows:
1. Assume power consumptions w.r.t.
communications, such as:
- CAN: data rate 500 Kbit/s. Eight bytes per
message are user data and 44 bits are
communication overhead;
- Ethernet: data rate 10 Mbit/s. Maximum 1500
bytes per message are user data and 144 -
communication overhead;
- ECU boot time: 250 ms (100-200 ms to boot,
then to receive messages to work with), 2.5 W.
To de-activate -2.5W;
- To activate/deactivate a SW component
5.61mWs (5ms for self-test, load data, 1122
MW power by microcontroller);
- To transfer a message between NWs: 0.12
mWs, transmission time 0.1s;
- Energy relevant parameters of the network
(including communication controllers and
transceivers);
2. Construct dependencies of functional elements
per function; account for transferred bytes;
3. Identify components needed for speed ranges;
4. Use a context parameter (speed from the New
European Driving Cycle);
5. Consider alternative allocations of [SW
component / sensor / actuator] structure.
Compare Function-based vs Energy-focused
allocations (as mentioned in Section 2.2);
6. Obtain energy demands for: CPU, ECU offset
(the energy demand of components such as the
power supply unit and the voltage regulator),
Sensors/actuators, Communication (Comm.
Connections, transfer, listener, energy saving
mode), and adaptivity (activation/de-activation
of SW components, activation/de-activation of
ECUs)).
4 ANALYSIS
The models outlined above work with different (types
of) inputs and address specific questions. They can be
linked to each other in a generalized sequence of
estimating partial networking benefits as follows:
1. Inputs: scenarios, functions, details of ECUs
(potential function deployments, energy
consumption or energy class), communication
details (if available); →
2. Processing as actions to: Consider (alternative)
function deployments to ECUs Identify
ECUs that can be put to sleep (by estimating
the amount of sleeping ECUs or identifying
Structuring of Methods to Estimate Benefits of Partial Networking
579
ECUs based on function-to-ECUs mappings]
Identify ECU consumption (measurements
or estimations individually or per ECU class);
3. Output as values of: fuel saved less CO
2
money saved by avoiding the CO
2
penalty.
The processing step heavily depends on the
adopted function/feature deployment (i.e., mapping)
alternatives. As the result of analyzing the literature
listed in this publication, the following (exemplary)
questions can be considered:
1. Can an ECU avoid sending those cyclic
messages that other ECUs rely on?
2. Can an ECU stay free from functions that
demand very low-latency replies?
3. Can repetitive tasks (that do not require the
processing power and flexibility of a
microprocessor) be offloaded? E.g., can an
ambient light system driven by Pulse Width
Modulation (PWM) run on a dedicated
hardware (not on the microcontroller)?
4. Can a mapping improve the response time?
5. If a group of sub-functions within an ECU is
not needed, can that ECU reduce its energy
consumption (in connection to overall structure
of energy savings mentioned in Section 2.1)?
6. Can a specific set of (sub-) functions be linked
to an ECU for energy saving:
- sub-functions not needed at specific
conditions (such as parking assist, if the
vehicle speed is greater than 20 km/h);
- a specific mode (e.g., an ECU with comfort
functionality within a parked vehicle)?
Some issues mentioned above need to be addressed
before adopting PartialNW, e.g.:
- state of ECUs after wake-up might need to be
renewed;
- errors shall be envisioned if ECUs don't know
whether other ECUs are sleeping or awake;
- potential need for extra Sleep Support that
tracks ECUs and supports wake-up/shut-down
sequences.
- timing needed (i.e., budgeting) for specific
functions, if some ECUs still need to wake up.
Related to the latter, analysing Ethernet wake-up
specifics can suggest ways to account for
performance and system segmentation based on
separating wake-up, go-to-sleep mechanisms, and in-
vehicle network segments. Specifically, it includes:
- Step-by-step wakeup. (Time to establish all
links shall be compared to the performance
needed);
- Need for an extra wire. This measure to
partition the system corresponds to a wake-up
mechanism that acts in parallel and helps to
avoid the delay of the step-by-step wakeup.
Such a measure is linked to the wiring
rationalization, when a power line or a relay is
used to wake up a group of ECUs;
- Global wake-up can complement a selective
go-to-sleep mechanism.
Altogether, moving away from the logic of having
an ECU per major (group of) functions, which can
restrict possibilities of selective sleep, Partial NW can
assist in energy savings. This section highlighted
several aspects to facilitate (cross-related) use of
relevant models, including a generalized sequence of
estimating PartialNW benefits, relevant processing-
related questions, and emerging budgeting and other
related issues.
5 CONCLUSIONS
Considering the potential energy savings that
PartialNW can provide, there is a need to understand
ways to capitalize on it. Unfortunately, the literature
on estimated potential savings sometimes misses the
description of the methodology steps, does not
explicitly outline the context, nor contain discussions
on applicability of such methods. Explicating
methods behind calculations and aligning existing
them can help to construct a structure for future
research on estimating potential energy savings. New
methods may address practical questions how to
assess specific cases by integrating (parts of) different
models and thus the decisions if specific approaches
should be adopted. While following the general
methodical steps, those new methods could be more
detailed and tailored to a specific implementation.
