Method of analyzing a design of an electronic device, system, computer program product, and computer-readable storage medium

Abstract

The present disclosure relates to a method (100) of analyzing a design of an electronic device. The method (100) comprises the following steps: obtaining (101), by a data processing device, a design file of the electronic device, wherein the design file comprises a design of the electronic device; determining (102), by the data processing device, electromagnetic compatibility, EMC, information of the electronic device based on the design of the electronic device of the obtained design file, wherein the determining of EMC information comprises: obtaining (103), by the data processing device, component information regarding at least one component comprised in the electronic device based on the design of the electronic device, and determining (104), by the data processing device, functional information based on the component information, wherein the functional information is determined using a graph neural network, GNN. The method comprises further: analyzing (106), by the data processing device, the EMC information to determine whether at least one EMC issue is detected in the EMC information, wherein the analyzing the EMC information comprises analyzing the functional information.

Description

[0001]P2024,0845 WO N August 25, 2025 - 1 - Description Method of analyzing a design of an electronic device, system, computer program product, and computer-readable storage medium This disclosure relates to a method of analyzing a design of an electronic device. The disclosure further relates to a corresponding system, a corresponding computer program product, and a corresponding computer-readable storage medium. Electronic devices of all kinds play an important role in most fields of our lives nowadays. In the context of this application, electronic devices are entire systems such as, for example, computer systems, monitors, televisions, etc., as well as parts of such entire systems such as, for example, printed circuit boards, PCBs, wirings of entire systems, casings, etc. An important aspect of both the entire systems and the parts, i.e., of all electronic devices considered in the context of this application, is their electromagnetic compatibility, EMC. The EMC has a significant impact on the electronic device’s functionality as well as it plays an important role for the electronic device to pass EMC testing regarding quality control. During the development process of an electronic device, firstly, a design of such device is drafted. Then, a prototype according to the drafted design is created. Afterwards, numerous tests for measuring the device’s qualities including the EMC are conducted. Based on tests results of those tests, the design is adapted to try to improve the qualities of the device. An adapted prototype is P2024,0845 WO N August 25, 2025 - 2 - created and tested again. This development process is commonly carried out in the form of a try-and-error approach, wherein changes to the design are implemented in order to achieve better results in the tests. To improve efficiency and simplify the design process of an electronic device, computerized methods to analyze a design of an electronic device are being developed. However, the information a computerized method may rely on during such analysis may be limited and, therefore, a result of a computerized analysis is often significantly less precise than the above-described cumbersome testing cycle. Hence, it is an object of this disclosure to present a method, system, computer program product and computer-readable storage medium with which at least a part of the development process of an electronic device may be improved. The above-mentioned object is achieved by the subject-matter of the independent claims. Further advantageous embodiments are disclosed in the dependent claims, as well as the attached figures and the following description. According to one aspect, a method of analyzing a design of an electronic device comprises the following steps: obtaining, by a data processing device, a design file of the electronic device, wherein the design file comprises a design of the electronic device, and determining, by the data processing device, electromagnetic compatibility, EMC, information of the electronic device based on the design of the electronic device of the obtained design file. The determining of EMC information comprises: obtaining, by the data processing device, component information regarding at least one component comprised in the electronic device based on the P2024,0845 WO N August 25, 2025 - 3 - design of the electronic device, and determining, by the data processing device, functional information based on the component information, wherein the functional information is determined using a graph neural network, GNN. The method further comprises: analyzing, by the data processing device, the EMC information to determine whether at least one EMC issue is detected in the EMC information, wherein the analyzing the EMC information comprises analyzing the functional information. An advantage of the above method is that EMC issues may be determined during the process of designing the electronic device with more precision and, hence, the result of the analysis is more reliable. This advantage is achieved in that not only the explicit information concerning components explicitly mentioned in the design file are considered, but from the explicit component information, additional functional information is derived. With other words, the information obtained from the design file directly is “enriched”. An advantage of using a GNN for determining the functional information based on the component information is that the structure of a GNN translates particularly well to complex structures of a design of an electronic device. A GNN is an artificial neural network for processing data that can be represented as graphs. Each connection between two components may be determined based on an adjacency matrix, wherein the connection between two components are represented by edges between nodes, the nodes being the components itself. To describe the nodes in the graph, i.e., to describe the components in the design, each node may have a certain number P2024,0845 WO N August 25, 2025 - 4 - of features, wherein the features of each node are represented in a feature matrix. A GNN handles graph data by performing the steps of message passing, aggregation, and updating on each node in the graph at each layer of the GNN. Given a graph, node embedding may be computed by using a graph with node features G = (V, E, X), where V and E are a set of nodes and edges, and xV∈ ℝdis a node feature for node v ∈ V. An output may be a graph- aware node embedding zV∈ ℝdout. Once zVis obtained, the nodes may be classified by applying Multi Perceptron Layer, MLP, for each zVindependently. GNNs initialize the node state by node features (0)← ^V∈ ℝd∀, v ∈ ^and then updates the state by aggregating information from the neighboring nodes hv(l+1)← ^^^^^,(^)(hv’(l), {hu(l)|u ∈ ^(v)}), where ^(^) is the set of neighboring nodes to v. ^^^^^,(^)is the aggregation function modelled by neural networks. This mechanism is called message passing. After initialization, the features are aggregated and transformed. The aggregation function is shared for all nodes. This may be repeated and thereby stacked corresponding to multi-layered neural networks. The GNN used in the context of this application may be trained based on training data, e.g., designs of corresponding electronic devices in which the to be detected functional information is marked for training purposes. This way, a GNN can be trained to detect the functional information from the explicit component information obtained from the design like in commonly known AI model training. P2024,0845 WO N August 25, 2025 - 5 - The component information in the context of this application is information regarding components that are comprised in the electronic device and that is directly obtained from the design file. For example, this component information is obtained from a component list of the electronic device. It is to be noted that, while the use of a GNN achieves the specific additional advantages discussed above, and while this application mainly focuses of the use of a GNN in this context, any other artificial neural network may be used as well for the determination of the functional information. Hence, the scope of this disclosure covers the method according to the first aspect mentioned above with the use of any other artificial neural network instead of a GNN as well. This applies accordingly to the other aspects, i.e., the system, computer program product, and computer-readable storage medium. The functional information is information that is determined based on the explicit component information, but that is not explicitly comprised in the design file itself. The functional information is information regarding aspects that impact the EMC performance. For example, the functional information comprises a filter information regarding the at least one component of the electronic device, and / or a power converter information regarding the at least one component of the electronic device. With other words, the functional information in those examples comprises information that the at least one component exhibits a filter and / or a power converter functionality in the design of the electronic device. The functional information signifies that a functional element, such as a filter or a power converter, is present in the design, even if the design itself does not P2024,0845 WO N August 25, 2025 - 6 - explicitly mention or show such element. The functional information may be, for example, in the form of a number of corresponding functional elements being detected in the design of the electronic device and / or in the form of identifying for components (some or all) comprised in the design of the electronic device whether they are part of such functional element or not. The functional information in the context of this application may relate, for example, to entire power converter circuits, such as buck converters, boost converters, flyback converters, etc., but may also relate, for example, to subcircuits, such as rectifier circuits, switching circuits, filters, etc.). The same as discussed in this application with regard to functional information relating to filters or power converters may further equally be applied to other aspects, such as amplifier circuits, active EMC filters, or zoning detection of functional areas, such as analog areas, digital areas, I / O areas, power areas, radio frequency areas, etc. The detection of such functional areas, identifying areas of the electronic device, e.g. a PCB, dedicated to a type of circuit or function may be advantageous. These areas are called zones and the data concerning those zones can be enriched by identifying the functional information concerning those zones. EMC is an important topic that engineers designing electronic devices must consider. EMC refers to the ability of electronic devices and systems to function properly in the presence of electromagnetic disturbances and the ability not to generate disturbances to others. EMC compliance may be a requirement for systems in a variety of applications including telecommunications, medical devices, computers, P2024,0845 WO N August 25, 2025 - 7 - automotive industries, aerospace industries, space industries, and consumer electronics, among others. Aside that EMC compliance is an important task while bringing new products to market as it is strongly and tightly regulated by legislation and authorities. EMC compliance can affect a final product's performance, as well as its liability. The design file may be any file comprising information regarding the design of the electronic device from which EMC information may be determined. For example in case of a PCB, the design of the PCB comprises information regarding PCB layers, drill holes, vias, tracks, wires, pads, mounted and / or mountable components, insulations, connectors, etc. For example in case of the electronic device being a wiring of a system (e.g. a computer system, a telecommunication device, a monitor, or any other), the design of such electronic device comprises information regarding a type of used wires, a shielding of the wires, a length of the wires, s number of poles of each wire, a material of each wire, a signal carried by each wire, etc. In particular, the design file may be a file of any of the following: - a set of all gerber files, NC drills, pick and place, optionally with a Bill of Materials (BOM); - ODB++ files; - Native CAD data - a schematic file, for example a PDF. Of the above-mentioned file types, ODB++ files may be advantageously used since they contain EMC relevant information and the information comprised in the files has a standardized structure, irrespective of, for example, a P2024,0845 WO N August 25, 2025 - 8 - vendor of the electronic device, which simplifies the obtaining of the relevant information from said files. Also, native CAD files may be advantageously used since they contain the most extensive information of the above. In general, the design may comprise any information of one or more of the following: a layout of the electronic device, schematics of the electronic device, a bill of material, BOM, component characteristics, and cable harness information. Hence, for determining the EMC information based on the design of the electronic device of the obtained design file, any model information in the design stage may be used to determine the EMC information: - schematics; - Bill of Material (BOM); - component datasheets; - CAD file of a housing and a system in which the electronic device is to be used; - measurement results from prototypes; - cable harness information. From those mentioned examples, the component information may be obtained, based on which the functional information may then be determined using the GNN. EMC information in the context of this application describes information based on which it may be determined