We propose RFLAM (Radio Frequency Large AI Model), a scalable generative AI framework that produces scene-conditioned, time-varying channel impulse response (CIR) estimates for mobile Massive MIMO (mMMIMO) Factory-of-the-Future (FoF) scenarios. Building on the SceneSense multimodal pipeline that provided proof-of-concept for static SISO CIR estimation, this positioning paper discusses design choices in scene-graph-conditioned CIR generation for (i) wideband, time-varying channels under realistic Doppler conditions, (ii) mMMIMO antenna arrays with up to 256 elements, and (iii) multi-user FoF environments with simultaneous mobile transceivers and scatterers. To systematically address scalability challenges with FoF mMMIMO, we propose and discuss divide-and-conquer strategies and designs of this framework, such as tensor decomposition and physics-informed partitioning of RFLAM submodules.

This paper investigates covert multi-hop communication in wireless networks where an adversary employs a cyclostationary (cycle) detector to reveal hidden transmissions. The covert route employs direct sequence spread spectrum (DSSS) signaling to ensure either maximum end-to-end covertness maximization or minimum latency minimization-under quality-of-service (QoS) and link budget constraints. Optimal bandwidth, transmit power, and spreading gain for each hop jointly satisfy reliability and either rate or covertness requirements. We show the equivalence between the covertness and the detection SNR gain-based widest-path formulations, and, hence, enabling efficient route computation. Numerical simulations in a realistic 3D environment illustrate that (i) end-to-end latency increases exponentially with the covertness requirement, (ii) the end-to-end latency increase is super-linear with the packet size M, and (iii) cycle and energy detectors impose different latency behavior as a function of the message length and the covertness requirement. The proposed framework provides important insights into resource allocation and routing design for covert networks against advanced detection adversaries.

Accurate and low-latency radio-frequency (RF) channel estimation is a fundamental requirement for next-generation (NextG) wireless systems. Traditional physics-based methods, such as ray tracing in NVIDIA Sionna, are computationally prohibitive for real-time and iterative use cases. We present SceneSense, a multimodal generative AI pipeline that learns the joint probability distribution between physical radio scenes and their corresponding channel impulse responses (CIRs). The trained pipeline ingests multimodal scene descriptions – comprising LiDAR point clouds, XML-annotated 3D scene graphs, and transmitter-receiver configurations – and produces CIR predictions at dramatically reduced latency. A graph neural network (e.g., Pytorch GINEconv or GPS) encodes rich scene representations into compact embeddings, which are aligned with Sionna-generated CIRs via contrastive learning. A downstream generative model then synthesizes CIR samples conditioned on the scene embedding. Our proof-of-concept demonstrates end-to-end latency of 9 ms versus 14 ms for Sionna ray tracing, with a clear path to 2ms by replacing the diffusion inference head with an autoregressive transformer.

Signal transmissions employ widely varying bandwidths, frequency carriers, modulation schemes, and may significantly overlap in frequency and time. Here we study blind strategies for characterizing transmitters operating in the same frequency band and same time in terms of their frequency, space, and time usage patterns. We achieve this using a combination of computer vision strategies based on spectrogram analysis and using source separation and pattern recognition strategies employing higher order statistical and machine learning analysis strategies. Following transmission characterization, the transmitters can be reconfigured at speed to avoid interference and ensure efficient usage of spectrum. Simulated results (using channel response derived from actual over-the-air measure ments) demonstrate the ability of our methods to accurately determine the start/end times and power spectra of multiple contemporaneous transmitters employing non-linear modulation schemes (including Bluetooth and WLAN) with SNR below 0 dB at several sensors in the receiver network.

Dithering is a technique commonly used to improve the perceptual quality of lossy data compression. In this work, we analytically and experimentally justify the use of dithering for ASR input compression. We formalize an understanding of optimal ASR performance under lossy input compression and leverage this to propose a parametric dithering technique for a low-complexity speech compression pipeline. The method performs well at 1-bit resolution, showing a 25% relative CER improvement, while also demonstrating improvements of 32.4% and 33.5% at 2- and 3-bit resolution, respectively, with our second dither choice yielding a reduced data rate. The proposed codec is adaptable to meet performance targets or stay within entropy constraints.

