A Neural Network TensorFlow Framework for PA Distortion Analysis and Mitigation

2025-10-29

We propose a user-friendly neural network framework on the open-source TensorFlow platform to analyze and mitigate power amplifier distortion. Using simulation data of a 2 W GaN power amplifier design in the 7-8 GHz band, we demonstrate how an Artificial Neural Network (ANN) can be trained to extract AM-AM, AM-PM profiles for the analysis of PA nonlinearities. Furthermore, the ANN can be utilized to mitigate PA distortion through post-compensating the output signal or predistorting the input signal. Despite its simple structure and the lightweight computation, the ANN model provides ~12 dB of ACPR improvement (at 25 dBm average output power) through output post-compensation, as well as ~6 dB of ACPR improvement (at 22 dBm) through input pre-distortion, on a test envelope signal with 100 MHz bandwidth and 11 dB PAR.

Shielded Relay Coil design to Optimize WPT and SAR for Distributed Wireless Brain Implants ^*

2025-07-14

This paper presents a shielded relay antenna to simultaneously enhance Wireless Power Transfer (WPT) and reduce Specific Absorption Rate (SAR) for a network of distributed brain microimplants. Through strategic placement of conductive features, Eddy currents are created to oppose high magnetic fields. This design advantageously equalizes and increases the field strength over the cortical surface area. This work has the potential to address the WPT/ SAR co-optimization challenges for biomedical implants in general. When applied to the target 2 × 2 cm<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> wireless brain-machine interface (BMI) system operating at 915 MHz, HFSS simulations show it provides 1.2 dB WPT enhancement and a 29% SAR reduction.

Author Correction: Patterned electrical brain stimulation by a wireless network of implantable microdevices

Nature Communications · 2025-01-09

Metamaterial Antenna Design to Enhance Near Field Inductive Coupling for Biomedical Implants

2025-01-07

This paper presents a concept for a double negative metamaterial (DNM)-based antenna to simultaneously enhance Wireless Power Transfer (WPT) and reduce Specific Absorption Rate (SAR) here for a network of distributed brain microim-plants. The DNM copper coils are integrated in a FR-4 substrate, which has a dielectric constant of 4.3 and tangent loss <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(\delta)$</tex> of 0.025. Occupying a <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$2 \times 2\text{cm}^{2}$</tex> area, the DNM structure is introduced into our target wireless brain-machine interface (BMI) system operating at 915 MHz. Preliminary HFSS simulations show it provides 2 dB WPT enhancement and a 20% SAR reduction. We believe the work has the potential to address the WPT/ SAR co-optimization challenges for biomedical implants in general.

Sub-Centimeter Probes for Low-Frequency Alternating Electric Field (AEF) Measurements in Rodent Models of Brain Cancer

2025-11-03

Analysis of Wireless Power Transfer Efficiency and Specific Absorption Rate for a Distributed Brain Implant System

2024-10-15 · 1 citations

Wireless sub-mm sized distributed brain implants have been proposed as the next frontier of Brain-Machine Interface (BMI) design to achieve untethered, high-density neural recording and stimulation. Simultaneously improving the wireless power transfer (WPT) efficiency and reducing the specific absorption rate (SAR) will be crucial for its clinical success. Towards these goals, we present an EM simulation method, a lumped equivalent circuit model, and a theoretical analysis to accurately predict the power delivered to the recording/ stimulating nodes, as well as the power dissipated in biological tissues and all other lossy elements within the system. This comprehensive framework also explains how increasing the distance between the transmit coil and the scalp can beneficially reduce the SAR without undermining the WPT efficiency. This work presents a rigorous prediction technique for transmission loss and tissue heating towards performance optimization.

Coil Orientation and Stimulation Threshold in Transcranial Magnetic Stimulation (TMS)

2024-07-15

Transcranial Magnetic Stimulation (TMS) is used to treat mental disorders and explore brain function via applied electromagnetic fields generated by a high current coil placed on the subject's scalp. While TMS has been in clinical use for decades, and continues to be a rapidly expanding therapeutic modality, there are still many unknowns regarding how stimulation parameters may affect treatment efficacy and how they may be optimized. One key parameter that is readily accessible is coil orientation and its effects on TMS stimulation threshold. In this work, a multi-scale modeling approach addressed the effects of coil orientation on the TMS stimulation threshold for neuronal activation with several TMS coil current pulse shapes and widths. The modeling tool that was used incorporated an anatomically realistic macro scale model of the human brain cortex, and a micro scale neuronal model based on a representative layer 5 pyramidal brain cell. Simulations were performed for the left primary motor cortical area. Coil orientations associated with a minimum stimulation threshold for various current pulses were identified and then validated using data collected and published by another research team. This multi-scale modeling approach is versatile and can potentially be used for various current pulses and TMS coil locations to predict and map optimum TMS coil orientations for a wide range of treatment applications.