whether and / or how strong electromagnetic radiation is emitted and / or absorbed by elements of the PCB. The EMC information comprises, for example, information regarding any of the following: a grounding of elements of the PCB, a shielding of elements of the PCB, a decoupling of elements of the PCB, a P2024,0845 WO N August 25, 2025 - 9 - filtering of elements of the PCB, a bypassing of elements of the PCB. The EMC information comprises any information derivable from the design of the PCB, which has an influence on the EMC of the PCB as well as the overall system, in which the PCB is to be used. For determining whether an EMC issue is detected, it may be determined whether the EMC information comprises any indication of an element of the design, whose contribution to EMC may be improved by changing characteristics of the element. This may be done using both the component information, obtained directly from the design file, and the functional information, determined using the GNN. For example, the design of the electronic device may not comprise explicit information regarding functional elements such as, for example, filters or power converters or other functional elements, but those elements may nevertheless be present in the design of the electronic device, being made up by explicit components that are shown in the component information, for example by resistors, capacitors, diodes, etc. For example, it may be determined whether an element’s EMC contribution may be changed by further grounding, shielding, decoupling, filtering, bypassing, adding additional components, removing components, changing the properties and / or characteristics of components, such as a capacitance or a type of element (types of elements being, for example, different capacitors: tantalum / ceramic / aluminum / film), and / or changing the relative position of elements to other elements in order to improve the overall EMC of the electronic device’s design. This list, however, is not to be understood to limit the understanding of the determining whether an EMC P2024,0845 WO N August 25, 2025 - 10 - issue is detected. The above may be done either with regard to the component information, i.e., whether the properties of an explicit component may be changed, or with regard to the functional information, i.e., whether the properties of a functional element such as a filter or power converter may be changed e.g., by changing certain explicit components belonging to said functional element, or with regard to both. The analysis may be done, for example, by checking for potential EMC issues in the EMC information. The outcome of the analysis may be that one or more than one EMC issue is detected in the EMC information, or that no EMC issue is detected. Optionally, after the analyzation of the EMC information, feedback information may be determined based on a result of the analyzing of the EMC information and a corresponding feedback may be output and then be used to improve the design of the electronic device directly, without creating and testing a prototype. The feedback information in the context of this application may comprise information indicating how well EMC requirements are met by the presented design, and / or the feedback information may be based on how many EMC issues were found and / or how critical regarding EMC the EMC issues that were found are. Additionally or alternatively, the feedback information may comprise direct information on what improvements with regard to EMC may be achieved in the design. With other words, the feedback information may comprise information regarding changes of the design with which the EMC of the design may be improved. Additionally or alternatively, the feedback information may comprise information concerning an EMC certification test, e.g. whether an EMC certification test is expected to be passed or P2024,0845 WO N August 25, 2025 - 11 - failed by the design, optionally with an indication of a probability, with which the EMC certification test is expected to be passed or failed. For example, it may be indicated in the feedback information, that a certain EMC certification test is expected, with a probability of 80%, to be passed. In any case, the feedback information may be used by the user or by a data processing device to improve the design with regard to EMC. The data processing device, on which the method is implemented, may be any computing device such as, for example, a local computing device, a server, a cloud computing device, etc. In at least one embodiment, the component information comprises at least one of the following: - an identifier of the at least one component; - a component type of the at least one component; - a number of pins of the at least one component; - a grounding state of the at least one component; and - a connection information of the at least one component. The component information mentioned in this context relates to the components themselves and, considering components as nodes in a graph, those pieces of component information may also be described as node level features. An advantage thereof is that such information can easily and efficiently be obtained from the design of the electronic device, thereby allowing an efficient analyzation of the design. P2024,0845 WO N August 25, 2025 - 12 - According to at least one embodiment, the determining of EMC information further comprises obtaining, by the data processing device, edge information regarding the at least one component comprised in the electronic device based on the design of the electronic device, wherein the edge information comprises information regarding a connection between the at least one component and at least one further component; and the determining, by the data processing device, functional information is additionally performed based on the edge information. An advantage thereof is that such edge information may particularly well be processed using a GNN and, therefore, using such edge information for determining the EMC information further improves accuracy of the analyzation of the design. The at least one further component relates to at least one further component that is comprised in the design of the electronic device, whose component information is obtained from the design file. The edge information relates to connections between two components or to connections between a component and a net of further components, to which the component is connected via at least one further component. Accordingly, those pieces of edge information may also be described as edge level features. For example, according to at least one embodiment, the edge information comprises at least one or any combination of two or more of the following: P2024,0845 WO N August 25, 2025 - 13 - - a trace length of a connection between the at least one component and at least one further component; - a trace width of a connection between the at least one component and at least one further component; - a trace type of a connection between the at least one component and at least one further component; - a trace inductance of a