RFI Mitigation Systems For Smart Phones by Automatic Calibration of MIPI Data Rate
Editor’s Note: The paper on which this article is based was originally presented at the 2025 IEEE International Symposium on Electromagnetic Compatibility & Signal/Power Integrity (EMC + SIPI), where it received recognition as the Best Symposium Paper. It is republished here with the gracious permission of the IEEE. Copyright 2025 IEEE
I
n recent years, the number of mounted semiconductor IC modules and the operating frequency have increased along with the enhancement of functions of smart phones. In the design of smart phones, the risk of electromagnetic interference (hereinafter referred to as EMI) generation is rising, leading to higher costs associated with EMC design.
Particularly, wireless receiver antennas are sensitive for unwanted electromagnetic radiation inside smart phones, posing a risk of radio frequency interference (hereinafter referred to as RFI), which can cause problems such as communication failures. One of the reasons for RFI is the noise radiated by semiconductor IC modules processing high-speed signals, located near wireless receiver antenna.1
Among the reasons, mobile industry processor interface (hereinafter referred to as MIPI) signals output by camera modules as image signals are prone to causing RFI. MIPI is one of the high-speed interface standards and transmits signals to motherboards via radiation-efficient flexible cables, thus causing relatively large noise radiation. Various studies have been conducted on analysis method2 and countermeasures3 for noise radiation caused by MIPI. Common countermeasures for noise radiation from MIPI signals propagated via flexible cables include conductive shielding and coating materials.4,5 These EMI countermeasure components are effective in suppressing noise radiation levels. However, these countermeasures require significant costs for component implementation and evaluation.
This paper proposes a low-cost system that automatically mitigates RFI in smart phones caused by noise radiated from flexible cables carrying MIPI signals. Figure 1 illustrates this concept. The application processor (hereinafter referred to as AP) calibrates the MIPI data rate through I2C communication with the image sensor, using received signal strength indicator (RSSI) within a channel as a feedback control variable, thereby mitigating the risk of RFI generation.
Figure 1: Smart phone RFI mitigation system through dynamic transition of MIPI data rate
The structure of this paper is as follows: The first section (“RFI Generation Mechanism Due to Noise Radiation from MIPI D-PHY”) explains the mechanism of RFI generation due to noise radiation from MIPI D-PHY. The second section (“Proposed Method”) proposes a system to mitigate RFI caused by noise radiation from MIPI D-PHY. The third section (“Construction of an RF Circuit and an Evaluation Board for Measurement”) describes the design method for the RF circuit and test board used in the evaluation of the proposed system. The fourth section (“Measurement System”) explains the measurement environment configuration for the proposed system. Finally, the fifth section (“Measurement Results”) demonstrates the effectiveness of the proposed system through measurement results.
RFI Generation Mechanism Due to Noise Radiation From MIPI D-PHY
The subject of evaluation is MIPI D-PHY, a standard communication protocol used for unidirectional image transmission between a camera module or a liquid crystal display (LCD) and an AP.6
Characteristics and Mechanism of Noise Radiation from MIPI D-PHY
Figure 2 shows voltage waveform and power spectrum of differential mode, as well as voltage waveform and power spectrum of common mode, for one data lane of MIPI D-PHY output from a camera module. The data rate of MIPI D-PHY is 2.5 Gbps. Since the camera is imaging under the dark condition, the output bits of MIPI D-PHY form a fixed pattern. This fixed pattern has periodicity at one-quarter of the operating clock frequency. Fixed patterns occur during normal shooting and more frequently under the dark condition. The frequency component of the fixed pattern has high peak levels and can cause RFI. Meanwhile, the radiated noise from the clock lane can also cause RFI because the frequency component of the clock has high peak levels.
Figure 2: Voltage waveforms and power spectra of MIPI D-PHY
The transmission method of MIPI D-PHY is differential mode. Ideally, MIPI D-PHY signals should not have common mode components. In practice, however, common mode components are generated in MIPI D-PHY transmission lines due to variations in the output timing of positive and negative signals driven by the transmission circuit inside the IC, and differences in the transmission line length. The common mode component creates a large electromagnetic loop because a return current passes through frame GND, resulting in significant noise radiation and thus causing smart phone RFI. Therefore, measures to mitigate the impact of the radiation are necessary.
