A compact V-band single-chip receiver compatible to flip-chip technology is demonstrated using 0.15 mu m mHEMT (metamorphic high electron mobility transistor) technology. The image rejection receiver consists of a three-stage low-noise amplifier, a 3xLO multiplier chain and an image rejection subharmonic downconverter. Only the input match network of the first-stage LNA is utilized with the coplanar waveguide with backside ground (CPWG) for compatibility with the flip-chip technology and the rest of matching networks are utilized with the microstrip (MS) transmission lines for a small size. mHEMT devices with through-substrate vias are employed as CPWG to MS transition. The flip-chip mHEMT V-band receiver with the chip size of 4.3 mm x 2.4 mm achieves conversion gain of 4.8 dB and noise figure is 7.7 dB at 59 GHz. The best image rejection ratio of 28 dB occurs at 60 GHz.
This paper presents a polyphase systolic array for implementing a 512-tap digital FIR filter on the Xilinx RF-SoC ZCU-111 platform using a matrix-vector multiplier core. The design is tailored for high-speed real-time applications and supports dynamic coefficient assignment. The proposed architecture is designed from scratch instead of using Xilinx-provided FIR cores to accommodate the configurability of high tap counts, parallel processing, adaptive coefficient updates in real-time RF applications, and portability for any FPGA or ASIC platforms. In addition, this custom design leverages multiplier symmetry and pipelining. Real-time operations across 8 parallel data paths processing at 256 MHz on RF-SoC digital fabric leading to 1 TeraMAC/s throughput. The digital architecture is verified using a low-pass filter and a bandpass filter. Thus, experimental results demonstrate successful realization of the custom designed FIR filter core in wide band signal processing scenarios in advanced wireless communication, RF AI/ML applications, spectrum sensing, electronic warfare (EW), and signal intelligence applications.
This paper presents a quasi-Yagi antenna direct antenna modulation (DAM) for wideband digitally controlled phase modulation. A compact transistor-based switching network, driven by two DC bias voltages, enables dynamic feed direction control, achieving 180 degrees far-field phase shifts over a wide frequency range. A 1D array of four DAM elements is fabricated and measured, showing strong agreement with simulations and effective beam steering across multiple digital states, within 1-GHz bandwidth covering the Wi-Fi 6E frequency range. The proposed quasi-Yagi DAM features low complexity and wideband digital reconfigurability, offering a promising solution for future 6G wireless communication and Internet-of-Things (IoT) applications.
The SpectrumX Mobile Experiment Platform (MEP) is an software radio instrument designed to support a wide range of spectrum experiments. It is capable of passive data acquisition as well as active transmission to enable research into signal propagation and waveform co-existence. The mobile instrument includes an analog front-end with calibration, positioning, and timing features, a field-programmable gate array (FPGA) radio, and an Artificial Intelligence (AI) compute platform for on-instrument data processing. Four synchronous receive channels and two transmit channels operate up to 6 GHz, with an optional wide-band tuner. The hardware, firmware, and software for the MEP are extensible and reproducible, with components available through a combination of off-the-shelf hardware and rapid prototyping fabrication. More than a dozen MEPs have been constructed through student workshops and are being used for ongoing field experiments. This paper details the design of the MEP, including component selection and initial testing.
Delicate components in the front ends of sensitive RF receivers are vulnerable to damage from high-powered signals. RF limiters are crucial in protecting components downstream of the receive antenna by attenuating high-amplitude inputs to safe levels while maintaining receiver sensitivity and dynamic range. This paper presents the characterization of eight limiters in passive, quasi-active, and active topologies. The gain, output power, and OIP3 of each limiter is measured with a -10 to 40 dBm input at 1 GHz. The results demonstrate that these designs and their cascade into multi-stage structures can provide a high-power tolerant and adjustable RF limiter with flexible performance required for receivers in integrated sensing and communications (ISAC) applications.