We consider a three party wireless network- Alice is a legitimate transmitter, Bob is a legitimate receiver, and Willie is a malicious adversary. Alice's goal is to communicate with Bob with a certain minimum Quality of Service (QoS) while evading detection from Willie. Alice transmits a Direct Sequence Spread Spectrum (DSSS) signal where the processing gain and transmit power are selected with the aim of evading detection by spreading the bit energy over a larger bandwidth while maintaining required QoS. Concurrently, Willie tries to detect the legitimate transmission by using either an energy or a cycle detector. We study this scenario within an adversarial optimization framework that aims at achieving a robust max-min covertness under link performance constraints. The framework is common to both detectors and hence allows for a comparative performance analysis that identifies common key performance parameters. The adversarial signal-to-noise ratio (SNR) detection gain enables direct trading of the DSSS processing gain for the (squared) channel quality ratio when aiming to improve link covertness, regardless of which detector Willie uses. The DSSS processing gain and SNR gain (and, hence, covertness) are limited by the bit rate and bit error rate link requirements, respectively. While both detectors' performance is limited by the SNR at Willie, the cycle detectors benefit significantly more from longer observation time of the legitimate transmission.

Efficient all-digital post-correction of low-resolution analog-to-digital converters can be achieved by using Look-Up Tables (LUTs). The performance of a LUT can be optimized by incorporating a parametric model for the expected input signal, noise level, and interference signals. We evaluate three analytical estimators for integration with parametrized LUTs, especially with applications to low-resolution, non-linear, or wideband quantizers. We also propose several approximations to improve tractability of the estimation problem for Phase-Shift Keyed input signals and Linear Frequency Modulated interference signals. Simulated results validate the ability of our estimator to recover the instantaneous value of the desired input signal in real-time with a high degree of accuracy. This includes cancellation of harmonic distortion that aliases into the desired signal bandwidth from front-end saturation due to high-power out-of-band interference. Our estimators are shown to achieve a significant gain over conventional linear-filtering techniques while also being robust to changes in input parameters, non-linear quantizers, and time-variant interference sources. For a tone input quantized to 3 bits and estimated with a fixed 12-tap model order we achieve >10 dB improvement in Mean Square Error and >20 dBc improvement in Spurious-Free Dynamic Range.

We propose a framework for the design, optimization, and implementation of Look-Up Tables (LUTs) used to recover noisy, oversampled, quantized signals given a parametric input model. The LUTs emulate the spectral effects of pre-quantization dithering through an all-digital solution applied after quantization. This methodology decomposes the intractable LUT design problem into four distinct stages, each of which is addressed analytically using a model-driven approach without reliance on training. Three dithering methods are studied to improve spectral purity metrics. Two novel indexing schemes are proposed to limit the LUT memory overhead shown to compress the LUT size by over four orders of magnitude with marginal performance loss. The LUT design is tested with an oversampled noisy sinusoidal input quantized to 3 bits and shown to improve its Spurious-Free Dynamic Range (SFDR) by over 19 dBc with only 324 bytes of memory while maintaining the same 3-bit fixed-point precision at the digital output. This correction can be implemented using two-level combinational logic ensuring ultra-low latency and, hence, suitable for low-resolution wideband devices.

This paper explores entropy-controlled dithering techniques in audio compression, examining the application of standard and modified TPDFs, combined with noise shaping and entropy-controlled parameters, across various audio contexts, including pitch, loudness, rhythm, and instrumentation variations. Perceptual quality metrics such as VISQOL and STOI were used to evaluate performance. The results demonstrate that TPDF-based dithering consistently outperforms RPDF, particularly under optimal alpha conditions, while highlighting performance variability based on signal characteristics. These findings suggest the situational appropriateness of using various TPDF distributions. This work emphasizes the trade-off between entropy and perceptual fidelity, offering insights into the potential of entropy-controlled dithering as a foundation for enhanced audio compression algorithms. A practical implementation as a Digital Audio Workstation plugin introduces customizable dithering controls, laying the groundwork for future advancements in audio compression algorithms.

SIMO Blind Channel Estimation and the problem it is encompassed by, Multichannel Blind Deconvolution, have been the topic of numerous works over the past several decades. Many methods have been developed to estimate the unknown channels and/or the signal based on assumptions made about them. One class of methods are the Subspace methods, which transform the problem into identifying the subspace of a matrix formed from the received signal observations across all channels. However, these approaches depend on a number of identifiability conditions that are often not met in practice, such as knowledge of the channel order; they do not perform well when conditions are not close to being met. These methods typically apply a quadratic, fixed-energy constraint to the estimated channels. However, results from previous works have noted that applying a linear constraint appears to make the Subspace methods more robust in certain cases, particularly when the channel is overmodeled. In this work, we compare the performance and robustness of Subspace methods subjected to the traditional quadratic constraint, as well as various sets of linear constraints, using both measured wireless communications channels, and simulated channels composed of varying numbers of Rayleigh fading multi-path components. We find that this type of constraint can indeed improve the robustness to overmodeling, and propose a constraint that does not depend on knowledge of the delay of the strongest path in the channel.