RFI Issue Due to Noise Radiation from MIPI D-PHY
Figure 3 overlays the measured frequency peaks of the common mode component generated by MIPI D-PHY operating at a data rate of 2.5 Gbps, on the 4G LTE communication bands in North America.7 Telecommunications carriers use different bands defined by law. Designing smart phones requires ensuring that the frequency peaks of the common mode component of MIPI D-PHY do not interfere with all the communication bands that communication systems use, in order to mitigate the risk of RFI. However, it is difficult to uniquely determine a data rate that prevents noise frequency peaks from overlapping with all communication bands of smart phones. Thus, designing smart phones have required significant efforts to adjust data rates of MIPI D-PHY.
Figure 3: MIPI D-PHY common mode noise component and 4G LTE communication bands
Proposed Method
MIPI Data Rate Automatic Adjustment System
Figure 4 illustrates a basic configuration of an AP and an RF circuit inside a smartphone, along with the detailed operational flow of the proposed system utilizing the functions of the AP and the RF circuit. Signals received by the built-in antenna inside the smart phone are input into the RF circuit, which outputs signals within a channel (hereinafter referred to as RF_OUT). RF_OUT is encoded by the AP and used to estimate communication quality as an RSSI. When the noise radiation from MIPI D-PHY interacts with the built-in antenna and the frequency component of the noise radiation from MIPI D-PHY overlaps with the channel, the RSSI increases. Thus, monitoring RSSI fluctuations enables the AP to detect communication interference caused by MIPI D-PHY noise radiation. As described in the first section of this paper, the frequency component of MIPI D-PHY noise radiation depends on MIPI D-PHY data rates. Calibrating data rates of MIPI D-PHY to minimize RSSI enables the AP to mitigate RFI risks caused by MIPI D-PHY noise radiation.
Figure 4: MIPI D-PHY data rate automatic adjustment system
Figure 5 shows a procedure for changing a MIPI D-PHY data rate. The camera module and the AP communicate via I2C signals (SCL and SDA). The AP overwrites the register values inside the camera module (image sensor) via I2C to change the multiplication and division settings of a phase‑locked loop (hereinafter referred to as PLL). This changes the input CLK frequency of the MIPI D-PHY interface, thereby changing the MIPI D-PHY data rate.
Figure 5: Procedure for changing MIPI D-PHY date rate
Figure 6 illustrates the timing for changing the MIPI D-PHY data rate. The data rate is changed during V-blank periods. A V-blank is set between each frame of data. During V-blank periods, the camera module does not send image data to the AP but sends signals called VSYNC to synchronize the timing of frame data transmission. Using VSYNC, the AP compensates for any timing deviations of image data transmission caused by changes in the MIPI D-PHY data rate. Thus, the proposed system calibrates the MIPI D-PHY data rate without affecting the image communication.
Figure 6: Data rate change timing of camera module
Algorithm
Figure 7 on page 40 presents a specific flowchart of the data rate automatic adjustment system as an example. First, it is determined whether the RSSI value is equal to or greater than the target value. If the RSSI is equal to or greater than the target value, the calibration of the MIPI D-PHY data rate begins. The MIPI D-PHY data rate is calibrated as follows: The AP checks the RSSI while changing the MIPI D-PHY data rate by the change of the PLL register values. This process determines the MIPI D-PHY data rate that minimizes the RSSI. Finally, the AP changes the PLL register values to output the MIPI D-PHY at the optimal data rate that minimizes the RSSI.
Figure 7: Algorithm for MIPI D-PHY data rate automatic adjustment
Construction of an RF Circuit and an Evaluation Board for Measurement
To evaluate the reduction in RSSI achieved by the proposed method, we designed test circuits emulating an RF circuit and a camera module inside a smart phone. This chapter presents the design results of these test circuits using simulations.
