About

Upamanyu Madhow is a Distinguished Professor in the Department of Electrical and Computer Engineering at UC Santa Barbara. His research interests include next-generation communication, sensing, and inference infrastructures centered around millimeter wave systems, signal processing algorithms, and robust machine learning. He is associated with the Wireless Communication and Sensornets Lab, where his work focuses on advancing communication technologies and sensing systems. His contact information includes a phone number, email, and office location at Harold Frank Hall, indicating his active engagement in research and academic activities within the university.

Research topics

• Artificial Intelligence
• Machine Learning
• Computer Science
• Telecommunications
• Materials science
• Real-time computing
• Statistics
• Electrical engineering
• Electronic engineering
• Mathematics
• Engineering

Selected publications

• Scaling Wideband Massive MIMO Radar via Beamspace Dimension Reduction

  ArXiv.org · 2025-08-15

  preprintOpen accessSenior author

  We present an architecture for scaling digital beamforming for wideband massive MIMO radar. Conventional spatial processing becomes computationally prohibitive as array size grows; for example, the computational complexity of MVDR beamforming scales as O(N^3) for an N-element array. In this paper, we show that energy concentration in beamspace provides the basis for drastic complexity reduction, with array scaling governed by the O(NlogN) complexity of the spatial FFT used for beamspace transformation. Specifically, we propose an architecture for windowed beamspace MVDR beamforming, parallelized across targets and subbands, and evaluate its efficacy for beamforming and interference suppression for government-supplied wideband radar data from the DARPA SOAP (Scalable On-Array Processing) program. We demonstrate that our approach achieves detection performance comparable to full-dimensional benchmarks while significantly reducing computational and training overhead, and provide insight into tradeoffs between beamspace window size and FFT resolution in balancing complexity, detection accuracy, and interference suppression.

  PublisherOA PDFDOI

• Single-Frame MIMO Radar Velocity Vector Estimation via Multi-Bounce Scattering

  IEEE Transactions on Computational Imaging · 2025-01-01 · 1 citations

  articleOpen access

  Radars are widely adopted for autonomous navigation and vehicular networking due to their robustness to weather conditions as compared to visible light cameras and lidars. However, radars currently struggle with differentiating static vs tangentially moving objects within a <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">single radar frame</i> since both yield the same Doppler along line-of-sight paths to the radar. Prior solutions deploy multiple radar or visible light camera modules to form a multi-“look” synthetic aperture for estimating the single-frame velocity vectors, to estimate tangential and radial velocity components of moving objects leading to higher system costs. In this paper, we propose to exploit multi-bounce scattering from secondary static objects in the environment, e.g., building pillars, walls, etc., to form an effective multi-“look” synthetic aperture for single-frame velocity vector estimation with a <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">single</i> multiple-input, multiple-output (MIMO) radar, thus reducing the overall system cost and removing the need for multi-module synchronization. We present a comprehensive theoretical and experiment evaluation of our scheme, demonstrating a <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$4.5 \times$</tex-math></inline-formula> reduction in the error for estimating moving objects' velocity vectors over comparable single-radar baselines.

  PublisherDOI

• Creating Spatial Degrees of Freedom for Long-Range LoS MIMO using Reflect-arrays

  2024-09-10 · 1 citations

  article

  A fundamental bottleneck for long-range Line-of-Sight (LoS) MIMO is transceiver form factor: for a given carrier frequency, the product of the transmit and receiver apertures required to sustain a given number of spatial degrees of freedom scales quadratically with the link distance. In this paper, we propose and investigate the feasibility of a novel approach for sidestepping this bottleneck by creating large effective apertures using multiple reflect-arrays (RAs) placed near the transceivers. The introduction of an RA between a transmitter and receiver results in an end-to-end gain scaling as $1/d_1^2d_2^2$, where d<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</inf> is the distance between the transmitter and RA and d<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> is the distance between the RA and the receiver. While this leads to poor scaling when d<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</inf> and d<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> are both comparable to the link range, this problem is alleviated in our proposed setting, in which RAs are deployed near the transmitter and receiver. By benchmarking the link budget of our system model against a SISO link using transceivers with comparable form factors, we provide analytical guidelines for choice of system parameters such as the required RA sizes and the distance of the RAs from the associated transceivers. Simulation results based on detailed modeling of the channel matrices validate our analytical framework. Two key findings are as follows: (a) in order to avoid excessive degradation in link budget relative to the SISO benchmark, we must deploy "large enough" RAs at "small enough" distances making near-field beam focusing between RAs and transceivers essential, (b) coarse (2-bit) quantization of RA phases suffices to implement the required combination of near-field focusing and long-range beamforming with relatively small performance loss. We illustrate our ideas for a 6.4 Gbps link at 1.5 km using 4-fold spatial multiplexing and 800 MHz of bandwidth operating at 28 GHz.

