Live State-of-Health Safety Monitoring for Safety-Critical Automotive Systems

Full text

Live State-of-Health Safety Monitoring for
Safety-Critical Automotive Systems
Andreas Strasser†, Philipp Stelzer†, Christian Steger†and Norbert Druml∗
†Graz University of Technology, Graz, Austria
{strasser, stelzer, steger}@tugraz.at
∗Infineon Technologies Austria AG, Graz, Austria
{norbert.druml}@infineon.com
Abstract—Autonomously driving vehicles require higher safety
and reliability standards than traditional human-driven vehicles
as they need to be able to handle safety-critical situations on
their own. Therefore, these systems needs to demonstrate fail-
operational behavior to ensure safety of the passengers by basic
car controls. Especially silent failures of semiconductor devices
can be critical from a safety point of view. Semiconductor devices
fail abruptly and cannot be detected in advance.
This paper presents a novel sensor approach to detect those
kind of silent failures ahead of time and to ensure safety
for future advanced driver-assistance systems (ADAS) such as
LiDAR (Light Detection and Ranging). We have evaluated the
design of our novel sensor concept in SystemC which will be
implemented in a LiDAR system to mitigate silent failures as
well as enable dynamic safety contracts.
Keywords-Safety, Safety Monitoring, Aging Monitor, Compo-
nent Reliability, Safety Integrated Circuits, Live FIT Estimation
I. INTRODUCTION
Autonomous driving is one of the next big steps of our
society and is the key enabler of Smart Mobility [1]. Smart
Mobility reinvents the urban environment by connecting in-
frastructure, vehicles and people to allow better quality of life,
efficient energy usage and reduced costs for everyone. As a
result, this era will disruptively change the daily routines of
individuals as well as urban life [2]. 50 years ago, the idea
of Smart Mobility started in Germany when Continental, a
leading German automotive manufacturing company, tested
tires on their test track Contidrom. Continental wanted to
ensure constant conditions for testing and developed a self-
driving car for this purpose [3]. This marked the beginning
Fig. 1. PRYSTINE’s concept view of a fail-operational urban surround
perception system [1].
of autonomous driving. Nowadays, self-driving cars have
already made their way to public streets. Tesla was the first
company to release a semi-autonomous driving function called
“Autopilot” [4]. Past accidents showed that it is hard to
ensure safe semi-autonomous driving in urban environments
by traditional methods [5]–[7]. Consequently, new Advanced
Driver-Assistance Systems (ADAS) such as LiDAR (Light
Detection and Ranging) need to be developed and combined
with established systems. This is also the PRYSTINE (Pro-
grammable Systems for Intelligence in Automobiles“ project’s
focus which aims at developing a comprehensive environment
perception system by using LiDAR, radar and vision cameras
as shown in figure 1 [1]. One of the key challenges of
autonomous driving is safety and reliability of before men-
tioned systems. Traditional human-driven vehicles are fully -
or supported by ADAS almost fully - controlled by the driver.
Therefore, the system can return control and responsibility to
the driver in critical situations. In future, vehicles with fully
autonomous driving functionality will not have this possibility
and need to be able to deal with critical situations on their
own. That’s one of the reasons why the impact of safety and
reliability in the automotive domain is steadily increasing [8].
Nowadays, safety-critical automotive systems are developed
in compliance with the ISO 26262 standard. This standard
covers the development of electrical and electronic compo-
nents for the automotive domain with a special focus on safe
hardware and software components [9]. The standard added
a guideline especially for semiconductor devices but does not
support or cover dynamic safety functions such as ”Conserts
M“ or ”Ontology-Based-Run-time-Reconfiguration“. Dynamic
safety functions are necessary to establish resilience and
flexibility to complex cyber-physical systems (CPS) [10].
Especially for future ADAS, such as the fail-operational urban
surround perception system of the PRYSTINE project, this
concept is vital to ensure fail-operational behavior during run-
time.
