The increasing demand for high-resolution displays and high-capacity data transmission has led to the rapid evolution of high-speed serial link technologies [1–3]. In such high-speed signal environments, significant degradation in signal quality is often observed as a result of frequency-dependent loss and inter-symbol interference (ISI), which reduces the noise margin and makes systems more susceptible to external electromagnetic noise. To overcome these limitations, previous studies have investigated the causes of such signal degradation and proposed various circuit designs, protection techniques, and countermeasures to mitigate electromagnetic interference (EMI) in high-speed digital systems [4, 5]. However, conventional electromagnetic immunity evaluations, such as those based on the IEC 61000-4-3 standard, rely primarily on qualitative assessments derived from the operational status of devices [6]. Meanwhile, probe-based injection methods, such as near-field or differential probe tests, exhibit limitations in quantitatively evaluating the electromagnetic immunity of the device [7]. Overall, these approaches are insufficient for quantitatively analyzing the effects of EMI in complex systems, such as display devices, that integrate multiple sources of noise.

This paper proposes a method for identifying immunity vulnerabilities in display devices through near-field emission measurements. The electromagnetic immunity is quantitatively evaluated by injecting near-field noise into the identified locations and analyzing the resulting pixel and frame error rates. During near-field noise injection, the pixel error ratio (PER) and frame error ratio (FER) of the screen are measured as performance metrics, and linear modeling is conducted using these metrics for the performance degradation zone. Furthermore, drawing on the linear model of performance degradation with regard to the injected noise amplitude, the display noise susceptibility index (DNSI) is proposed to quantitatively evaluate the immunity of the display device at various testing frequencies. The proposed method is validated by carrying out near-field noise injection testing at the clock frequency and at adjacent frequency bands of a high-definition multimedia interface (HDMI) under operating conditions. In addition, the effectiveness of the proposed method is experimentally validated by quantitatively evaluating electromagnetic noise immunity at testing frequencies of interest.

Since the coupling path and magnitude of near-field noise are determined by the location where the EMI is generated in a target evaluation system, the vulnerable parts of a system should first be identified to ensure repeatable and quantitative immunity evaluation. In this study, the identification process was conducted based on the principle of electromagnetic reciprocity, which assumes that regions with high near-field noise emissions are vulnerable to external electromagnetic noise interference. Fig. 1 depicts a near-field emission measurement setup, where a high-speed serial link with an HDMI is the device under test (DUT), and a Raspberry Pi 4B is used as the signal transmission system. The area near the cable that connects the HDMI source and link, denoted using the yellow dashed lines in Fig. 1, is designated for emission measurement, which was conducted by transmitting a video signal with 1,080p resolution and 60 Hz refresh rate using transition minimized differential signaling (TMDS). To capture noise emission, a magnetic probe connected to a spectrum analyzer was placed at a height of 1 mm above the DUT surface, and near-field noise scanning was conducted over the measurement area at a spatial resolution of 1 cm × 1 cm [8]. The emission strength at each measurement point was determined using Max Hold mode. The frequency-dependent electromagnetic emission distribution on the DUT surface was quantitatively analyzed, revealing strong emission at the HDMI clock frequency of 148.5 MHz [9].

To validate the assumption that regions with high emissions are vulnerable to EMI, an electromagnetic immunity test setup was configured, as shown in Fig. 2. In this immunity test setup, near-field noise at the HDMI clock frequency was generated by a signal generator, amplified by a power amplifier, and then injected into the DUT, with the probe configuration and testing area being equivalent to those in the noise emission measurement, starting from a low amplitude and increasing in 1 dBm steps. The noise amplitude at which the sub-pixel illumination defect (Type 3)—a defect specified in the ISO 9241-307 standard that frequently occurs in near-field noise injection conditions—first appeared on the video output was considered the electromagnetic immunity threshold of the DUT [10].

Based on the noise emission measurement, it was confirmed that the HDMI channel and connector parts, which connect the processor generating the HDMI clock signal and the external display, are the primary sources of electromagnetic emissions. Subsequent electromagnetic immunity testing further verified that these areas were vulnerable to injected noise, thereby validating the assumption regarding reciprocity considered in the proposed method (Fig. 3).

In this section, a method for quantitatively evaluating the electromagnetic immunity of display devices under near-field noise injection is proposed. To ensure a quantitative evaluation with high repeatability, electromagnetic immunity testing was conducted on the most vulnerable areas of the system, as identified in Section II. In EMI environments, visual errors on display devices may occur momentarily or persist in localized screen areas, making it difficult to quantitatively evaluate a device’s immunity level based on subjective judgment. To overcome this limitation, variations in pixel- and frame-level error rates under external noise conditions were measured in this study. Subsequently, these rates were employed as performance metrics to quantify the subtle errors occurring in display devices. For this purpose, the electromagnetic immunity testing setup shown in Fig. 2 was supplemented with a camera and an analysis PC to construct the display error detection and evaluation setup illustrated in Fig. 4. The DUT outputs a 1080p/60 Hz video signal to the external monitor via an HDMI interface, and the screen is captured by a camera maintaining the same resolution and refresh rate to detect pixel and frame errors.

