arXiv subjects

Satish Mulleti

Publications and source records attributed to Satish Mulleti.

At least 19 recordsLinked to original sources

arXiv2026

Fold First, Detect Directly: Communication Symbol Detection Without Unfolding for Low-Bitrate Modulo-ADCs

Modulo-folding ADCs reduce power consumption by restricting the dynamic range of the sampled signal before quantization, at the cost of an unfolding step needed to recover the true samples before any further processing. We show that this unfolding step can be skipped entirely for symbol detection, even in a realistic, oversampled setting where additive noise is present before the modulo operator and becomes correlated as it passes through the receiver's front-end filter, and separate quantization noise is further introduced due to the ADC. We show that a specific residual, formed from the folded, quantized observations and a candidate symbol hypothesis, exactly cancels the unknown integer wrap introduced by folding, so that the likelihood of a hypothesis is the density of the folded noise evaluated at that residual. Starting from this exact likelihood an intractable lattice sum over all integer wrap vectors, we show that the wrap vector is ternary and sparse with high probability whenever the folding threshold exceeds the noise standard deviation, so that at a threshold-to-noise ratio of three or more the sum is well approximated by a single Gaussian term. The resulting Mahalanobis maximum-likelihood detector works directly on the folded, quantized samples, and a block-structured search keeps detection tractable for long symbol sequences. Simulations confirm that in this regime our detector tracks the accuracy of a conventional, non-folding ADC closely across a wide range of SNRs, while the unfolding-based baselines need substantially higher oversampling to reach comparable accuracy.

eess.SP↗

arXiv2026

Tunable Domain Adaptation Using Unfolding

Machine learning models often struggle to generalize across domains with varying data distributions, such as differing noise levels, leading to degraded performance. Traditional strategies like personalized training, which trains separate models per domain, and joint training, which uses a single model for all domains, have significant limitations in flexibility and effectiveness. To address this, we propose two novel domain adaptation methods for regression tasks based on interpretable unrolled networks--deep architectures inspired by iterative optimization algorithms. These models leverage the functional dependence of select tunable parameters on domain variables, enabling controlled adaptation during inference. Our methods include Parametric Tunable-Domain Adaptation (P-TDA), which uses known domain parameters for dynamic tuning, and Data-Driven Tunable-Domain Adaptation (DD-TDA), which infers domain adaptation directly from input data. We validate our approach on compressed sensing problems involving noise-adaptive sparse signal recovery, domain-adaptive gain calibration, and domain-adaptive phase retrieval, demonstrating improved or comparable performance to domain-specific models while surpassing joint training baselines. This work highlights the potential of unrolled networks for effective, interpretable domain adaptation in regression settings.

cs.LG↗

arXiv2026

Adaptive Non-Uniform Sampling of Bandlimited Signals via Algorithm-Encoder Co-Design

We propose an adaptive non-uniform sampling framework for bandlimited signals based on an algorithm-encoder co-design perspective. By revisiting the convergence analysis of iterative reconstruction algorithms for non-uniform measurements, we derive a local, energy-based sufficient condition that governs reconstruction behavior as a function of the signal and derivative energies within each sampling interval. Unlike classical approaches that impose a global Nyquist-type bound on the inter-sample spacing, the proposed condition permits large gaps in slowly varying regions while enforcing denser sampling only where the signal exhibits rapid temporal variation. Building on this theoretical insight, we design a variable-bias, variable-threshold integrate-and-fire time encoding machine (VBT-IF-TEM) whose firing mechanism is explicitly shaped to enforce the derived local convergence condition. To ensure robustness, a shifted-signal formulation is introduced to suppress excessive firing in regions where the magnitude of the signal amplitude is close to zero or the local signal energy approaches zero. Using the proposed encoder, an analog signal is discretely represented by time encodings and signal averages, enabling perfect reconstruction via a standard iterative algorithm even when the local sampling rate falls below the Nyquist rate. Simulation results on synthetic signals and experiments on ultrasonic guided-wave and ECG signals demonstrate that the proposed framework achieves substantial reductions in sampling density compared to uniform sampling and conventional IF-TEMs, while maintaining accurate reconstruction. The results further highlight a controllable tradeoff between sampling density, reconstruction accuracy, and convergence behavior, which can be navigated through adaptive parameter selection.