This paper presents a DC-120 GHz low-noise, single-stage distributed amplifier (DA) implemented in a 70-nm GaAs pHEMT process (PP07) with peak f(T)/f(max) of 160/360 GHz. The DA consists of seven cascode gain cells, each incorporating an inherently-grounded transconductance (g(m)) stage to achieve superior high-frequency performance. In addition, ribbon inductors are employed for series gain peaking to further enhance gain-bandwidth product (GBW). The DA achieves a broadband gain (S-21) of 11.9 dB with a GBW of 472 GHz while consuming 297 mW from a 3.3-V supply. The saturated output power (P-SAT) is 15.2 dBm and 10.3 dBm at 50 GHz and 80 GHz, respectively, with a noise figure (NF) <7 dB up to 75 GHz. The DA exhibits an average group delay (tau) of 25.3 ps with a group delay variation (Delta tau) of +/- 24.9 ps up to 120 GHz. The DA occupies a chip area of 1.33 mm x 0.575 mm (0.76 mm(2)), including the pads. To the best of the authors' knowledge, this DA achieves the highest GBW, along with excellent NF and output power, among all previously published GaAs and GaN HEMT-based DAs.
This paper presents an experimental comparison of classical machine learning (ML) and deep learning (DL) models for detecting and classifying drones based on radio frequency (RF) features. Although previous studies have explored individ-ual models or partial comparisons, inconsistent datasets have hindered direct performance evaluation between approaches. To enable a fair and comprehensive comparison, we collected a consistent RF dataset from three popular DJI drone models. Using this data set, we evaluated the performance of classical ML algorithms: Support Vector Machines (SVM), Random Forest, Gradient Boosting (XGBoost) and K-Nearest Neighbors (KNN); along with advanced DL architectures such as a shallow and deep Multi-Layer Perceptron (MLP). Our results demonstrate that DL models offer superior accuracy and robustness, whereas classical ML methods are more efficient in terms of training and inference time. This trade-off provides critical insight for selecting the appropriate models in practical drone detection and classification scenarios.
Wireless connectivity within vehicles is becoming increasingly attractive due to advances in communications technologies and user expectations for these technologies. Thus, there is a great need for models of wireless channels inside vehicles. In this paper, we present a novel method for deriving wireless channel models in highly reflective environments, with a particular focus on intra-vehicular communication at 2.4 GHz. In contrast to traditional grid-based channel measurements, the proposed approach employs a time-efficient area-based measurement technique combined with statistical analysis of received signal strength distributions. The Cumulative Distribution Functions (CDFs) of the conducted measurements with different channel bandwidths reveal significant variation in small-scale fading behavior. To model these CDFs accurately, distribution functions, and a custom function are fitted on the channel behavior. The resulting channel model allows for reduced measurement complexity while enabling fast and reliable characterization of wireless channels in dynamic and reflective environments such as vehicle interiors.
Blind source separation (BSS) algorithms have shown promise in isolating individual signal components from mixed observations without prior knowledge of the sources. Although extensive research has been conducted in simulated and controlled environments, the practical feasibility of BSS in real-world wireless systems, particularly in multiple-input multiple-output (MIMO) configurations with over-the-air (OTA) wave-form excitations, remains underexplored. This study provides a feasibility analysis of BSS performance under OTA conditions, taking into account factors, such as antenna coupling, self-interference, multipath propagation, and signal-to-noise ratio. OTA measurements were performed using representative wave-form scenarios from 4G, 5G, radar, satellite communications, and land mobile radio. The results indicate that BSS algorithms can achieve effective source separation under certain conditions, though performance may be sensitive to signal characteris-tics and environmental impairments. This work is intended to support ongoing research efforts aimed at developing robust signal processing frameworks for future wireless MIMO systems, especially in scenarios where channel state information is limited or unavailable.
