< Back to previous page

Project

System-level hardware-based design techniques for EM Resilience: a necessity for safe and reliable programmable electronics

With the advent of autonomous vehicles, Smart Cities, Industry 4.0 and many more Internet-of-Things related applications, our future society and lives become highly dependent on high-tech electronics. Unfortunately, all high-tech electronics are sensitive to ElectroMagnetic Interference (EMI), while the increasing electrification of, amongst others, vehicles and machines unavoidably means a much harsher electromagnetic environment. In addition, the continuing miniaturization of electronics and decreasing supply voltages makes new electronic products even more vulnerable to EMI. It is therefore of utmost importance to develop the required knowledge and techniques to assure that safety- or mission-critical systems will not suffer from unacceptable risks when being exposed to both intentional and unintentional EM disturbances. This challenge goes well beyond what is needed for compliance to the EMC Directive for CE certification for normal household applications. While for those applications, one malfunction in every 2-3 years might be perfectly acceptable, safety- or mission-related applications with possibly critical consequences might need a mean-time-between-failure of more than \SI{100}{} or even \SI{10000}{} years! For automotive applications, one even aims at only one dangerous failure in every 1 million years of operation due to the huge number of vehicles on the road.

The aim of the study leading to this PhD manuscript was to create techniques and measures to help achieve resilience to EM disturbances in safety- or mission-critical systems. 'Resilience' as used here means that in case of disturbance, the developed techniques and measures should make the system 'real-time fault-tolerant' for EMI so that the system continues to work as intended in a safe manner. In practice, the study for this PhD manuscript focused on hardware-based techniques and measures to minimize the Bit-Error-Rate (BER) within crucial communication channels. This was done by modifying some commonly used techniques from the discipline of Functional Safety, such as redundancy in combination with majority voting, with the appropriate EMC knowledge to make them much more performant to cope with EMI.  Within Functional Safety, redundancy is mainly used to cope with random failures. However, EMI is a complex phenomenon which has to be seen as a systematic, common cause failure. Indeed, 'systematic' because a given system design in a given digital state will always behave in the same way when a given EM disturbance is applied. 'Common cause' because EMI influences many different components at the same time. A typical redundant system is to have two or more identical sets of hardware and software with the same inputs, and performing the same operations on them. When a malfunction occurs in one of these 'parallel channels', a comparator/voter detects that their outputs no longer agree and triggers appropriate actions to maintain safety. Unfortunately, the malfunctions that EMI creates in identical channels can easily be so similar that the comparator/voter cannot tell that there is a problem at all. In this PhD manuscript, several ways are presented to achieve that the parallel paths in a redundant system exhibit a different behaviour ('EM-diversity') when subjected to the same EMI. 

To validate the performance of the proposed EM-diversity techniques, an efficient simulation framework is used. This simulation framework allows to apply a large variation of EMI disturbances (incoming fields, transient disturbances, ESD, etc.) to (simplified) models of safety-critical systems. The post-processing integrates statistical analysis to check how electromagnetic disturbances affect e.g. the BER. Thanks to this, the effectiveness of different types of diverse redundancy (inversion, spatial, frequency, time, etc.) for various types of EMI can be compared in depth.

The first part of this manuscript introduces two types of harsh Electromagnetic (EM) environments, namely a plane wave environment and reverberation environment. The first type can be compared with an open space environment in real life or an anechoic chamber as an EMC test environment. This type of environment subjects the system-under-test only to a single plane wave at a time. The second type of environment can be compared with a real life environment which has a lot of reflections occurring on e.g. as buildings, cars, humans, etc. In the EMC testing, this is mimicked in a reverberation chamber. This type of environment subjects the system-under-test to many plane waves, coming from many random directions, at the same time.
The experiments that have to be performed to analyse the EM-diversity properties of the proposed techniques and measures are incorporated in an in-house built simulation framework. This simulation framework is optimised for efficiency and applicability. The effect of the two EM environments is modelled by a limited set of full-wave simulations of the geometry under consideration. The results from that simulation are implemented in the framework and the effect of the disturbances is calculated by using an efficiently implemented reciprocity-based technique. In addition, all properties of the encoding and decoding methods for the data which is communicated over the subjected geometry can be modified efficiently. By using sets of random data and varying the parameters within the model using a Monte-Carlo method, statistically relevant metrics are achieved and can be used to compare the effectiveness of the introduced techniques and measures (T\&Ms) to create EM-resilience. The metrics comprises the BER and the number of false negatives or undetectable errors.

In this PhD manuscript, several new hardware EM-diverse T\&Ms are introduced. These T\&Ms are based on several properties of the hardware that can be changed. First, the possibility of matching or not matching the impedances of micro-strips is investigated on redundant and non-redundant geometries. Next, the use of an extra communication channel with inverted data is used to see if it could introduce EM-diverse behaviour. Furthermore, using three micro-strips in different orientations to create spatial diversity is investigated and effectively creates EM-diverse systems. Two final methods which change the timing of the data going over the communication channels is analysed. The transmission start time is changed to create time diversity and the transmission data rate is changed to create frequency diversity. Both methods show that they effectively introduce EM-diverse properties to the system, each with their own specific properties.

In addition, this manuscript studies the use of a matched filter as a possible measure to create EM-resilience. The matched filter is a well-known digital processing technique in receivers to maximise the signal-to-noise ratio. This technique was never before investigated in the light of EM-resilience. Additionally, the matched filter method is compared with a majority voter. It is shown that using a matched filter could even be more effective than using a majority voter under some condition. 
The last part of this manuscript compares the proposed techniques in several ways and concludes which type of diversity to create EM resilience can be used in which situation. The comparison is based on the two main metrics used in this manuscript, namely the BER and the number of false-negatives or undetectable errors when using redundancy.

Date:12 Sep 2016 →  18 Sep 2020
Keywords:Electromagnetic Compatibility, Risk Management
Disciplines:Sensors, biosensors and smart sensors, Other electrical and electronic engineering, Electromagnetism and antenna technology
Project type:PhD project