connection between the at least one component and at least one further component; - a net ID of a net comprising the at least one component and at least one further component; and - an edge type of a connection between the at least one component and at least one further component, wherein the edge type defines whether a connection between the at least one component and an individual at least one further component is considered or whether a connection between the at least one component and a net of multiple further components via the at least one further component is considered. Different versions of combining different pieces of the above mentioned component information and edge information may be used, which will be described in more detail below, with reference to the figures. According to at least one embodiment, the GNN for determining the functional information is at least one of the following: - a graph convolutional network, GCN; - a graph attention network, GAT; and - a graph attention network version 2, GATv2. A GCN is a specific type of GNN that uses convolutional operations to propagate information between nodes in a graph. GCNs leverage a localized aggregation of neighboring node P2024,0845 WO N August 25, 2025 - 14 - features to update the representations of the nodes. An advantage of using a GCN is that it provides an efficient and easy to implement solution for the GNN. A GAT is a specific type of GNN that incorporates attention mechanisms to capture important relationships between nodes in a graph. A GAT assigns attention weights to the neighboring nodes of a target node, allowing the network to focus on the most relevant neighbors during the information propagation process. An advantage is that with a GAT, a performance of the GNN is improved and false positives in determining the functional information are reduced. A GATv2 concatenates both edge and node features and then applies GAT's attention mechanism. An advantage is that with GATv2, a performance of the GNN is further improved and false positives in determining the functional information are further reduced. According to at least one embodiment, the component information of each of the at least one component and / or the edge information of each of the at least one component comprises at least one categorical value and / or at least one numerical value, wherein: - the at least one categorical value belonging to each of the at least one component is processed using one-hot encoding to determine an affiliation of the respective component to a component group; and / or - the at least one numerical value belonging to each of the at least one component is processed using a normalization preprocessing technique to normalize the at least one numerical value of the respective component. P2024,0845 WO N August 25, 2025 - 15 - An advantage of using one-hot encoding for categorical values is that an easy to implement and efficient way of categorizing the components is provided. It creates an additional feature for each group of the categorical feature and mark each observation belonging (Value=1) or not (Value=0) to that group. Therewith, the information obtained directly from the design file may further be categorized, e.g. in a feature matrix, and the determining of the functional information may thereby further be increased with regard to its accuracy. An advantage of using normalization preprocessing technique for the numerical values is that thereby numerical features in a dataset are transformed to ensure they are on a similar scale. Normalization is used to rescale numerical features in a dataset to have a mean of 0 and a standard deviation of 1. This may further improve accuracy of the output of the GNN, i.e., of the determination of functional information. According to at least one embodiment, the functional information is divided into at least two classes and, for the analyzing the EMC information, at least one of the classes is weighted using a Weighted Random Sampler. An advantage thereof is that an imbalance between classes in the data may be overlooked therewith and, therefore, accuracy of the output of the GNN and therefore of the determination of functional information may be improved. For example, in the case of functional information being “filters”, classes may be, for example, RC lowpass filter, LC lowpass filter, high pass filter, and non-filter. In such case, commonly, the class of “non-filters” could be expected to be significantly higher, as a majority of components in the design would not P2024,0845 WO N August 25, 2025 - 16 - relate to the functional information “filter”. This may create an imbalance, which may be resolve with the WeightedRandomSampler. The WeightedRandomSampler allows to specify weights for each class, which can be used to oversample the minority classes or under sample the majority classes. The method may be a computer implemented method. According to another aspect, a system for analyzing a design of an electronic device is disclosed, the system comprising a data processing device, wherein the data processing device is configured to: - obtain a design file of the electronic device, wherein the design file comprises a design of the electronic device; - determine electromagnetic compatibility, EMC, information of the electronic device based on the design of the electronic device of the obtained design file, wherein to determine the EMC information, the data processing device is configured to: - obtain component information regarding at least one component comprised in the electronic device based on the design of the electronic device; - determine functional information based on the component information, wherein the functional information is determined using a graph neural network, GNN; - analyze the EMC information to determine whether at least one EMC issue is detected in the EMC information, wherein the analyzing the EMC information comprises analyzing the functional information. Advantages and embodiments of the system correspond, in general, to those discussed above regarding the corresponding P2024,0845 WO N August 25, 2025 - 17 - method. For the sake of a concise description, the above discussed advantages and embodiments are not repeated herein, however may equally apply to the system as well. According to yet another aspect, a computer program product comprises instructions which, when executed on a computing device, implements the above-discussed method or any embodiment(s) thereof. According to yet another aspect, a computer-readable storage medium comprises the above-mentioned computer program product. Advantages and embodiments of the computer program product and the computer-readable storage medium correspond, in general, to those discussed above regarding the corresponding method and system. For the sake of a concise description, the above discussed advantages and embodiments are not repeated herein, however may equally apply to the computer program product and the computer-readable storage medium. A basic concept of analysis of an electronic device is disclosed in the EP patent application