Design of the RF Circuit
Figure 8 shows a configuration of the RF circuit and transition of the amount of received RF power (hereinafter referred to as RF budget). The RF circuit is mainly composed of filters, amplifiers, a mixer, and an RF detector, which are constructed as discrete circuits with SMA connectors as input/output terminals. The functions of each circuit are as follows: the filters before the mixer perform band limitation, the filters after the mixer have a passband of approximately 10 MHz, determining the channel, the mixer down-converts the frequency of the received signals, the amplifiers improve the noise figure or adjust the gain, and the RF detector outputs the DC voltage corresponding to the total power of signals within the channel. The power divider before the RF detector is used for signal monitoring. The antenna connected to the RF circuit is a half-wave dipole antenna for 840 MHz band.
Figure 8: Configuration of RF circuit and transition of received RF power
The RF budget shows the transition of signal levels existing in the channel range from 877.5 to 887.5 MHz at each I/O port of the RF circuit. The changes in signal levels are calculated based on the gain and the loss of each amplifier and mixer of the circuit. Using the RF budget, we designed the RF circuit. First, to evaluate the noise near the noise floor of the RF circuit, we determined the antenna reception level range to be -104 to -54 dBm. Next, we selected an RF detector with a detection range that matches the antenna reception level range. Finally, using the RF budget, we selected filters, amplifiers, and a mixer, considering the gain and the noise figure, to ensure a one-to-one correspondence between the antenna reception level and the RF detector input level. In the evaluation of measurements, the power target for the antenna reception level was set to -64 to -54 dBm, considering the upper input level limit of the RF detector. The next section presents the design and simulation results of the evaluation board for noise radiation that meets the power target at the antenna output terminal.
Design and Simulation Results of the Evaluation Board
To receive noise radiation from MIPI D-PHY with the antenna, the evaluation board was designed with MIPI D-PHY wiring routed only on the surface layer. The MIPI D-PHY wiring consists of five lanes, with the clock lane positioned at the center and the data lanes arranged on both sides of the clock lane. Figure 9 shows the results of analyzing the amount of power received by the antenna using a full-wave 3D electromagnetic field simulator. The data rate of MIPI D-PHY, acting as a noise source, was set to 1680 Mbps so that the noise peak would interfere with the reception band of the 840 MHz half-wave dipole antenna. Through such simulations, the MIPI wiring length was designed to ensure that the noise radiation from MIPI D-PHY falls within the target values at the antenna output terminal. The final estimated signal power level was approximately -48 dBm. Properly adjusting the distance between the antenna and the evaluation board keeps the input power within the target range of -64 to -54 dBm. The reason the received signal strength does not increase when the antenna is placed close to the board is that the increased parasitic capacitance changes the resonant frequency of the antenna, thereby reducing its gain.
Figure 9: Estimated radiation level from the evaluation board
Measurement System
Figure 10 on page 42 illustrates a measurement environment. The environment is used to measure the mitigation of RFI obtained by the proposed system operation. The environment emulates the configuration of the internal components of a smart phone, such as a camera module, a receiver antenna, an RF circuit, and an AP.
Figure 10: Measurement evaluation environment
Evaluation Board
Figure 11 shows a measurement environment for noise radiation from MIPI D-PHY using an evaluation board equipped with an image sensor for smart phones. Only MIPI D-PHY wiring is placed on the surface of the evaluation board, whereas all components are mounted on the backside. The evaluation board is housed in an aluminum shield case during measurements. Only the MIPI D-PHY wiring is exposed through an aperture in the shield case. The shield case is placed inside an electromagnetic anechoic box with a shielding effectiveness of 70 dB at 2.5 GHz band. This setup allows the half-wave dipole antenna positioned above to receive only noise radiation from MIPI D-PHY. Register signals applied from outside the evaluation board controls the data rate of MIPI D-PHY. The image sensor is set to the dark condition.
Figure 11: Environment for measuring noise radiation from MIPI D-PHY
RF Circuit
Figure 12 shows the RF circuit constructed using the design method described in the previous chapter. The RF circuit configured with these discrete circuits has a noise figure of 3.5 dB. The RF circuit outputs DC voltages corresponding to the total power of signals existing in the channel (approximately 10 MHz bandwidth) among the signals received by the half-wave dipole antenna. The measurement band used in this study is set to 840 MHz band, used by multiple telecommunications carriers in Japan.