  PublisherDOI

• Zero-Shot Accurate mmWave Antenna Array Calibration in the Wild

  2024-12-04 · 2 citations

  articleOpen accessSenior author

  mmWave antenna array calibration is a necessary yet tedious and costly process in manufacturing to capture the non-idealities in phased arrays, in order to obtain codebooks for accurate and stable beam steering. Unfortunately, predefined codebooks provided by manufacturers to steer beams in a given set of directions do not support the arbitrary beam shapes required for various mmWave communication, sensing, and security applications. To create arbitrary beam patterns, one needs to first find the unknown calibration vector for the particular phased array in use. In this paper, we introduce EiCal, a novel zero-shot technique that leverages the beamforming codebook advertised by the manufacturer to extract the calibration vector at zero cost (i.e., with no additional measurements). The key idea is that the unknown desired calibration vector can be obtained via an appropriately designed eigen-decomposition of the given codebook. We experimentally demonstrate the efficacy of EiCal on a 60 GHz mmWave array for two scenarios: angle estimation using compressive pseudorandom beams, and simultaneous steering of beams and nulls. Our results also point to potential simplifications in the calibration process at the manufacturer.

  PublisherOA PDFDOI

• Scaling mmWave MU-MIMO: Information-Theoretic Guidance Using Real-World Data

  2024-10-27 · 1 citations

  articleSenior author

  Advances in millimeter wave (mmWave) radio frequency integrated circuits (RFICs) potentially enable the realization of one RF chain per antenna for arrays with hundreds of antenna elements, opening up the possibility of fully digital beamforming for truly massive multiuser (MU-)MIMO. In order to scale adaptive multiuser detection to such regimes, an attractive option, explored in several recent studies, is to reduce complexity via “beamspace” techniques which exploit the sparsity of mmWave channels. In this paper, we derive information-theoretic benchmarks using measured mmWave channels, comparing the capacity with ideal channel knowledge without complexity constraints, and several related benchmarks, with that attainable with low-complexity linear adaptive multiuser detection strategies in beamspace. We hope that the resulting insights regarding the price paid for various levels of simplification in processing will guide performance/complexity tradeoffs in the design of scalable signal processing algorithms.

  PublisherDOI

• Tiled Beamspace Processing for Scaling mmWave Massive MU-MIMO

  2024-10-07 · 1 citations

  articleSenior author

  Recent progress in millimeter wave (mmWave) silicon technologies has given rise to a new possibility: digital beamforming for truly massive multiuser (MU-) MIMO. However, there are two key challenges in scaling: packaging a vast array of antennas with the corresponding radio frequency integrated circuits (RFICs), and controlling the complexity of digital signal processing (DSP) for MU detection. In this paper, we show that modular tiled architectures, which simplify the task of RF packaging, also enable significant reduction of the communication and computational burden of DSP for MU-MIMO by utilizing beamspace techniques that take advantage of the sparsity of the mmWave channel. Specifically, we propose and investigate Linear Minimum Mean Squared Error (LMMSE) adaptive MU detection via novel tiled beamspace architectures in which the bulk of the DSP occurs in-place at each tile. The dimensionality reduction and parallelization enabled by such architectures not only reduce the computational burden of inference and training relative to a traditional "full array" baseline, but also significantly reduce the length of the required training period. We consider three different training strategies with differing requirements for computation and inter-tile communication: independent training for each tile, coordinated training across tiles, and hierarchical training based on independent training as a first stage. Simulation results show that these approaches can actually outperform the full array baseline when we limit the length of the training period.