EP24151345.6, which is hereby incorporated by reference. The aspects disclosed in this application may be combined with any aspect of the analysis disclosed in said EP patent application. Further advantageous embodiments are disclosed in the attached figures, the description thereof, and the attached claims. In the figures, entities with in general the same function may be indicated having the same reference number. Those entities having the same reference numbers, however, P2024,0845 WO N August 25, 2025 - 18 - may slightly differ from each other and do not have to be identical in all detail. Figure 1 shows a method of analyzing a design of an electronic device according to one embodiment of the application; Figure 2 shows a system for analyzing a design of an electronic device according to one embodiment of the application; Figure 3 shows an extract of a design of a PCB with an example of functional information and component information according to this application; Figures 4A, 4B show symbolic visualizations of two examples of graph-based representations of components of an electronic device according to an embodiment of the application; Figures 5A, 5B show an example of a design of a PCB and a symbolic visualization of a graph-based representation of the design of the PCB according to one embodiment of the application; and Figure 6 shows an example of a classification of functional information, output by a GNN, according to one embodiment of the application. P2024,0845 WO N August 25, 2025 - 19 - Figure 1 shows a method 100 of analyzing a design of an electronic device according to one embodiment of the application. In a step 101, a data processing device obtains a design file of the electronic device, wherein the design file comprises a design of the electronic device, whose design is to be analyzed. The design file may be any file comprising information regarding the design of the electronic device. The design comprises explicit information regarding components that are comprised in the electronic device. Further details hereof have been discussed above and will not be repeated herein but apply accordingly. In a step 102, the data processing device determines EMC information of the electronic device based on the design of the electronic device of the obtained design file. EMC information in the context of this application describes information based on which it may be determined whether and / or how strong electromagnetic radiation is emitted and / or absorbed by elements of the electronic device and / or the overall system. Elements in the context of this entire application refer to both the explicit components shown in the design of the electronic device, but also to functional elements, which are created from one or more of those explicit components but are not explicitly mentioned or shown in the design of the electronic device. This is, for example, the case for filters and power converters, which are created in a design of an electronic device e.g. from resistors, P2024,0845 WO N August 25, 2025 - 20 - capacitors, etc., but are not marked as such in the design of the electronic device. EMC further describes information relating to conducted emissions, immunity (radiated and conducted), electrostatic discharge (ESD), power integrity, and signal integrity of the electronic device. The EMC information comprises, for example, information regarding any of the following: a grounding of elements of the electronic device, a shielding of elements of the electronic device, a decoupling of elements of the electronic device, a filtering of elements of the electronic device, a bypassing of elements of the electronic device. The step 102 of determining the EMC information of the electronic device comprises in this embodiment the two following sub steps 103 and 104. In step 103, the data processing device obtains component information regarding at least one component comprised in the electronic device based on the design of the electronic device. The component information is directly obtained from the information explicitly mentioned / shown in the design. The component information may, for example, comprise at least one of the following: an identifier of the at least one component, a component type of the at least one component, a number of pins of the at least one component, a grounding state of the at least one component, and a connection information of the at least one component. P2024,0845 WO N August 25, 2025 - 21 - In step 104, the data processing device determines functional information based on the component information, wherein the functional information is determined using a graph neural network, GNN. In this context, any GNN may be used. For example, a graph convolutional network, GCN, a graph attention network, GAT, and / or a graph attention network version 2, GATv2, may be used. Also, any combination thereof may be used. Further details regarding GNNs are discussed above and are not repeated herein, but apply accordingly. The functional information is information that is determined based on the explicit component information, but that is not explicitly comprised in the design file itself. The functional information is information regarding aspects that impact the EMC performance, for example in case of a PCB design, the EMC performance of a circuit. For example, the functional information comprises a filter information regarding the at least one component of the electronic device, and / or a power converter information regarding the at least one component of the electronic device. With other words, the functional information in those examples comprises information that the at least one component exhibits a filter and / or a power converter functionality in the design of the electronic device. The functional information signifies that a functional element, such as a filter or a power converter, is present in the design, even if the design itself does not explicitly mention or show such element. The functional information may be, for example, in the form of a number of corresponding functional elements being detected in the design of the P2024,0845 WO N August 25, 2025 - 22 - electronic device and / or in the form of identifying for components (some or all) comprised in the design of the electronic device whether they are part of such functional element or not. Optionally, step 102 of determining the EMC information may further comprise a step 105: obtaining, by the data processing device, edge information regarding the at least one component comprised in the electronic device based on the design of the electronic device, wherein the edge information comprises information regarding a connection between the at least one component and at least one further component. If this step 105 is present, the determining of functional information in step 104 is further performed based on the edge information, additionally to the component information. The edge information comprises, for example, at least one of the following: a trace length of a connection between the at least one component and at least one