Figure 12: Photograph of RF circuit configuration
A/D Converter and FPGA
DC voltages output by the RF circuit are converted to digital values by an A/D converter and an FPGA in the subsequent stage of the system. The DC voltage value data are decoded by a PC, then evaluated as RSSI.
Sequence of the Measurement System
Figure 10 shows the FPGA and the PC for controlling the image sensor emulate the operation of the AP inside a smart phone. First, the image sensor is activated and set to capture mode. Then, the noise radiation generated by the MIPI D-PHY operation is received by the antenna. The RF circuit detects this noise and then the PC recognizes it as an RSSI through the subsequent A/D converter and FPGA processing. The PC searches for the optimal data rate of the MIPI D-PHY to minimize the RSSI. This sequence moves the peak of the noise radiation from MIPI D-PHY outside the channel. The next section presents the results of operation verification and RSSI performance evaluation in this measurement environment.
Measurement Results
Operation Verification of the RF Circuit
Figure 13 shows the results of verifying the RF circuit operation. The data rate of the MIPI D-PHY is 1680 Mbps.
Figure 13: Noise power existing in RF circuit (a) antenna output, (b) RF detector input
Under the condition where the noise radiation from the MIPI D-PHY received by the half-wave dipole antenna at 840 MHz was input into the RF circuit, each power spectrum of the signal at the antenna output terminal and the RF detector input terminal was observed by a spectrum analyzer. The frequency of the target noise peak was set to 840 MHz. The local oscillator frequency to be input to the mixer was adjusted to move the noise peak frequency to 240 MHz, the center of the passband of the band-pass filter. Finally, the power spectrum of the noise to be input to the RF detector also yielded reasonable results. The measurement results obtained in Figure 13 closely match the calculated results (Figure 8) previously described, confirming the correct operation of the RF circuit we designed.
Reduction in RSSI by the Proposed System
The reduction in RSSI by the proposed system was evaluated through measurements. Figure 14 (a) shows the power spectra of the input signal to the RF detector, obtained using a spectrum analyzer. The initial data rate of the MIPI D-PHY was 2460 Mbps. The proposed system was operated on the assumption that the common-mode noise peak (799.5 MHz) generated by the MIPI D-PHY data and clock signal was mixed into the center of the channel range from 794.5 to 804.5 MHz. Our system automatically searched for the optimal data rate of the MIPI D-PHY to minimize the RSSI, and in this case, the optimal data rate was 2424 Mbps. The data rate shift of 36 Mbps confirmed that the noise peak of the MIPI D-PHY moved outside the channel.
Figure 14 (b) shows the relationship between the MIPI D-PHY data rate and the RSSI detected by the RF detector. When the noise peak was at the center of the channel, the RSSI value was -3.2 dBm. On the other hand, applying our proposed system reduced the RSSI to -12.3 dBm, confirming the improvement of 9.1 dB. These results demonstrated the effectiveness of the proposed system in improving RSSI.
Figure 14 (a) and (b): Spectra of input signal to RF detector, (b) detected RSSI values by RF detector
Impact of the Proposed System on Image Quality
Even if the RFI caused by the image sensor inside a smart phone is mitigated, any impact on image quality will be problematic. Here, we examined whether changing the MIPI D-PHY data rate would affect image quality.
Figure 15 shows the results of measurements to check the relationship between the MIPI D-PHY data rate and image quality characteristics of the image sensor. The imaging condition was set too dark. The image quality characteristics are as follows: Dark shading is defined as the peak-to-peak values of the average signal levels of each vertical and horizontal pixel line. Random line noise is defined as the standard deviations of the signal levels of each vertical and horizontal line divided by square root of two. The values of each image quality characteristic are normalized based on the values at the data rate of 2496 Mbps.
Figure 15: Correlation between MIPI D-PHY data rate and image noise (a) dark shading, (b) random line noise
When the MIPI D-PHY data rate was changed within the range of ±200 Mbps from 2496 Mbps, the variation in image noise output by the sensor was confirmed to be negligibly small. This study verified that our proposed system maintains image quality characteristics even when the MIPI D-PHY data rate varies during actual operations.