This paper reviewed articles related to partial
networking and illuminated their (sometimes not
explicitly documented) methods. Equipped with this
information, a designer can estimate power savings
for their system of interest using these methods and
first order values listed in this paper, as well can
further study the mentioned literature. A researcher
may further investigate the methods and construct
new models and methods by tabulating and linked
existing approaches.
Within a future research agenda, several aspects
missing in the relevant literature can be addressed,
such as connections between the context of this
research and moving-to-sleep and waking-up
mechanisms. The relation to design science could be
further strengthened. To provide such a
comprehensive view, future research may study the
generalizability limits of methods to estimate benefits
VEHITS 2018 - 4th International Conference on Vehicle Technology and Intelligent Transport Systems
580
of selective sleep. It can describe a case on applying
generalized sequence of estimating partial
networking benefits, including relating it to concerns
of stakeholders. Moreover, the links between the in-
vehicle partial networking and other energy-
conscious domains (and their approaches) can be
investigated. On example is the applicability of the
methods described in this paper to other systems, such
as smart grids or smart cities.
ACKNOWLEDGEMENTS
This work was partially supported by industrial
funding and TKI (Topconsortia voor Kennis en
Innovatie) program. I would also like to thank Teun
Hendriks for comments and suggestions.
REFERENCES
Butzkamm C, Bollati, D., 2012, Partial Networking for
CAN bus systems: Any saved gram CO2/km is
essential to meet stricter EU regulations, Online
resource, available at: https://www.can-cia.org/
fileadmin/resources/documents/proceedings/2012_but
zkamm.pdf.
Heinrich, P., Oswald, E., Knorr, R., 2016, Energy saving
potential of adaptive, networked, embedded systems: A
case study. In The Sixth International Conference on
Smart Grids, Green Communications and IT Energy-
aware Technologies. ENERGY, June 26 - 30, 2016,
Lisbon, Portugal.
Heinrich, P., Prehofer, C., 2013, Network-wide energy
optimization for adaptive embedded system, In
SIGBED Rev. 10, 1 (February 2013), 33-36.
Heinrich, P., Prehofer, C., 2013b, Early energy estimation
in the design process of networked embedded systems”,
In 3rd International Conference on Pervasive and
Embedded Computing and Communication Systems,
PECCS 2013.
Hong, W., Viehl, A., Lin, J., Bringman,, O., Rosenstiel, W.,
2016, Impact analysis of AUTOSAR energy saving
mechanisms for automotive networks. In IEEE
Intelligent Vehicles Symposium (IV), Gothenburg,
2016, pp. 1097-1102.
Huber, H., 2010, Partial Networking: In-Vehicle Networks
Can Reduce Costs and CO2 Emissions. Online
resource, available at: https://www.ecnmag.com/
article/2010/03/partial-networking-vehicle-networks-
can-reduce-costs-and-co2-emissions.
Lingadahalli, S. Maydiga, S., Darin, M., 2016, Model
Based Approach for Analysis of In-Vehicle CAN
Partial Networks Power Consumption. In SAE
Technical Paper 2016-01-0064.
Monetti, V.A., Otter, T., Ulshöfer, N., 2012, Spritverbrauch
senken, Reichweite erhöhen System-Basis-Chip für den
Teilnetzbetrieb am CAN-Bus. Online resource,
available at: http://www.elektroniknet.de/elektronik-
automotive/sonstiges/system-basis-chip-fuer-den-
teilnetzbetrieb-am-can-bus-85914.html.
NXP, 2018, Automotive products. Online resource,
available at: https://www.nxp.com/products/
automotive-products:MC_50802?tid=FSH.
Schmutzler, C., Krüger, A., Schuster, F., Simons, M., 2010,
Energy efficiency in automotive networks: Assessment
and concepts. In International Conference on High
Performance Computing & Simulation, Caen, France,
2010, pp. 232-240.
Suermann, T., Müller, S., 2014, Power Saving in
Automotive Ethernet. Online resource, available at:
https://www.amaa.de/previous-conferences/amaa-
2014/presentations-2014/AMAA201406PowerSaving
inAutomotiveEthernetNXPMueller_Steffenv1.2.pdf.
Walla, G., Gabriel, D., Barthels, A., Ruf, F., Michel, H.U.,
Herkersdorf, A., 2012, ITE-Sim: A simulator and
power evaluation framework for electric/electronic
architectures. In IEEE Vehicle Power and Propulsion
Conference, Seoul, 2012, pp. 869-874.
Yi, C.H., Jeon, J.W., 2015, Power saving using Partial
Networking in automotive system. In IEEE
International Conference on Information and
Automation, Lijiang, 2015, pp. 148-152.
Structuring of Methods to Estimate Benefits of Partial Networking
581