In this study, two performance metrics—PER and FER—are considered to quantitatively evaluate the electromagnetic immunity of display devices. PER_j is the ratio of erroneous pixels to the total number of pixels in the j-th frame during near-field noise injection, as formulated in Eq. (1):

Fig. 5 shows that the pixel error PE(i, j) is a binary value indicating whether an error occurs at the i-th pixel location in the j-th frame, measured as 1 if an error occurs and 0 otherwise. Furthermore, NoP refers to the total number of pixels within a single frame, defined as M × N based on the display resolution. In this study, Max(PER_j), which is the pixel error rate in the frame with the highest number of pixel errors, is used as a performance metric to evaluate the transient error characteristics of the display device. Since this metric reflects the most severe error state experienced by the system during the entire observation period under EMI, it serves as a quantitative criterion for evaluating the immunity limit under worst-case transient EMI conditions.

The proposed evaluation method also employs another metric, FER, to quantitatively evaluate persistent errors occurring at specific locations on the display screen. Since high-resolution displays have finer pixel structures, the entire screen was divided into pixel blocks of size B × B based on the resolution perceptible to the human eye. Upon near-field noise injection, block error BE(k, j) refers to an error occurring at the k-th block position in the j-th frame among the total number of frames (NoF), as illustrated in Fig. 5. Meanwhile, FER_k refers to the ratio of NoF in which an error occurs at the k-th block position to the total NoF, as expressed in Eq. (2):

In this study, Mean(FER_k) is used as one of the performance metrics to evaluate persistent errors on the display screen. It is calculated as the average FER_k across all NoB block positions, where NoB refers to the number of blocks in a frame. This metric serves as a criterion for evaluating immunity levels against persistent and accumulated errors on the display screen under EMI conditions.

The performance metrics Max(PER_j) and Mean(FER_k) reflect the transient and cumulative error characteristics of the display device, respectively, together serving as a comprehensive criterion for evaluating the electromagnetic immunity level, both temporally and spatially. Notably, since both metrics are defined as ratios, the evaluation results remained consistent regardless of the absolute resolution, pixel size, or number of pixels and blocks, ensuring general applicability across various display configurations.

Fig. 6 shows the Max(PER_j) and Mean(FER_k) measured at 151.5 MHz under varying near-field noise amplitudes, with three repeated runs conducted under identical conditions. All experiments were conducted using fixed cables, probes, display, and camera settings while also maintaining identical test environments to ensure repeatability. Both metrics converged with an average error of 5%, confirming the repeatability of the measurements. Based on this verification, subsequent experiments were analyzed using a single measurement under the same conditions.

Figs. 7 and 8 present the measurement results of Max(PER_j) and Mean(FER_k) with regard to the injected noise amplitude at various testing frequencies. It can be observed that as the injected noise amplitude increased, the quality of the display output degraded, with both performance metrics exhibiting an increasing trend. Based on the specific threshold values for each performance metric, a normal operation zone unaffected by noise injection and a linear increase zone were observed. Notably, the susceptibility of each metric to noise, represented by the slope of the linear region, varied depending on the frequency.

Based on these observations, a method for quantitatively evaluating electromagnetic immunity according to the injected near-field noise frequency was devised, as shown in Fig. 9. In the proposed method, electromagnetic immunity is evaluated by dividing the response into two zones—the normal operation zone, where no significant effect is observed, and the performance degradation zone, where the metrics increase linearly—and then applying the corresponding performance metrics to each zone. For the performance metrics employed in the proposed evaluation method, the threshold values for Max(PER_j) and Mean(FER_k)—PER_TH and FER_TH, respectively—were set as criteria for stable video playback performance. The range between A_TH, which refers to the first injected near-field noise amplitude that exceeds the threshold, and A_max, which is the maximum injected noise amplitude, was considered the performance degradation zone, which was approximated using a linear model. The linear model for the performance degradation zone was derived by applying first-order linear approximation using the least squares method to the performance metric values measured within that zone. The resulting slope, denoted as S(A), can be considered a quantitative indicator of susceptibility to performance degradation.

In this study, a new metric—the DNSI—is proposed as a representative metric for evaluating pixel and frame error rates under various near-field noise injection frequencies in display devices. As expressed in Eq. (3), the DNSI quantifies electromagnetic immunity in the performance degradation zone by calculating the triangular area formed by the threshold separating the normal operation zone from the degradation zone in the linear model. This evaluation metric was identically applied to each performance metric—Max(PER_j) and Mean(FER_k).