Verification of the Proposed System Immunity to External Noise
The calibration of the MIPI D-PHY data rate by the proposed system is expected to be performed in a factory before product shipment. In this operating environment, smart phones do not communicate wirelessly with the outside. However, the wireless receiver antenna can receive noise radiation from other IC modules besides the MIPI D-PHY. Therefore, the proposed system must operate correctly even when the RSSI changes due to factors other than the MIPI D-PHY noise radiation.
Figure 16 shows the measured probability that the proposed system could calibrate the MIPI D-PHY data rate when additive white Gaussian noise (hereinafter referred to as AWGN) was input to the RF circuit. AWGN, which is the thermal noise of a resistor amplified by an LNA, was input to the output terminal of the half-wave dipole antenna via a directional coupler. The increase in the power intensity of AWGN decreases the S/N ratio. Under these conditions, the reduction in RSSI when the noise peak of the MIPI D-PHY moves out of the channel becomes smaller, and calibrating the data rate of the MIPI D-PHY becomes difficult for an AP. In calculating the S/N ratio, a signal power was defined as the maximum power of a single peak of the MIPI D-PHY mixed into the center of the channel. On the other hand, a noise power was defined as the maximum total power of AWGN existing in the channel. The number of attempts to search for the optimal MIPI D-PHY data rate by the proposed system was set to 20. The measurements confirmed that 0 dB or higher S/N ratio allows the proposed system to calibrate the MIPI D-PHY data rate with 100% probability.
Figure 16: Success probability of RSSI reduction system under external noise interference
Conclusion
This paper presented a system that automatically mitigates RFI risk caused by MIPI D-PHY noise radiation. We demonstrated that with enhanced AP functionality, exploring dynamically the optimal data rate of MIPI D-PHY suppresses interference in the channel caused by the noise radiation from MIPI D-PHY. This system is realized only through the software functionality of the AP, requiring no additional component costs. Additionally, the proposed system can be applied to IC products controllable by an AP, beyond just camera modules. The proposed system operates at any time after smart phone manufacturing, allowing for noise radiation countermeasures tailored to the characteristics of each smart phone, thereby potentially reducing the turnaround time for product design.
References
- S. Kim et al., “Simulation-based analysis on EMI effect in LPDDR interface for mitigating RFI in a mobile environment,” 2016 IEEE 20th Workshop on Signal and Power Integrity (SPI), 2016, pp. 1-4.
- Y. Han, S. Lee, H. A. Huynh, and S. Kim, “Radio Frequency Interference Analysis of Camera Module and Antenna in Smartphones,” 2019 International Conference on Electronics, Information, and Communication (ICEIC), 2019, pp. 512-515.
- Y. Chien, H. Liu, C. Chou, and T. Wu, “An Embedded Common-Mode Filter and Mixed-Mode Scattering Parameters for the MIPI C-PHY Interface,” IEEE Transactions on Electromagnetic Compatibility, vol. 66, no. 1, pp. 143-152, Feb. 2024.
- R. Kumar, A. Kumar, and D. Kumar, “RFI/EMI/microwave shielding behaviour of metallized fabric-a theoretical approach,” International Conference on Electromagnetic Interference and Compatibility ‘99 (IEEE Cat. No. 99TH 8487), 1997, pp. 447-450.
- S. Yang et al., “Effects of Shielding Materials on EMI Performance for 5G Wi-Fi Applications,” 2022 17th International Microsystems, Packaging, Assembly and Circuits Technology Conference (IMPACT), 2022, pp. 9-12.
- MIPI alliance specification for D-PHY version 3.0, July 2021.
- “3GPP TS 36.101: User Equipment (UE) Radio Transmission and Reception (Release 12),” September 2024.
Share this story:
The Authors
Akihiro Tsukioka is part of the Design Platform Technology Division at Sony Semiconductor Solutions Corporation and can be reached at akihiro.tsukioka@sony.com.
Akihiro Tsukioka is part of the Design Platform Technology Division at Sony Semiconductor Solutions Corporation and can be reached at akihiro.tsukioka@sony.com.
Yasuhiro Ochiai works within the Design Platform Technology Division at Sony Semiconductor Solutions Corporation and is available at
yasuhiro.ochiai@sony.com.
yasuhiro.ochiai@sony.com.