eess.SP↗

arXiv2025

Verifiable Deep Quantitative Group Testing

We present a neural network-based framework for solving the quantitative group testing (QGT) problem that achieves both high decoding accuracy and structural verifiability. In QGT, the objective is to identify a small subset of defective items among $N$ candidates using only $M \ll N$ pooled tests, each reporting the number of defectives in the tested subset. We train a multi-layer perceptron to map noisy measurement vectors to binary defect indicators, achieving accurate and robust recovery even under sparse, bounded perturbations. Beyond accuracy, we show that the trained network implicitly learns the underlying pooling structure that links items to tests, allowing this structure to be recovered directly from the network's Jacobian. This indicates that the model does not merely memorize training patterns but internalizes the true combinatorial relationships governing QGT. Our findings reveal that standard feedforward architectures can learn verifiable inverse mappings in structured combinatorial recovery problems.

eess.SP↗

arXiv2025

Two-Dimensional Tomographic Reconstruction From Projections With Unknown Angles and Unknown Spatial Shifts

In parallel beam computed tomography (CT), an object is reconstructed from a series of projections taken at different angles. However, in some industrial and biomedical imaging applications, the projection geometry is unknown, completely or partially. In this paper, we present a technique for two-dimensional (2D) tomography in which both viewing angles and spatial shifts associated with the projections are unknown. There exists literature on 2D unknown view tomography (UVT), but most existing 2D UVT algorithms assume that the projections are centered; that is, there are no spatial shifts in the projections. To tackle these geometric ambiguities, we first modify an existing graph Laplacian-based algorithm for 2D UVT to incorporate spatial shifts, and then use it as the initialization for the proposed three-way alternating minimization algorithm that jointly estimates the 2D structure, its projection angles, and the corresponding shifts. We evaluate our method on noisy projections of ribosome images and demonstrate that it achieves superior reconstruction compared to the baseline that neglects shifts.

eess.IV↗

arXiv2025

Linear-Bias Time Encoding for Low-Rate Quantized Representation of Bandlimited Signals

Integrate-and-fire time encoding machines (IF-TEMs) provide an efficient framework for asynchronous sampling of bandlimited signals through discrete firing times. However, conventional IF-TEMs often exhibit excessive oversampling, leading to inefficient encoding for signals with smoothly distributed information. This letter introduces a linear-bias IF-TEM (LB-IF-TEM), where the bias dynamically tracks the input signal to maintain a nearly constant integrator input, thereby localizing the firing intervals. The resulting concentrated distribution enables effective non-uniform quantization with reduced distortion. Theoretical analysis establishes explicit bounds on the achievable oversampling range, while experimental results demonstrate that the proposed method attains comparable reconstruction accuracy at significantly lower bitrate than existing IF-TEM variants. The LB-IF-TEM thus provides a low-power, communication-efficient, and analytically tractable framework for time-based signal encoding and reconstruction.

eess.SP↗

arXiv2025

A Non-Uniform Quantization Framework for Time-Encoding Machines

Time encoding machines (TEMs) provide an event-driven alternative to classical uniform sampling, enabling power-efficient representations without a global clock. While prior work analyzed uniform quantization (UQ) of firing intervals, we show that these intervals are inherently non-uniformly distributed, motivating the use of non-uniform quantization (NUQ). We derive the probability distribution of firing intervals for a class of bandlimited signals and design a power-law-based NUQ scheme tailored to this distribution. Simulations demonstrate that NUQ significantly outperforms UQ under the same bit budget. We also compare TEMs with non-uniform sampling (NUS), where both amplitudes and timings require quantization, and show that TEM--NUQ achieves lower error at half the transmission cost. These results highlight the advantages of distribution-aware quantization and establish TEM--NUQ as an efficient alternative to conventional UQ and NUS schemes.