Fail-operational systems require information about the com-
mon reliability and safe state of each system. Up to now, there
is no possibility to retrieve live information about component
reliability. Usually, components are designed for a specific
utilization profile and safety is dimensioned for this profile.
If there are substantial deviations to this profile, components
could be undersized from a safety point of view [9]. It
102
2019 22nd Euromicro Conference on Digital System Design (DSD)
978-1-7281-2862-7/19/$31.00 ©2019 IEEE
DOI 10.1109/DSD.2019.00025
Authorized licensed use limited to: UNIVERSITA MODENA. Downloaded on January 27,2022 at 08:19:17 UTC from IEEE Xplore. Restrictions apply.

would be beneficial to enable live monitoring of semiconductor
devices’ component reliability to communicate the state-of-
health of individual components.
This paper will address the following research questions:
•Is it possible to detect component reliability of semi-
conductor systems during run-time?
•How can component reliability be measured for semi-
conductor devices?
II. RE LATE D WOR K
In general, detecting safety-related issues of mechanical
components is rather trivial as it often involves vibration or
noise during the operation [11]. For electrical or electronic
components, detecting safety-related issues is much more com-
plex. These systems fail silently and abruptly [9]. Especially
for fully-autonomous vehicles, this fact poses a substantial risk
as these systems need to handle every safety-critical situation
on their own and any failure could trigger fatal road accidents.
If we consider tucks carrying ecologically harmful substances,
accidents may also lead to environmental disasters.
In general, designers of safety-critical semiconductor de-
vices construct and dimension components for specific utiliza-
tion profiles. These profiles cover the worst case utilization of
the component to ensure component reliability during lifetime.
Especially for semiconductor companies that design “Safety
Elements out of Context”, this design philosophy is difficult
as they need to find the best compromise between cost and
reliability. Overdimensioning hardware leads to higher costs,
which may be the decisive factor for making business or not.
Nowadays, every semiconductor device contains additional
safety-related monitoring circuits. For digital circuits, common
monitors are error correction codes (ECC) or Built-In-Self-
Test (BIST), analog circuits use monitors such as the Built-
In-Current Sensor (BICS). These monitors mitigate specific
problems: For instance, ECC control single event upsets
(SEU), BIST checks correct functionality [9]. Shaheen et
al. [12] describe common ECC practices in the automotive
domain such as Parity Bit, Single Error Correction, Single
Error Correction and Detection to detect and correct SEU
during run-time [12]. Sargsyan [13] describes different BIST
technologies that ensure correct functionality of digital semi-
conductor devices such as Production Mode Testing, Power-
on Mode Testing and Mission Mode Testing. These tests are
executed at startup or during idle time and compare the result
with deposited patterns [13]. For analog circuits, Smith et al.
describe the BICS that can detect current leakage [14]. Beckler
et al. [15] introduce the On-Chip Diagnosis for early life and
wear-out failures [15]. All these approaches only focus on
testing the specific circuit’s functionality in a specific moment
and can not give any information on the current state-of-
health. Therefore, it is necessary to have historical data about
the device such as temperature, for instance. Szekely et al.
[16] introduce a sensor for on-line temperature monitoring of
safety-critical Integrated Circuits (IC). However, this sensor
focuses on observing and communicating current temperature
to external systems but does not cover temperature history
[16]. Especially temperature history has a big impact on
component reliability and needs to be considered from a safety
point of view because higher temperature relates to higher
component stress and this negatively influences the reliability.
Component reliability is one of the key requirements for
safety-critical hardware devices. Nowadays, the automotive
industry’s approved safety methods are compiled in the ISO
26262 standard [9]. In general, these methods quantify hard-
ware devices’ component reliability in the failure in time
(FIT) Rate. The FIT Rate represents the amount of failures
that statistically arises within one billion operating hours. The
FIT Rate is calculated or statistically determined by specific
standards such as the IEC TR 62380 [17]. Usually, each
semiconductor manufacturer publishes the specific FIT Rates
for their devices in the component reliability data sheet [18].