Design and Validation of Miniaturized Repetitive Transcranial Magnetic Stimulation (rTMS) Head Coils

Sensors · 2024-02-29 · 6 citations

Repetitive transcranial magnetic stimulation (rTMS) is a rapidly developing therapeutic modality for the safe and effective treatment of neuropsychiatric disorders. However, clinical rTMS driving systems and head coils are large, heavy, and expensive, so miniaturized, affordable rTMS devices may facilitate treatment access for patients at home, in underserved areas, in field and mobile hospitals, on ships and submarines, and in space. The central component of a portable rTMS system is a miniaturized, lightweight coil. Such a coil, when mated to lightweight driving circuits, must be able to induce B and E fields of sufficient intensity for medical use. This paper newly identifies and validates salient theoretical considerations specific to the dimensional scaling and miniaturization of coil geometries, particularly figure-8 coils, and delineates novel, key design criteria. In this context, the essential requirement of matching coil inductance with the characteristic resistance of the driver switches is highlighted. Computer simulations predicted E- and B-fields which were validated via benchtop experiments. Using a miniaturized coil with dimensions of 76 mm × 38 mm and weighing only 12.6 g, the peak E-field was 87 V/m at a distance of 1.5 cm. Practical considerations limited the maximum voltage and current to 350 V and 3.1 kA, respectively; nonetheless, this peak E-field value was well within the intensity range, 60-120 V/m, generally held to be therapeutically relevant. The presented parameters and results delineate coil and circuit guidelines for a future miniaturized, power-scalable rTMS system able to generate pulsed E-fields of sufficient amplitude for potential clinical use.

Patterned electrical brain stimulation by a wireless network of implantable microdevices

Nature Communications · 2024-11-21 · 14 citations

Transmitting meaningful information into brain circuits by electronic means is a challenge facing brain-computer interfaces. A key goal is to find an approach to inject spatially structured local current stimuli across swaths of sensory areas of the cortex. Here, we introduce a wireless approach to multipoint patterned electrical microstimulation by a spatially distributed epicortically implanted network of silicon microchips to target specific areas of the cortex. Each sub-millimeter-sized microchip harvests energy from an external radio-frequency source and converts this into biphasic current injected focally into tissue by a pair of integrated microwires. The amplitude, period, and repetition rate of injected current from each chip are controlled across the implant network by implementing a pre-scheduled, collision-free bitmap wireless communication protocol featuring sub-millisecond latency. As a proof-of-concept technology demonstration, a network of 30 wireless stimulators was chronically implanted into motor and sensory areas of the cortex in a freely moving rat for three months. We explored the effects of patterned intracortical electrical stimulation on trained animal behavior at average RF powers well below regulatory safety limits. Transmitting information directly into the brain is a challenge for future brain-computer interfaces. Here, the authors present a patterned electrical microstimulation protocol using an epicortically-implanted network of silicon microchips to target specific areas of the cortex.

An All-Digital Synthesizer Enabled by a Convolutional Neural Network

2024-06-16

An “All-Digital” synthesizer is proposed comprised of an inverter-based vector modulator and a Convolutional Neural Network (CNN) controller - all integrable within a digital CMOS process. The modulator is broadband tunable (GHz's of bandwidth), broad phase range (≈ 100°), and high resolution (< 0.1°) enabling precision tracking. Each channel of the synthesizer has an expansive control space, in the order of 2<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">28</sup> states, and exhibits a highly non-linear response to the control code. A vector mapping encoding is proposed to characterize the control state-space by a CNN producing a 1.44° RMS prediction error with only 0.01 % of the input state-space as a training set. A four-channel modulator was fabricated in a 12 nm FinFet CMOS process achieving an operating range of 4 GHz to 8 GHz, an Image Rejection Ratio (IRR) of 58 dBc at 8 GHz, a power consumption of 50 mW, and an active area of 280 um × 180 um.