Ensuring the security of data transmitted over wireless links is a critical concern across civilian and military applications. In contemporary times, the protection of data in wireless networks against interception by adversaries is commonly achieved through diverse cryptographic methods. Despite these efforts, conventional cryptographic security often falls short, particularly when facing intelligent adversaries who possess in-depth knowledge of the systems and techniques in use. In such circumstances, the necessity for low probability of detection (LPD) communication arises to prevent the initial detection of transmissions.

We study a three node code division multiple access (CDMA) network that maintains secrecy in the presence of a sophisticated adversary. It relies on a custom testbed of software-defined radios (SDRs). We evaluate the adversaries’ detection performance employing cyclostationarity and energy detection strategies while the two friendly CDMA links maintain the intended performance. We illustrate the tradeoff as a function of the detection error and link bit error as a function of the CDMA codeword length for the two detection schemes. It is observed that while energy detectors are sensitive to noise and other interference in the band that can degrade their performance with spread spectrum systems, the cyclostationary detectors are very sensitive to receiver imperfections.

In this article, we investigate covert communication in intelligent reflecting surface (IRS) assisted networks with a friendly jammer where the jammer radiates random power jamming signals to confuse Willie in detecting the existence of the communication between Alice and Bob. We propose a novel technique that jointly optimizes the transmission probability, transmit power at Alice, IRS reflection matrix, and the interval of jamming power with the aim of maximizing the expected achievable rate at Bob while maintaining the covertness of the communication. In addition, we analyze the impact of the number of IRS elements on the detection error probability (DEP) at Willie and study the trade-off between the number of IRS elements and the covertness of the communication. We show through numerical simulations that the developed technique achieves near-optimal performance with low computational complexity and the analysis on the DEP at Willie is accurate.

Autonomous teams of unmanned ground and air vehicles rely on networking and distributed processing to collaborate as they jointly localize, explore, map, and learn in sometimes difficult and adverse conditions. Co-designed intelligent wireless networks are needed for these autonomous mobile agents for applications including disaster response, logistics and transportation, supplementing cellular networks, and agricultural and environmental monitoring. In this paper we describe recent progress on wireless networking and distributed processing for autonomous systems using a low frequency portion of the electromagnetic spectrum, here defined as roughly 25 to 100 MHz with corresponding wavelengths of 3 to 12 meters. This research is motivated by the desire to support autonomous systems operating in dense and cluttered environments by harnessing low frequency propagation, where meters long wavelengths yield significantly reduced scattering and enhanced penetration of obstacles and structures. This differs considerably from higher frequency propagation, requiring different low frequency propagation models than those widely employed for other bands. Progress in use of low frequency for autonomous systems has resulted from combined advances in low frequency propagation modeling, networking, antennas and electromagnetics, geolocation, multi-antenna array distributed beamforming, and mobile collaborative processing. This article describes the breadth and the depth of interaction between areas, leading to new tools and methods, especially in physically complex indoor/outdoor, dense urban, and other challenging scenarios. We bring together key results, models, measurements, and experiments that describe the state of the art for new uses of low frequency spectrum for multi-agent autonomy.

Dithering reduces error correlation and improves spectral purity of quantizers by injecting shaped noise at their input prior to quantization. Despite its ability to improve the Spurious Free Dynamic Range (SFDR) and Total Harmonic Distortion (THD) of quantized signals, dithering can be impractical for implementation in analog-to-digital conversion systems due to its high analog-domain complexity. We propose a method to emulate the effects of dithering exclusively in the digital domain using a Look-Up Table (LUT) architecture, allowing efficient and low-cost performance improvement to existing quantizers through digital post-processing. We present analytical results for how to optimally design the LUT in both the Mean-Square and Maximum-Likelihood sense. When simulated, our proposed technique improves SFDR by 15 dBc and reduces THD by 20 dBc while using less than 1 kB of memory and maintaining the same fixed-point output resolution as the input quantized data.