These data sheets usually provide the FIT Rate for a specific
test temperature which can be used to calculate equivalent FIT
Rates for specific temperatures using the Arrhenius equation
as seen in (1).
DF =e
Ea
k·(1
Tuse
−1
Tstress )) (1)
where:
DF is Derating Factor
Eais Activation Energy in eV
k is Boltzmann Constant (8.167303 x 10-5 ev/K)
Tuse is Use Junction Temperature in K
Tstress is Stress Junction Temperature in K
The Derating Factor (DF) represents the positive or negative
feedback of the specific temperature on the semiconductor
device and depends on the Junction Temperatures that need
to be determined with equation (2).
Tj=Tamb +Pdis ·θja (2)
where:
Tamb is Ambient Temperature
Pdis is Power Dissipation
θja is Package Thermal Resistance Value
Equation (2) shows that the component reliability depends on
the power dissipation as well as on the ambient temperature
of the integrated circuit. The Derating Factor can be used for
calculating the specific FIT Rate for a specific temperature as
seen in (3).
F IT Base =DF ·F I TDS (3)
where:
DF is Derating Factor as seen in (1)
FITDS is Base FIT Rate of Component Reliability Data
sheet
The idea of Beckler et al. [15] and Szekely et al. [16] with
these equations could be used for live component reliability
monitoring.
Therefore, this paper’s contribution to existing research is:
•Developing a novel method for enabling live safety
monitoring of safety-critical automotive systems.
103
Authorized licensed use limited to: UNIVERSITA MODENA. Downloaded on January 27,2022 at 08:19:17 UTC from IEEE Xplore. Restrictions apply.

•Implementing the novel method in SystemC to prove
feasibility.
•Describing the integration of the novel method in a safety-
critical LiDAR sensor system for autonomous driving.
III. USE CA SE OV ERVIEW
Self-driving vehicles handle safety-critical situations on
their own without any control of a driver. Consequently, a
high safety and reliability standard is necessary to ensure
fail-operational behavior. In the next few years LiDAR will
become common in middle-class cars and will be an im-
portant part of self-driving functionality [19]. LiDAR is an
environment perception systems in combination with Radar
and Vision [1].
The 1D MEMS LiDAR system of Druml et al. [19] is
a novel approach to develop an inexpensive ADAS that is
suitable for the mass. Novel technologies are always related
to unknown failures [11]. Especially in the domain of self-
driving cars, these failures are not tolerable because they result
in severe road accidents.
To increase the learning curve and to evolve safer and
more reliable LiDAR systems as fast as possible, component
reliability should be monitored live to get real-time data of a
single vehicle as well as of a complete fleet. This will enable
functions that increase the overall safety level of an individual
driver as well as the overall road safety. Both scenarios will
be described in our use case that is divided into two sections:
•Live Reliability Data for Customers
•Live Reliability Data for Original Equipment Manufac-
turer (OEM)/Suppliers
A. Live Reliability Data for Customers
The reliability data of a single vehicle can be used to
determine the overall usage level of a specific system as well as
of the complete car. This could be used for enabling predictive
maintenance like in the aircraft industry. If a specific FIT Rate
is reached and the safety-critical device is dropping in the
Automotive Safety Integrity Level (ASIL), this could trigger
the replacement of the specific device. Especially for self-
driving cars this approach could ensure a specific safety level
of all self-driving road vehicles.
Another use case is the review of the complete car if
individual maintenance repairs are worth to accomplish. If a
certain amount of systems has reached a specific FIT Rate, this
would suggest that these systems will also fail in the next few
months. This will support the customer during his decision, if
a repair is useful or not.
B. Live Reliability Data for OEM/Supplier
For OEMs and suppliers, the reliability data is valuable to
understand whether the systems are designed for their use
cases and whether there are any problems that could arise
during warranty time. By using real-time data, suppliers can
interfere to adapt the software parts of the devices to ensure
a specific FIT Rate until the end of lifetime.