further component; a trace width of a connection between the at least one component and at least one further component; a trace type of a connection between the at least one component and at least one further component; a trace inductance of a connection between the at least one component and at least one further component; a net ID of a net comprising the at least one component and at least one further component; and an edge type of a connection between the at least one component and at least one further component, wherein the edge type defines whether a connection between the at least one component and an individual at least one further component is considered or whether a connection between the at least one component and a net of multiple further components via the at least one further component is considered. P2024,0845 WO N August 25, 2025 - 23 - According to the method 100 shown in Figure 1, the component information of each of the at least one component and / or, if step 105 is present, the edge information of each of the at least one component comprises at least one categorical value and / or at least one numerical value, wherein: the at least one categorical value belonging to each of the at least one component is processed using one-hot encoding to determine an affiliation of the respective component to a component group; and / or the at least one numerical value belonging to each of the at least one component is processed using a normalization preprocessing technique to normalize the at least one numerical value of the respective component. Details hereof are discussed above and are not repeated herein but apply accordingly. In a step 106, the data processing device analyzes the EMC information to determine whether at least one EMC issue is detected in the EMC information, wherein the analyzing the EMC information comprises analyzing the functional information. The analyzing may further comprise analyzing the component information. Details hereof are discussed above and are not repeated herein but apply accordingly. In a further optional step 107, the data processing device may determine feedback information based on a result of the analyzing of the EMC information. The feedback information in the context of this application may comprise an indicator of how well EMC requirements are met by the presented design, and / or the feedback information may be based on how many EMC issues were found and / or how critical with regard to EMC the EMC issues that were found P2024,0845 WO N August 25, 2025 - 24 - are. Additionally or alternatively, the feedback information may comprise direct information on what improvements with regard to EMC may be achieved in the design. Additionally or alternatively, the feedback information may comprise an adapted design in which the detected EMC issues are reduced or solved. In any case, the feedback information may be used by the user to improve the design regarding EMC. The method 100 as shown in Figure 1 may be implemented in any otherwise known development process of an electronic device. EMC issues may be determined during the process of designing the electronic device with this method 100. Such issues may be recognized, in particular before a prototype of the electronic device is created and tested. By determining EMC issues based on the design file, changes may be implemented in the design before the prototype is created, thereby saving resources, saving time of the development process, and saving cost of the development of the electronic device. The feedback information may then be used to improve the PCB design directly, without creating and testing a prototype. Figure 2 shows a system 1 for analyzing a design of an electronic device according to one embodiment of the application. In the embodiment shown herein, the system 1 comprises a data processing device 2. The data processing device 2 is configured to implement, in particular, the method 100 described with regard to Figure 1. The data processing device 2 may be, for example, any processor. In Figure 2, the system 1 is shown as a single device. The system 1, however, may also be implemented using e.g. cloud computing, wherein the P2024,0845 WO N August 25, 2025 - 25 - data processing device 2 may rely on distributed processors etc. The system 1 may alternatively also comprise dedicated modules for obtaining the design file, determining the EMC information, and analyzing the EMC information as well as optionally modules for implementing the optional steps of Figure 1, or any modules combining two or more of those tasks. Also a combination of a local device and a cloud or server based device may be used. Figure 3 shows an extract of a design of a PCB with an example of functional information and component information according to this application. The extract of Figure 3 shows only a part of a PCB design, which may be present in any larger design. The extract shows a first resistor 3, a second resistor 4, and a capacitor 5. The components 3, 4, and 5 are interconnected in such a way that a filter 6 can be detected in this design, even if no explicit filter is shown herein. In this example, the first resistor 3 would be identified as being part of the filter 6, and in this extract, one filter 6 would be detected to be present as a functional element. The information regarding the filter 6 being present and / or the first resistor 3 being part of the filter 6 is considered being functional information in the context of this application. The filter 6 being present and the first resistor 3 being part of the filter 6, in the context of this application, is determined using the GNN described in detail above. Since a filter has a larger impact on EMC than a mere resistor or capacitor, this additional information, i.e., the functional information, which is used in the above-discussed P2024,0845 WO N August 25, 2025 - 26 - step of analyzing the EMC information, allows a more accurate determination of EMC issues. In this entire application, in the above-discussed step of analyzing the EMC information, it may not only be considered whether changing properties of a single component but whether, by changing properties of a component, may change the impact of the functional element on the EMC. In the example of Figure 3, changing the properties of the first resistor 3, therefore, may have a larger impact on EMC since it is part of the filter 6 than if it would be an individual resistor bot being part of a filter. By considering the functional information, therefore, the EMC analyzation is more accurate. Figures 4A and 4B show symbolic visualizations of two examples of graph-based representations of components of an electronic device according to an embodiment of the application. Figure 4A discloses an example in which only single components are considered in the design, such as a resistor 7, a first capacitor 8, a second capacitor 9, and an inductor 10, as well as the connections 11 between each of those components. In this example, a component information regarding an ID of each component, a number of pins of each