eess.SP↗

arXiv2025

Dictionary-Based Deblurring for Unpaired Data

Effective image deblurring typically relies on large and fully paired datasets of blurred and corresponding sharp images. However, obtaining such accurately aligned data in the real world poses a number of difficulties, limiting the effectiveness and generalizability of existing deblurring methods. To address this scarcity of data dependency, we present a novel dictionary learning based deblurring approach for jointly estimating a structured blur matrix and a high resolution image dictionary. This framework enables robust image deblurring across different degrees of data supervision. Our method is thoroughly evaluated across three distinct experimental settings: (i) full supervision involving paired data with explicit correspondence, (ii) partial supervision employing unpaired data with implicit relationships, and (iii) unsupervised learning using non-correspondence data where direct pairings are absent. Extensive experimental validation, performed on synthetically blurred subsets of the CMU-Cornell iCoseg dataset and the real-world FocusPath dataset, consistently shows that the proposed framework has superior performance compared to conventional coupled dictionary learning approaches. The results validate that our approach provides an efficient and robust solution for image deblurring in data-constrained scenarios by enabling accurate blur modeling and adaptive dictionary representation with a notably smaller number of training samples.

eess.IV↗

arXiv2025

Learnable Kernels for FRI -- Joint Kernel Encoder Optimization and Hardware Validation

Finite Rate of Innovation (FRI) sampling techniques provide efficient frameworks for reconstructing signals with inherent sparsity at rates below Nyquist. However, traditional FRI reconstruction methods rely heavily on pre-defined kernels, often limiting hardware implementation and reconstruction accuracy under noisy conditions. In this paper, we propose a robust, flexible, and practically implementable framework for FRI reconstruction by introducing novel learnable kernel strategies. First, we demonstrate effective reconstruction using known, fixed kernels such as truncated Gaussian and Gaussian pair kernels, which mitigate the requirement that the samples should have a sum-of-exponentials (SoE) form. Next, we extend this concept by jointly optimizing both the sampling kernel and reconstruction encoder through a unified learning approach, yielding adaptive kernels that significantly outperform traditional methods in resolution and noise robustness, with reduced sampling rates. Furthermore, we propose a practical hardware realization by representing kernels as sums of two exponential decay signals with jointly optimized poles, facilitating compact, efficient analog implementations. Our approach is validated experimentally through hardware implementations using a unity-gain Sallen-Key analog filter, achieving accurate real-world signal recovery. The developed convolutional neural network-based encoder substantially reduces computational complexity, demonstrating competitive performance with fewer parameters, making our method particularly suitable for resource-constrained, edge-based deployments.

eess.SP↗

arXiv2025

Self-Calibrating Integrate-and-Fire Time Encoding Machine

In this paper, we introduce a novel self-calibrating integrate-and-fire time encoding machine (S-IF-TEM) that enables simultaneous parameter estimation and signal reconstruction during sampling, thereby effectively mitigating mismatch effects. The proposed framework is developed over a new practical IF-TEM (P-IF-TEM) setting, which extends classical models by incorporating device mismatches and imperfections that can otherwise lead to significant reconstruction errors. Unlike existing IF-TEM settings, P-IF-TEM accounts for scenarios where (i) system parameters are inaccurately known and may vary over time, (ii) the integrator discharge time after firings can vary, and (iii) the sampler may operate in its nonlinear region under large input dynamic ranges. For this practical model, we derive sampling rate bounds and reconstruction conditions that ensure perfect recovery. Analytical results establish the conditions for perfect reconstruction under self-calibration, and evaluation studies demonstrate substantial improvements - exceeding 59dB - highlighting the effectiveness of the proposed approach.

eess.SP↗

arXiv2024

Compressed Sensing Based Residual Recovery Algorithms and Hardware for Modulo Sampling