To obtain a high data rate density (Gb/s/mm) in arrays of mm-wave dielectric waveguide (DWG) interconnects, the interconnect pitch must be small. Unfortunately, small pitch results in significant field coupling, hence crosstalk, between adjacent waveguides. To quantify the effect of coupling on interconnect density, we first determine crosstalk between two parallel unshielded DWGs and calculate the minimum pitch required for acceptable crosstalk (-30 dB) in 1-10 m links. We find that DWGs using alumina cores, with expanded PTFE (ePTFE) cladding, can tolerate smaller pitch at a given crosstalk than DWGs using PTFE cores. Then we analyze a DWG design where a metallic shield surrounding the e-PTFE cladding suppresses crosstalk; with appropriate cladding size, this shield only negligibly increases DWG attenuation. Constraining crosstalk to -30 dB, for a 10 m link, the shielded designs provide 14:1 greater interconnect density (guides/mm(2)) for PTFE-core DWGs and 8:1 greater interconnect density for alumina-core DWGs.
Electromagnetic metastructure-based interconnects offer a potential solution to parasitic effects, however, their integration into microelectronic packaging has been hindered by a fundamental trade-off between their physical size and operating frequency, leading to impractically large footprints for gigahertz operation. This work introduces a novel approach to overcome this challenge by vertically folding the metastructure into a multilayer, via-connected structure. Through comprehensive simulation and experimental validation, we demonstrate a tenfold footprint reduction to 575 mu m, making the size of proposed waveguide comparable to the pitch of state-of-the-art ball-grid-array packaging, while maintaining operation at 6 GHz, compatible to VLSI clock frequency. The optimized design achieves an 18 dB reduction in cross-talk noise compared to a buried microstrip line of the same footprint, effectively halving the cross-talk noise spikes as confirmed by time-domain analysis. This work validates a viable pathway for integrating high-performance, compact metastructure-based interconnects into commercial chip designs.
A unified RF rectifier-demodulator front end has been developed for wireless neural implants and other low-power wireless sensors operating in sub-GHz bands, including the 902-928MHz ISM band. The architecture integrates both blocks to enable efficient power harvesting and data demodulation, essential for neural stimulators, channel-selective neural recorders, and compact IoT or RFID nodes. Implemented in 180nm CMOS, the design employs a 4-stage fully differential cross-coupled differential-drive (CCDD) rectifier with a low-power DC demodulator supporting 1 MHz data rate. Including matching and packaging losses, a maximum power harvesting efficiency (PHE) of 32.2% is measured at 4-6dBm input, with efficiency above 20% for inputs over 0dBm. Although parasitic effects shifted the optimum frequency to 750MHz, functionality is preserved across 750-915MHz by tuning the off-chip matching network. These results demonstrate a compact, reconfigurable rectifier-demodulator front end suitable for energy-constrained neural interfaces and ISM-band IoT and RFID applications.
To meet the high-capacity and low-latency demands of 6G networks, small cell backhaul must be significantly enhanced. Orbital angular momentum (OAM) multiplexing enables multiple data streams over a single aperture pair and, when combined with MIMO and uniform circular arrays (UCAs), can improve spectral efficiency. However, OAM systems are vulnerable to inter-mode interference (IMI) caused by beam axis misalignment. This paper proposes an OAM-MIMO scheme that mitigates IMI by selecting among even-numbered, odd-numbered, and all mode groups based on channel state information (CSI). The scheme integrates stream separation using successive interference cancellation (SIC) and adaptive mode group selection according to IMI severity. Simulation results demonstrate improved capacity and robustness compared to fixed-mode approaches.
This work presents novel classes of dual-band filters comprising cascaded network configurations with all positively coupled resonators. The identification of these networks greatly simplifies the realization of the microwave filters and improves their reliability. The EM implementation of an X-Band dual-band waveguide filter is presented displaying close agreement with the circuit simulations.
This paper introduces novel oscillators for direct frequency synthesis at millimeter-wave frequencies. The first approach targeting ultra-low phase noise adopts a reduced filling factor electromagnetic bandgap (EBG) resonator fabricated from high-resistivity silicon (HRS). This was combined with a SiGe-BiCMOS chip comprising a two-stage loop amplifier, coupler, electronic phase shifter and an output amplifier. Using a phasedlocked loop (PLL) circuit incorporating a commercial sampling phase detector (SPD) and a low-noise reference source operating at 520 MHz, phase noise values of -81, -98 and -119 dBc/Hz were measured at offset frequencies of 0.1, 1 and 10 kHz, respectively, at a carrier frequency of 45.8 GHz. Additionally, we present approaches to improve the temperature drift of the oscillator by using EBG-resonators fabricated from temperature-stabilized microwave ceramics. The impact on tuning range and oscillator phase noise are addressed.