Dithering is a technique that can improve human perception of low-resolution data by reducing quantization artifacts. We hypothesize that the perceptual prominence of quantization artifacts is proportional to the magnitude of the quantization error autocorrelation vector. Under this hypothesis we derive two parametric dither distributions that trade-off between minimizing mean square error and minimizing an upper bound on the quantization error autocorrelation vector magnitude in the ℓ 1 sense $\left( {{f_{{V_{1,\alpha }}}}(v) = \alpha {\Pi _{\alpha \Delta }}(v) + (1 - \alpha )\frac{1}{2}\left[ {\delta \left( {v - \frac{{\alpha \Delta }}{2}} \right) + \delta \left( {v + \frac{{\alpha \Delta }}{2}} \right)} \right]} \right)$ or ℓ 2 sense $\left( {{f_{{V_{2,\alpha }}}}(v) = {\Pi _{\alpha \Delta }}(v)} \right)$ where ${\Pi _a}(v) \triangleq \frac{1}{a}, - \frac{a}{2} \leq v \leq \frac{a}{2}$ and ∆ is the width of the quantization region. The application of these distortion-controlling dithers to an example low-rate image recompression problem (using Lena) reveals optimal performance with partial dithering (0 < α ∝ λ < 1) as per Fig. 1 while our novel ℓ 1 -optimized dither produces a new Pareto front for the quality-entropy trade-off shown in Fig. 2 .

In this paper, we experimentally investigate the performance of diversity-based Low Probability of Detection (LPD) wireless communication using a custom Software-Defined Radio (SDR) system. We illustrate that polarization diversity can dramatically reduce a key metric for LPD against feature detection schemes: the Degree of Cyclostationarity (DCS) of the received signal. This reduces the probability of signal detection at an adversarial receiver. Moreover, random sequence selection and time-dithering are both proven to be potent methods for reducing cyclic correlations and can combine to provide a 94.6% reduction in DCS at the receiver. To facilitate communication with asynchronous SDRs several post-correction measures are described which enable demodulation of the received data. These are tested and shown to be effective even in the absence of an explicit preamble, which is critical to maintain the LPD nature of the communication signal.

The vehicle-to-vehicle (V2V) distance ranging system is a critical component of transportation automation, including the railroad industry. Although the mainline rail traffic is strictly regulated by sophisticated train control and dispatching systems, self-propelled rail vehicles, such as maintenance-of-way equipment and light rail vehicles, still rely on the safety discretion of human operators. To support both safety protection and autonomous operations for these vehicles, a fine-grained V2V ranging system is needed. This paper presents our design of a decentralized, high-resolution, short distance V2V ranging system based on ultra-wideband (UWB) impulse radio (IR) technology, implemented by commercial UWB sensors and customized algorithms. A field-testing case study has been conducted to validate the system performance on full-sized selfpropelled rail vehicles. The test results quantified the concerned performance parameters of UWB technology for the chosen use case and proved both the operational concepts and the functional design in this study. Our methodology in algorithm design, fieldtesting, and analytical data metrics should provide implications for applying UWB to other evolving use cases of intelligent rail transportation. At the end of this paper, we evaluate the system capability using ROS simulation and conclude the lessons learned from the system design

In this paper, we develop an approach to exploit multiple disparate wireless communication technologies simultaneously to enhance covertness of a communication link. Specifically, given two available communication modalities between a pair of friendly nodes (Alice and Bob), the goal is to evade detection by an adversary (Willie) who is equipped with a radiometer covering the frequency bands of both modalities. We propose a joint detection threshold optimization technique from Willie's point of view. We also develop a joint transmit power optimization strategy for Alice to maximize covertness while meeting the throughput requirement at Bob. Through numerical simulations we show that the proposed scheme matches the performance of exhaustive search method while reducing the computational time by 98% and also improves the covertness by 56% compared to a naïve benchmark scheme.

This letter investigates reduction in the degree of cyclostationarity (DCS) of digital communication signals in the presence of polarization diversity. Reduction in DCS decreases the probability of detection of an intruder that searches for periodic patterns in the transmitted signal. We propose a DCS reducing scheme that exploits dual-polarization diversity when correlated wide sense stationary information streams with different rates and codes are transmitted over polarizations assuming a uni-polarized adversary. The effects of depolarization and data correlation across polarizations on both the DCS experienced by the adversary, and the resulting probability of detection are investigated.