Analog-to-Digital Converters (ADCs) are essential components in modern data acquisition systems. A key design challenge is accommodating high dynamic range (DR) input signals without clipping. Existing solutions, such as oversampling, automatic gain control (AGC), and compander-based methods, have limitations in handling high-DR signals. Recently, the Unlimited Sampling Framework (USF) has emerged as a promising alternative. It uses a non-linear modulo operator to map high-DR signals within the ADC range. Existing recovery algorithms, such as higher-order differences (HODs), prediction-based methods, and beyond bandwidth residual recovery (B2R2), have shown potential but are either noise-sensitive, require high sampling rates, or are computationally intensive. To address these challenges, we propose LASSO-B2R2, a fast and robust recovery algorithm. Specifically, we demonstrate that the first-order difference of the residual (the difference between the folded and original samples) is sparse, and we derive an upper bound on its sparsity. This insight allows us to formulate the recovery as a sparse signal reconstruction problem using the least absolute shrinkage and selection operator (LASSO). Numerical simulations show that LASSO-B2R2 outperforms prior methods in terms of speed and robustness, though it requires a higher sampling rate at lower DR. To overcome this, we introduce the bits distribution mechanism, which allocates 1 bit from the total bit budget to identify modulo folding events. This reduces the recovery problem to a simple pseudo-inverse computation, significantly enhancing computational efficiency. Finally, we validate our approach through numerical simulations and a hardware prototype that captures 1-bit folding information, demonstrating its practical feasibility.

eess.SP↗

arXiv2024

Super-Resolution via Learned Predictor

Frequency estimation from measurements corrupted by noise is a fundamental challenge across numerous engineering and scientific fields. Among the pivotal factors shaping the resolution capacity of any frequency estimation technique are noise levels and measurement count. Often constrained by practical limitations, the number of measurements tends to be limited. This work introduces a learning-driven approach focused on predicting forthcoming measurements based on available samples. Subsequently, we demonstrate that we can attain high-resolution frequency estimates by combining provided and predicted measurements. In particular, our findings indicate that using just one-third of the total measurements, the method achieves a performance akin to that obtained with the complete set. Unlike existing learning-based frequency estimators, our approach's output retains full interpretability. This work holds promise for developing energy-efficient systems with reduced sampling requirements, which will benefit various applications.

eess.SP↗

arXiv2024

Blind-Adaptive Quantizers

Sampling and quantization are crucial in digital signal processing, but quantization introduces errors, particularly due to distribution mismatch between input signals and quantizers. Existing methods to reduce this error require precise knowledge of the input's distribution, which is often unavailable. To address this, we propose a blind and adaptive method that minimizes distribution mismatch without prior knowledge of the input distribution. Our approach uses a nonlinear transformation with amplification and modulo-folding, followed by a uniform quantizer. Theoretical analysis shows that sufficient amplification makes the output distribution of modulo-folding nearly uniform, reducing mismatch across various distributions, including Gaussian, exponential, and uniform. To recover the true quantized samples, we suggest using existing unfolding techniques, which, despite requiring significant oversampling, effectively reduce mismatch and quantization error, offering a favorable trade-off similar to predictive coding strategies.

eess.SP↗

arXiv2024

Subsampling of Correlated Graph Signals

Graph signals are functions of the underlying graph. When the edge-weight between a pair of nodes is high, the corresponding signals generally have a higher correlation. As a result, the signals can be represented in terms of a graph-based generative model. The question then arises whether measurements can be obtained on a few nodes and whether the correlation structure between the signals can be used to reconstruct the graph signal on the remaining nodes. We show that node subsampling is always possible for graph signals obtained through a generative model. Further, a method to determine the number of nodes to select is proposed based on the tolerable error. A correlation-based fast greedy algorithm is developed for selecting the nodes. Finally, we verify the proposed method on different deterministic and random graphs, and show that near-perfect reconstruction is possible with node subsampling.