The proposed DNSI quantification model accounted for both the slope of the linear model in the performance degradation zone and the amplitude of the external noise at which degradation first occurred. Accordingly, the DNSI value increased when the slope representing susceptibility to an increase in the error rate was steeper, and when degradation began at a relatively lower noise amplitude. In other words, the DNSI value was observed to be larger when performance degradation started at a lower noise amplitude, with the system becoming more susceptible as the noise amplitude increased.

Overall, the proposed evaluation metric serves as a representative metric reflecting both the external noise amplitude and susceptibility to performance degradation, and thus can be effectively used to quantitatively compare and evaluate the electromagnetic immunity of a target device across various frequencies.

This section presents the experimental validation results of the electromagnetic immunity evaluation method proposed in Section III. The performance metrics of the display device were measured under varying noise amplitudes, and the proposed method was applied to quantitatively evaluate electromagnetic immunity at each injection frequency. The measurement was conducted using the same near-field noise injection setup described in Fig. 3.

Figs. 10 and 11 show the measurement results of the proposed DNSI when using the Max(PER_j) and Mean(FER_k) metrics at 151.5 MHz—a frequency adjacent to the TMDS clock frequency of 148.5 MHz—where immunity is observed to be relatively weak. In this experiment, the pixel block size B was set to 8, adhering to the commonly used standard in JPEG compression [11]. The threshold value PER_TH, employed to define the normal operation zone of Max(PER_j), was set to 0.0005 based on the international standard ISO 9241-307, which is widely adopted by display manufacturers as a quality control criterion. Meanwhile, the threshold value FER_TH for Mean(FER_k) was set to 0.0017, considering the flicker detection limit of the human visual system [10, 12–14]. The dotted lines in Fig. 10 show that the performance metrics measured at each frequency degraded linearly as the injected noise amplitude increased. Meanwhile, the solid lines represent the linear modeling results for the performance degradation zone based on the proposed method. These results clearly demonstrate that the proposed segmentation method based on the linear model is capable of successfully distinguishing the normal operation zone from the performance degradation zone. This, in turn, allows for the identification of the degradation threshold and enables a quantitative analysis of the susceptibility to performance degradation.

Fig. 12 presents the DNSI results derived from Max(PER_j) and Mean(FER_k) for all test frequencies, while their detailed values are summarized in Table 1. It is observed that the DNSI at the HDMI clock frequency of 148.5 MHz is quite low, indicating strong immunity, while higher DNSI values are observed at the adjacent side channel frequencies, indicating increased susceptibility. These results confirm that the proposed immunity evaluation metric successfully quantified the degradation in the output quality of the display device upon the injection of electromagnetic noise at and around the HDMI clock frequency. Moreover, the DNSI values based on Max(PER_j) and Mean(FER_k) showed similar trends to the measurement results, proving that the use of both performance metrics enables reliable electromagnetic immunity evaluation that reflects both transient and cumulative error characteristics. It was also confirmed that the proposed method can be effectively used to clearly identify design-level vulnerabilities that may be overlooked when relying on a single metric, such as degradation phenomena that occur intermittently or repeatedly at specific locations under EMI conditions.

The experimental validation establishes that the proposed electromagnetic immunity evaluation method enables frequency-dependent quantitative analysis of display immunity levels rather than simply determining functional operations, as is the case with conventional methods. In particular, the proposed method offers the advantage of effectively identifying the vulnerable frequency bands of display devices by providing a representative immunity metric for each frequency.

Notably, the threshold used for the measurement and validation conducted in this study was established based on international standards and human visual perception characteristics. Depending on a device’s performance margin, system engineers can flexibly adjust this value to predict electromagnetic immunity against external noise in the early design stage and apply the results as feedback, thereby enabling a sophisticated electromagnetic immunity design.

This study proposes a method for quantitatively evaluating the electromagnetic immunity of display devices based on near-field noise injection testing. The proposed evaluation approach identifies the vulnerable parts of a system based on external noise by locating areas with strong electromagnetic noise emissions. To quantify this vulnerability, near-field noise characterized by various frequencies and amplitudes was injected into the identified vulnerable parts of the target evaluation system, following which the resulting pixel and frame error rates were measured as performance metrics. Based on the measurement results, the immunity characteristics of the device were classified into the normal operation zone or the performance degradation zone, depending on the sensitivity of the metrics to the injected noise. Furthermore, in this study, DNSI is proposed as a quantitative evaluation metric for the linear modeling of the performance metrics in the performance degradation zone. The proposed evaluation method was experimentally validated by inspecting the electromagnetic immunity of an HDMI display device at the clock frequency and adjacent frequencies. The validation results confirmed that the proposed method has the capacity to successfully quantify the noise immunity level of a target display depending on the noise frequency.

In conclusion, the proposed method can serve as a practical electromagnetic immunity evaluation system for various display devices because it enables the quantitative comparison of immunity levels across frequencies, as well as the effective identification of vulnerable frequency bands.