Especially software updates are changing the behavior of
Self-Driving Car
Self-Driving Car
OEM Backend
Self-Driving Car #1
LiDAR System
Communication
Server C2I
Radar System
Signal Processing
Vision Camera
Car 2 Infrastructure
Analyzing Reliability
LiDAR
Status
SP
Status
Radar
Status
Vision
Status
Fig. 2. Use case overview of the live FIT Monitor for safety-critical LiDAR
sensor systems.
devices and may have a big impact on the overall safety level.
By collecting reliability data of these live monitors, it will be
possible to investigate and evaluate changes of these updates
from a safety point of view.
IV. RET ROFIT - LIVE SAFETY MONITORING SEN SO R
DF =e
Ea
k·(1
Tuse
−1
Tamb+Pdis ·θja )) (4)
By combining both equations of the Related Work on com-
ponent reliability, it becomes obvious that it is possible to
calculate the theoretical FIT Rate for a specific temperature
as seen in (3). However, component temperature is changing
over time which results in different FIT Rates. Therefore,
considering these temperature profiles as a time slice in a
whole mission profile [9] will be used and integrated in our
novel approach of live safety monitoring.
The idea behind our novel approach is to sample the power
dissipation and the actual case temperature at a specific time
interval. The power dissipation measurements are averaged
and saved in a register which represents the average power
dissipation of the whole lifetime. The temperature values are
classified in a specific temperature range and integrated in a
histogram. This histogram represents the whole temperature
history of the integrated circuit during lifetime and can be
used for further component reliability computations.
For calculating the FIT Rate at a specific time, the following
steps are necessary:
1) Calculate FIT Rate for each Histogram Bin
2) Determine the time span percentage of each Histogram
Bin
3) Calculate the FIT Rate for each Histogram Bin
104
Authorized licensed use limited to: UNIVERSITA MODENA. Downloaded on January 27,2022 at 08:19:17 UTC from IEEE Xplore. Restrictions apply.

4) Sum up each individual Bin FIT Rates to the overall FIT
Rate
5) Determine and check with theoretical lifetime FIT Rate
A. Calculate FIT Rate for each Histogram Bin
Each Histogram Bin represents a specific temperature. In
our case, we chose a temperature range between 0◦C and
140◦C. For each Bin, the specific FIT Rate can be calculated
by using equation (3) and (4). These FIT Rates represent the
statistical lifetime FIT Rate assuming this device would run
on this specific temperature for the whole lifetime.
B. Determine the time span percentage of each Histogram Bin
As a first step, the run-time of the device until this moment
is determined. For this purpose, all samples of the whole
Histogram are summed up as seen in (5).
TOR =Pn·TSR
3600 (5)
where:
TSR is sampling rate of the measurements.
The overall run-time can be used to determine the specific
amount of run-time for each Histogram Bin as seen in (6).
TRun =TSR
3600 ·TOR
·n(6)
where:
TSR is sampling rate of the measurements.
TOR is the whole run-time of the device as calculated in
(5).
The equation (6) is used to calculate the run-time for each His-
togram Bin. In the next step, the specific FIT Rate considering
the specific run-time is calculated.
C. Calculate the FIT Rate for each Histogram Bin
In the next step, the FIT Rate of the whole lifetime of each
Histogram Bin is calculated.
F IT Bin =F ITRB
TEL
·TRun (7)
where:
FITRB is FIT Rate of the specific temperature of the Bin as
calculated in (4).
TRun is the whole run-time of the device as calculated in
(5).
TEL is the expected lifetime of the semiconductor device
that has been selected during design phase.
D. Sum up each individual Bin FIT Rates to the overall FIT
Rate
In the last step, all individual FIT Rates of the Bins are
summed up to an overall FIT Rate.
F IT TS =XF IT Bin (8)
where:
FITBin is FIT Rate of each Bin as calculated in (7).