This work presents a synthesis methodology for sensors based on nonlinear resonators. The new procedure enables high sensitivity near a specific value of the sensing parameter by tracing turning-point loci and identifying cusp bifurcations. As a key advantage, the method enables reconfigurability, since the sensitivity may be optimized around different target values. The sensor can also detect variations in the sensed parameter through discontinuous jumps observed when sweeping either the input amplitude or frequency. The concept has been validated using a prototype that was experimentally characterized.
We have been conducting research and development on a wireless transmission scheme using the 42-GHzband in millimeter waves, which enables large-capacity transmission with a wide frequency bandwidth, to realize the wireless transmission required for content production such as 3D and ultra-wide field-of-view video that provides a sense of presence and immersion. This paper reports a prototype modem that implements a multi-channel 2 x 4 MIMO SC-FDE system that is capable of 700-Mbps-class signal transmission and the results of evaluating its basic performance. By comparing the bit error rate characteristics of the AWGN channel between the computer simulation and prototype, it was confirmed that the prototype achieves the desired performance and provides a prospect for realizing 700-Mbps-class immersive video transmission using a 42-GHz band multi-channel 2 x 4 MIMO SC-FDE system.
This paper presents the experimental demonstration of a low-cost metasurface lens for near-field wireless power transfer (WPT) at 24 GHz. The lens is horn-fed and engineered to embed quantized spherical-phase control and Gaussian amplitude tapering directly into its physical structure. Both functionalities are realized through geometric variation in cross-shaped unit cells distributed across a 10 lambda x 10 lambda aperture. This dual-control strategy enhances near-field focusing by mitigating quantization artifacts and edge diffraction effects. Fabricated using standard PCB processes, the metasurface achieves efficient energy convergence without auxiliary feed shaping. Measurement results confirm a tightly confined beam sustained over a 7.6 lambda depth and a 5.8 dB enhancement in transmission coefficient |S-21| relative to free-space propagation. These findings validate the effectiveness of geometry-driven amplitude tapering in improving spatial energy localization for compact WPT systems.
This paper presents the concept, development, and validation of an integrated sensing module for Time Division Duplex (TDD) Radio Access Networks (RANs), exploring the guard period inherent in these systems. Our approach achieves a tight integration between the RAN and the sensing node using a lightweight energy-accumulation block that operates during this period. The design allows simultaneous communication and spectrum monitoring, supporting real-time Dynamic Spectrum Sharing (DSS) scenarios. The proposed sensing mechanism enables secondary TDD users to opportunistically access the spectrum without causing interference to primary users. A proof of concept was implemented using an open-source 5G system and a Software-Defined Radio (SDR) transceiver, with results showing a stable RAN performance in the absence of the primary users. This supports the feasibility of embedding sensing into communication systems for future DSS-oriented and Integrated Sensing and Communications (ISAC) applications.
This paper reports approaches for reducing the power consumption for a passive switching mixer first receiver that are demonstrated in a 200-GHz low-IF (3.75 GHz) BiCMOS receiver prototype. The receiver employs a 2(nd) order sub-harmonic mixer that allows the use of a 100-GHz LO signal generated by HBT's to reduce DC power consumption. Additionally, the receiver employs a single mixer instead of I/Q down converter to further lower DC power consumption by reducing the number of LO chains and eliminating the need for the generation of I/Q LO signals. The receiver fabricated in a GF BiCMOS process with 45-nm CMOS and HBT f(T)/f(max) of 410/610 GHz exhibits a 16-GHz bandwidth, a single side band noise figure of 23 dB, OP1dB of 0.5 dBm and an out-of-band gain roll-off of similar to 10 dB over 15 GHz at RF while consuming 96 mW not including that for an ADC.