component, whether a component is connected to ground, as well as a component type (resistor, capacitor, etc.). An ID of the resistor 7 in this example is r41, of the first capacitor 8 is c53, of the second capacitor 9 is c43, and of the inductor 10 is l2. Number of pins and whether a component is connected to ground is not shown in the symbolic P2024,0845 WO N August 25, 2025 - 27 - visualization herein, but may be obtainable from the corresponding design file. Furthermore, connections 11 between each of the components 7, 8, 9, 10 are shown in this symbolic visualization. Regarding each connection 11, a trace length, a trace width, and a trace type (power, signal, clock, etc.) may be obtainable from the corresponding design file. This information disclosed herein relates to the component information and edge information discussed in detail above. Details will not be repeated herein, but apply accordingly. The example shown in Figure 4B also discloses the individual components, such as the resistor 7 with ID r41, the first capacitor 8 with ID c53, the second capacitor 9 with ID c43, and the inductor 10 with ID l2. Additionally, in Figure 4B, further elements are considered as components, i.e., nodes, in the graph: a voltage supply 12 with ID 3V3, a ground contact 13 with ID GND, a first net 14 of multiple component not shown individually with ID EM, and a second net 15 of multiple components not shown individually, the net 15 having the ID NetL2_1. The connections 11 are corresponding to the example of Figure 4A, however, for a better visualization, only a few of the connections 11 have reference signs in Figure 4B, the others applying correspondingly. The example of Figure 4B provides additional information in the graph directly, regarding the connections of the nodes to ground 13, to a voltage supply 12, or whether components are connected via the same net 15. This allows a more detailed analyzation with the GNN of the corresponding information than in the example of Figure 4A, in which, for example, the P2024,0845 WO N August 25, 2025 - 28 - grounding information is merely comprised as a parameter for the nodes and not as an explicit node in the graph, and in which the connection regarding nets is not represented at all. In the example of Figure 4B, the component information regarding each node may comprise the node ID, a node type (component or net), a number of pins, a component type, and a connection to ground. The edge information may comprise a trace length, a trace inductance, and an edge type (whether the connection 11 refers to a component-component connection or to a component-net connection). This version comprises significantly more topological information than the version of Figure 4A and, hence, allows a more accurate determination of the functional information using a GNN. Another version, not shown in the Figures, may use as component information the component ID, the number of pins, whether the component is connected to ground, and the component type, and as edge information a trace length, a trace inductance, a net ID, and a trace type. Any other combination of component information and edge information, however, may also be used. Figures 5A and 5B show an example of a design 16 of a PCB and a symbolic visualization of a graph-based representation 17 of the design of the PCB according to one embodiment of the application. The PCB design 16 shown in Figure 5A shows a large number of components placed on the PCB, as well as wires connecting those components. The PCB design 16 shown herein is merely used as an example and shall not be discussed in further detail. P2024,0845 WO N August 25, 2025 - 29 - The components shown in this PCB design 16 provide the explicit component information discussed in this application. The wires between the components shown in this PCB design 16 provide the explicit edge information discussed in this application. Those components and wires may be represented in the graph-based representation 17 shown in Figure 5B. Those Figures 5A and 5B are to be understood symbolic, to provide an understanding of the graph-based representation 17 of a PCB design 16, but are not to be discussed with respect to each individual component and connection shown herein. The information comprised in the graph-based representation 17 may then be analyzed with a GNN, and therefrom, the functional information discussed in this application may be obtained. Any of the above-discussed GNNs and any of the above-discussed versions of component information / edge information, or any other combination thereof, may be used for this analyzation. Regarding details, reference is made to the above-discussed disclosure, which herein applies accordingly. Figure 6 shows an example of a classification of functional information 18, output by a GNN, according to one embodiment of the application. The functional information in this example refers to filters. Any other functional information, however, may be treated accordingly. In this example, the functional information 18 is classified into a class RC_LPF 19 (i.e., low pass filters comprising at least one resistor and at least one capacitor), a class LC_LPF 20 (i.e., low pass filters comprising at least one inductor and at least one capacitor), a class RC_LPF_Pi 21 P2024,0845 WO N August 25, 2025 - 30 - (i.e., low pass filters comprising at least one resistor and at least one capacitor, following a PI structure), a class RLC_LPF 22 (i.e., low pass filters comprising at least one inductor, at least one resistor, and at least one capacitor), a class LC_LPF_T 23 (i.e., low pass filters comprising at least one inductor and at least one resistor, following a T structure), a class LC_LPF_Pi 24 (i.e., low pass filters comprising at least one inductor and at least one capacitor, following a PI structure), a class High_PF 25 (i.e., any high pass filter), and a class Non_Filter 26 (i.e., anything not belonging to a filter). At least one of the classes may be weighted using a WeightedRandomSampler, in order to cope with the imbalance between the classes, since, as can be seen in Figure 6, the class Non-Filter 26 is significantly larger than the other classes. The functional information in many of the example and embodiments described herein refers to filters. Correspondingly, the same aspects disclosed herein may also be applied to functional information relating to power converters, or any other corresponding functional elements that may be present in the designs of electronic devices. Moreover, many of the examples and embodiments discussed herein rely on the electronic device being a PCB. Correspondingly, the same aspects may apply, however, also to any other electronic devices. P2024,0845 WO N August 25, 2025 - 31 - List of reference signs 1 system 2 data processing device 3 first resistor 4 second resistor 5 capacitor 6 filter 7 resistor 8 first capacitor 9 second capacitor 10 inductor 11 connection 12 voltage supply 13 ground 14 first net 15 second net 16 PCB design 17 graph-based representation 18 functional information 19 to 26 classes 100 method 101 to 107 steps