This value represents the FIT Rate to this specific timestamp
and can be compared to the theoretical FIT Rate up to this
timestamp as well as the theoretical FIT Rate until the end of
the expected lifetime.
E. Determine and check with theoretical lifetime FIT Rate
In the last step, we observe if the FIT Rate of the current
timestamp exceeds the theoretical FIT Rate that was chosen
during design phase.
F IT TT S =F I TDS ·TOR
TEL
(9)
where:
FITDS is theoretical FIT Rate for a specific temperature as
seen in (4).
TOR is run-time of the device until this timestamp as seen
in (6).
TEL is the expected lifetime of the semiconductor device
that has been selected during design phase.
The ratio between the theoretical FIT Rate and the calculated
FIT Rate gives a tendency about the usage of the device and
whether there should be any concern due to predicted over-
stress until the end of the lifetime.
F IT Ratio =F ITT S
F IT TT S
(10)
Ratios that are greater than one indicate that the device was
used too extensively and that there could be over-stress until
the end of the expected lifetime. This also increased the
theoretical amount of failures until the end of the lifetime. The
amount of statistical failures can be determined with equation
(11).
F IT LT =F ITT S ·TEL
TOR
(11)
where:
FITTS is the calculated FIT Rate for a specific timestamp
as seen in (8).
TOR is run-time of the device until this timestamp as seen
in (6).
TEL is the expected Lifetime of the semiconductor device
that has been selected during design phase.
V. RESU LTS
We will implement the “RetroFIT” method in a LiDAR
system as seen in Figure 3. To evaluate the functionality and
behavior of this methodology we implemented this approach
in SystemC.
In Figure 4 the architecture of the implemented FIT Monitor
can be seen. The architecture consists of the “Environmental
and Integrated Circuit Simulation Model” that contains the
temperature profile (as seen in Figure 5) curve that will
stimulate the FIT Monitor. The histogram will save each
sampled value of the temperature as well as the average power
dissipation. The last part is the signal processing where the FIT
Rates are calculated as described in Section IV. In Figure 5 the
upper diagram is showing temperature profile that have been
105
Authorized licensed use limited to: UNIVERSITA MODENA. Downloaded on January 27,2022 at 08:19:17 UTC from IEEE Xplore. Restrictions apply.

System
Safety
Controller
(AURIX)
Laser Illumination
MEMS Mirror
MEMS
Driver
ASIC
Sensing
Reflected
Signal
Photo Diodes
dt
Emitted
Signal Point
Cloud
Data
Trigger and
Laser Powe r Setting
FPGA / Dedic ated
LiDAR Ha rdware
Accelerator s
Receiver
Circuits
Raw Data
Emitter Path
Receiver Path
Trigger and
Gain Settin g
Config
and
Status
FIT Mon.
FIT Mon.
FIT Mon.
FIT Mon.
FIT Mon.
FIT Mon.
Actuation
Component
Reliability
Report
Fig. 3. Live FIT Monitor integration into the safety-critical LiDAR sensor
system to enable live safety monitoring [19].
used for our simulation. The lower diagram shows the specific
temperature values for each sampling point. In our simulation
we have sampled with a frequency of 0.05 Hertz. The related
Histogram of our simulation can be seen in Figure 6. Each
Histogram Bin represents a 1◦C and is distributed on the x-
Axis. The amount of samples can be read out on the y-Axis.
In our simulation the most samples could be found between
100◦C and 110◦C. Compared with the temperature profile of
Figure 5 this looks plausible.
TABLE I
FIT RES ULTS O F OUR SY ST EMCMODEL SIMULATION WITH
TEM PE RATU RE PRO FIL E INP UT A S SEE N IN FIGURE6.