eess.SP↗

arXiv2023

Power-Efficient Sampling

Analog-to-digital converters (ADCs) facilitate the conversion of analog signals into a digital format. While the specific designs and settings of ADCs can vary depending on their applications, it is crucial in many modern applications to minimize their power consumption. The significance of low-power ADCs is particularly evident in fields like mobile and handheld devices reliant on battery operation. Key parameters of the ADCs that dictate the ADC's power are its sampling rate, dynamic range, and number of quantization bits. Typically, these parameters are required to be higher than a threshold value but can be reduced by using the structure of the signal and by leveraging preprocessing and the system application needs. In this review, we discuss four approaches relevant to a variety of applications.

eess.SP↗

arXiv2023

Modulation For Modulo: A Sampling-Efficient High-Dynamic Range ADC

In high-dynamic range (HDR) analog-to-digital converters (ADCs), having many quantization bits minimizes quantization errors but results in high bit rates, limiting their application scope. A strategy combining modulo-folding with a low-DR ADC can create an efficient HDR-ADC with fewer bits. However, this typically demands oversampling, increasing the overall bit rate. An alternative method using phase modulation (PM) achieves HDR-ADC functionality by modulating the phase of a carrier signal with the analog input. This allows a low-DR ADC with fewer bits. We've derived identifiability results enabling reconstruction of the original signal from PM samples acquired at the Nyquist rate, adaptable to various signals and non-uniform sampling. Using discrete phase demodulation algorithms for practical implementation, our PM-based approach doesn't require oversampling in noise-free conditions, contrasting with modulo-based ADCs. With noise, our PM-based HDR method demonstrates efficiency with lower reconstruction errors and reduced sampling rates. Our hardware prototype illustrates reconstructing signals ten times greater than the ADC's DR from Nyquist rate samples, potentially replacing high-bit rate HDR-ADCs while meeting existing bit rate needs.

eess.SP↗

arXiv2023

Unlabelled Sensing with Priors: Algorithm and Bounds

In this study, we consider a variant of unlabelled sensing where the measurements are sparsely permuted, and additionally, a few correspondences are known. We present an estimator to solve for the unknown vector. We derive a theoretical upper bound on the $\ell_2$ reconstruction error of the unknown vector. Through numerical experiments, we demonstrate that the additional known correspondences result in a significant improvement in the reconstruction error. Additionally, we compare our estimator with the classical robust regression estimator and we find that our method outperforms it on the normalized reconstruction error metric by up to $20\%$ in the high permutation regimes $(>30\%)$. Lastly, we showcase the practical utility of our framework on a non-rigid motion estimation problem. We show that using a few manually annotated points along point pairs with the key-point (SIFT-based) descriptor pairs with unknown or incorrectly known correspondences can improve motion estimation.

eess.SP↗

arXiv2023

Greedy Selection for Heterogeneous Sensors

Simultaneous operation of all sensors in a large-scale sensor network is power-consuming and computationally expensive. Hence, it is desirable to select fewer sensors. A greedy algorithm is widely used for sensor selection in homogeneous networks with a theoretical worst-case performance of (1-1/e) ~ 63% of the optimal performance when optimizing submodular metrics. For heterogeneous sensor networks (HSNs) comprising multiple sets of sensors, most of the existing sensor selection methods optimize the performance constrained by a budget on the total value of the selected sensors. However, in many applications, the number of sensors to select from each set is known apriori, and solutions are not well-explored. For this problem, we propose a joint greedy heterogeneous sensor selection algorithm. Theoretically, we show that the worst-case performance of the proposed algorithm is bounded to 50% of the optimum for submodular cost metrics. In the special case of HSNs with two sensor networks, the performance guarantee can be improved to 63% when the number of sensors to select from one set is much smaller than the other. To validate our results experimentally, we propose a submodular metric based on the frame potential measure that considers both the correlation among the sensor measurements and their heterogeneity. We prove theoretical bounds for the mean squared error of the solution when this performance metric is used. We validate our results through simulation experiments considering both linear and non-linear measurement models corrupted by additive noise and quantization errors. Our experiments show that the proposed algorithm results in 4-10 dB lower error than existing methods.

eess.SP↗