Claims

P2024,0845 WO N August 25, 2025 - 32 - Claims 1. A method (100) of analyzing a design of an electronic device, the method (100) comprising the following steps: - obtaining (101), by a data processing device, a design file of the electronic device, wherein the design file comprises a design of the electronic device; - determining (102), by the data processing device, electromagnetic compatibility, EMC, information of the electronic device based on the design of the electronic device of the obtained design file, wherein the determining of EMC information comprises: - obtaining (103), by the data processing device, component information regarding at least one component comprised in the electronic device based on the design of the electronic device; - determining (104), by the data processing device, functional information based on the component information, wherein the functional information is determined using a graph neural network, GNN; - analyzing (106), by the data processing device, the EMC information to determine whether at least one EMC issue is detected in the EMC information, wherein the analyzing the EMC information comprises analyzing the functional information.

2. The method (100) according to claim 1, wherein the electronic device is a printed circuit board, PCB.

3. The method (100) according to claim 1 or 2, wherein the component information comprises at least one of the following: - an identifier of the at least one component;P2024,0845 WO N August 25, 2025 - 33 - - a component type of the at least one component; - a number of pins of the at least one component; - a grounding state of the at least one component; and - a connection information of the at least one component.

4. The method (100) according to any of claims 1 to 3, wherein the functional information comprises at least one of the following: - a filter information regarding the at least one component of the electronic device; and - a power converter information regarding the at least one component of the electronic device.

5. The method (100) according to any of claims 1 to 4, wherein: - the determining (102) of EMC information further comprises obtaining (105), by the data processing device, edge information regarding the at least one component comprised in the electronic device based on the design of the electronic device, wherein the edge information comprises information regarding a connection between the at least one component and at least one further component; and - the determining (104), by the data processing device, functional information is additionally performed based on the edge information.

6. The method (100) according to claim 5, wherein the edge information comprises at least one of the following: - a trace length of a connection between the at least one component and at least one further component; - a trace width of a connection between the at least one component and at least one further component;P2024,0845 WO N August 25, 2025 - 34 - - a trace type of a connection between the at least one component and at least one further component; - a trace inductance of a connection between the at least one component and at least one further component; - a net ID of a net comprising the at least one component and at least one further component; and - an edge type of a connection between the at least one component and at least one further component, wherein the edge type defines whether a connection between the at least one component and an individual at least one further component is considered or whether a connection between the at least one component and a net of multiple further components via the at least one further component is considered.

7. The method (100) according to any of claims 1 to 6, wherein the GNN for determining the functional information is at least one of the following: - a graph convolutional network, GCN; - a graph attention network, GAT; and - a graph attention network version 2, GATv2.

8. The method (100) according to any of claims 1 to 7, wherein the component information of each of the at least one component and / or, if referring back to claim 5, the edge information of each of the at least one component comprises at least one categorical value and / or at least one numerical value, wherein: - the at least one categorical value belonging to each of the at least one component is processed using one-hot encoding to determine an affiliation of the respective component to a component group; and / orP2024,0845 WO N August 25, 2025 - 35 - - the at least one numerical value belonging to each of the at least one component is processed using a normalization preprocessing technique to normalize the at least one numerical value of the respective component.

9. The method (100) according to any of claims 1 to 8, wherein the functional information is divided into at least two classes and, for the analyzing the EMC information, at least one of the classes is weighted using a Weighted Random Sampler.

10. A system (1) for analyzing a design of an electronic device, the system (1) comprising a data processing device (2), wherein the data processing device (2) is configured to: - obtain a design file of the electronic device, wherein the design file comprises a design of the electronic device; - determine electromagnetic compatibility, EMC, information of the electronic device based on the design of the electronic device of the obtained design file, wherein to determine the EMC information, the data processing device is configured to: - obtain component information regarding at least one component comprised in the electronic device based on the design of the electronic device; - determine functional information based on the component information, wherein the functional information is determined using a graph neural network, GNN; - analyze the EMC information to determine whether at least one EMC issue is detected in the EMC information, wherein the analyzing the EMC information comprises analyzing the functional information.P2024,0845 WO N August 25, 2025 - 36 - 11. A computer program product comprising instructions which, when executed on a computing device, implements the method (100) according to any of claims 1 to 9.

12. A computer-readable storage medium comprising the computer program product according to claim 11.

Images