FITTS FITTTS FITLT FITRB FITRatio
FIT
in [1] 2.36E-9 2.111E-9 8.5 7.6 1.118
FIT Monitor
Test Class
Environment & Integrated Circuit
Simulation Model
Temperature
Sensor
Power
Dissipation
Sensor
FIT Signal Processing
Histogram
Trigger
Sensor
Values
Read
Values
FIT
Data
Create
Histogram
Calculate
FIT / BIN
Add Up
FIT
Memory
Estimate
Lifetime FIT
Fig. 4. SystemC model overview of the “RetroFIT” methodology to enable
live safety monitoring for safety-critical LiDAR sensor systems.
0 100 200 300 400 500 600 700 800 900 1000
Time [s]
0
20
40
60
80
100
120
140
Temperature [°C]
Temperature Profile
0
20
40
60
80
100
120
140
Temperature [°C]
Sampling 20s
0 100 200 300 400 500 600 700 800 900 1000
Time [s]
Fig. 5. Measurement results of the “RetroFIT” monitor.
Histogram 20s
0 20 40 60 80 100 120 140
Temperature [°C]
0
1
2
3
4
5
6
Frequency [1]
Fig. 6. Histogram results of the “RetroFIT” monitor.
In Table I the FIT results of our SystemC simulation can be
seen. The device has an FITRB value of about 7.6 in 1 Billion
operating hours at 100◦C. The provided temperature profile,
as seen in 5, over-stresses the component and this results in
a higher FITLT of about 8.5. As a result the device has been
over-stressed by 11.8%. Consequently, a continuously operated
device with this temperature profile would result in a higher
FIT Rate than from the designer of the device expected.
VI. SUMMARY
In Section IV, this paper introduces the novel “RetroFIT”
sensor to support live safety monitoring of electrical and
electronic devices. Nowadays, electronic components such as
sensors and micro-controllers fail without any prior indication.
Especially for fully automated driving, this circumstance may
cause disastrous consequences such as deadly accidents. For
future autonomous driving vehicles, our novel method can
communicate the actual component reliability.
To give an overview about the application of our novel
sensor, we have introduced two common use cases from the
customer point of view as well as from the OEM/Supplier
point of view. Both cases show that “RetroFIT” has a big
impact on the overall road safety as the sensor may for instance
trigger component replacement. The values could be obtained
106
Authorized licensed use limited to: UNIVERSITA MODENA. Downloaded on January 27,2022 at 08:19:17 UTC from IEEE Xplore. Restrictions apply.

by qualified car repair shops as well as displayed inside the
driver’s cabin including service deactivation.
In section V, we prove that the sensor concept is feasible
and that it is possible to live monitor component reliability for
electronic devices.
Fail-operational systems become increasingly essential. Our
novel “RetroFIT” sensor enables dynamically changing con-
tracts during run-time. This concept is one of the key enablers
of advanced fail-operational systems. Our sensor enables the
communication of the actual ASIL level of components and
communicates these values to other systems. This will detect
ASIL degradation during run-time and trigger safety related
functions to increase the overall system safety.
ACKNOWLEDGMENTS
The authors would like to thank all national funding au-
thorities and the ECSEL Joint Undertaking, which funded the
PRYSTINE project under the grant agreement number 783190.
PRYSTINE is funded by the Austrian Federal Ministry
of Transport, Innovation and Technology (BMVIT) un-
der the program ”ICT of the Future” between May 2018
and April 2021 (grant number 865310). More information:
https://iktderzukunft.at/en/.
REFERENCES
[1] N. Druml, G. Macher, M. Stolz, E. Armengaud, D. Watzenig, C. Steger,
T. Herndl, A. Eckel, A. Ryabokon, A. Hoess, S. Kumar, G. Dim-
itrakopoulos, and H. Roedig, “Prystine - programmable systems for
intelligence in automobiles,” in 2018 21st Euromicro Conference on
Digital System Design (DSD), Aug 2018, pp. 618–626.
[2] R. Faria, L. Brito, K. Baras, and J. Silva, “Smart mobility: A survey,”
in 2017 International Conference on Internet of Things for the Global
Community (IoTGC), July 2017, pp. 1–8.
[3] Mmpro. [Online]. Available: https://publicarea.admiralcloud.com/p/
a49d3d8ba92f3cdbfa864f
[4] M. Dikmen and C. Burns, “Trust in autonomous vehicles: The case of
tesla autopilot and summon,” in 2017 IEEE International Conference on
Systems, Man, and Cybernetics (SMC), Oct 2017, pp. 1093–1098.
[5] “Accident investigations.” [Online]. Available: https://www.ntsb.gov/
investigations/Pages/HWY18FH011.aspx
[6] F. Lambert, Fred, and Electrek, May 2018. [Online]. Available:
https://electrek.co/2018/05/09/tesla-model- s-fatal- crash-fire- national/
[7] S. Levin, “Tesla fatal crash: ’autopilot’ mode sped up
car before driver killed, report finds,” Jun 2018. [On-
line]. Available: https://www.theguardian.com/technology/2018/jun/07/
tesla-fatal- crash-silicon- valley-autopilot-mode- report
[8] R. Mariani, “An overview of autonomous vehicles safety,” in 2018 IEEE
International Reliability Physics Symposium (IRPS), March 2018, pp.
6A.1–1–6A.1–6.
[9] I. n. E. ISO, “Draft 26262 2nd Edition: Road vehicles-Functional safety,”
International Standard ISO/FDIS, vol. 26262, 2018.
[10] T. Amorim, D. Ratasich, G. Macher, A. Ruiz, D. Schneider, M. Driussi,
and R. Grosu, “Runtime safety assurance for adaptive cyber-physical
systems: Conserts m and ontology-based runtime reconfiguration applied
to an automotive case study,” in Solutions for Cyber-Physical Systems
Ubiquity. IGI Global, 2018, pp. 137–168.
[11] N. Leveson, Engineering a safer world: Systems thinking applied to
safety. MIT press, 2011.
[12] H. Shaheen, G. Boschi, G. Harutyunyan, and Y. Zorian, “Advanced
ECC solution for automotive SoCs,” in 2017 IEEE 23rd International
Symposium on On-Line Testing and Robust System Design (IOLTS), July
2017, pp. 71–73.
[13] D. Sargsyan, “Iso 26262 compliant memory bist architecture,” in 2017
Computer Science and Information Technologies (CSIT), Sept 2017, pp.
78–82.
[14] P. A. Smith and D. V. Campbell, “A practical implementation of bics
for safety-critical applications,” in Proceedings 2000 IEEE International
Workshop on Defect Based Testing (Cat. No.PR00637), April 2000, pp.
51–56.
[15] M. Beckler and R. D. Blanton, “On-chip diagnosis for early-life and
wear-out failures,” in 2012 IEEE International Test Conference, Nov
2012, pp. 1–10.
[16] V. Szekely, M. Rencz, J. M. Karam, M. Lubaszewski, and B. Courtois,
“Thermal monitoring of safety-critical integrated systems,” in Proceed-
ings of the Fifth Asian Test Symposium (ATS’96), Nov 1996, pp. 282–
288.
[17] T. IEC, “62380,” Reliability data handbook–universal model for reliabil-
ity prediction of electronics components, PCBs and equipment (emerged
from UTEC 80-810 or RDF 2000), 2004.
[18] “Reliability report,” Jul 2018. [Online]. Avail-
able: https://www.intel.com/content/www/us/en/programmable/support/
quality-and- reliability/reports-tools/reliability- report/rel-report.html
[19] N. Druml, I. Maksymova, T. Thurner, D. Van Lierop, M. Hennecke, and
A. Foroutan, “1D MEMS Micro-Scanning LiDAR,” in Conference on
Sensor Device Technologies and Applications (SENSORDEVICES), 09
2018.
107
Authorized licensed use limited to: UNIVERSITA MODENA. Downloaded on January 27,2022 at 08:19:17 UTC from IEEE Xplore. Restrictions apply.