  PublisherDOI

• A Fourier Analysis of Digital Beamforming with Severely Quantized mmWave Arrays

  2023-10-29 · 1 citations

  articleSenior author

  Recent progress in silicon radio frequency integrated circuits (RFICs) has opened the possibility of fully digital massive MIMO with hundreds of antennas in millimeter wave (mmWave) bands. A critical bottleneck in “mostly digital” processing is the cost and power consumption of analog-to-digital converters (ADCs). We consider here a mmWave massive MIMO receiver employing L-blt ADCs, a particularly energy-efficient choice, in a regime where prior work on Bussgang linearization does not apply: a small number of users propagating over spatially sparse mm Wave channels. We investigate beamspace techniques based on a spatial FFT across antenna elements, which concentrates the energy of each user to a small number of FFT bins. We provide a Fourier analysis of the spatial harmonics for one user through one path, characterizing the impact of the ADC nonlinearity, along with the aliasing and spectral spread due to sampling and windowing corresponding to an array with a finite, discrete number of antennas. The analysis provides guidance on training sequence design for isolating the “fundamental” spatial frequency corresponding to the true angle of arrival. Simulations show that the design succeeds in suppressing higher-order harmonics for two users with disparate power levels.

  PublisherDOI

• Improving Robustness via Tilted Exponential Layer: A Communication-Theoretic Perspective

  arXiv (Cornell University) · 2023-11-02

  preprintOpen accessSenior author

  State-of-the-art techniques for enhancing robustness of deep networks mostly rely on empirical risk minimization with suitable data augmentation. In this paper, we propose a complementary approach motivated by communication theory, aimed at enhancing the signal-to-noise ratio at the output of a neural network layer via neural competition during learning and inference. In addition to standard empirical risk minimization, neurons compete to sparsely represent layer inputs by maximization of a tilted exponential (TEXP) objective function for the layer. TEXP learning can be interpreted as maximum likelihood estimation of matched filters under a Gaussian model for data noise. Inference in a TEXP layer is accomplished by replacing batch norm by a tilted softmax, which can be interpreted as computation of posterior probabilities for the competing signaling hypotheses represented by each neuron. After providing insights via simplified models, we show, by experimentation on standard image datasets, that TEXP learning and inference enhances robustness against noise and other common corruptions, without requiring data augmentation. Further cumulative gains in robustness against this array of distortions can be obtained by appropriately combining TEXP with data augmentation techniques. The code for all our experiments is available at https://github.com/bhagyapuranik/texp_for_robustness.

  PublisherOA PDFDOI

• All-Digital LoS MIMO With Low-Precision Analog-to-Digital Conversion

  IEEE Transactions on Wireless Communications · 2022-01-20

  preprintOpen accessSenior author

  Line-of-sight (LoS) multi-input multi-output (MIMO) systems exhibit attractive scaling properties with increase in carrier frequency: for a fixed form factor and range, the spatial degrees of freedom increase quadratically for 2D arrays, in addition to the typically linear increase in available bandwidth. In this paper, we investigate whether modern all-digital baseband signal processing architectures can be devised for such regimes, given the difficulty of analog-to-digital conversion for large bandwidths. We propose low-precision quantizer designs and accompanying spatial demultiplexing algorithms, considering <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$2 \times 2$ </tex-math></inline-formula> LoS MIMO with QPSK for analytical insight, and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$4 \times 4$ </tex-math></inline-formula> MIMO with QPSK and 16QAM for performance evaluation. Unlike prior work, channel state information is utilized only at the receiver (i.e., transmit precoding is not employed). We investigate quantizers with regular structure whose high-SNR mutual information approaches that of an unquantized system. We prove that amplitude-phase quantization is necessary to attain this benchmark; phase-only quantization falls short. We show that quantizers based on maximizing per-antenna output entropy perform better than standard Minimum Mean Squared Quantization Error (MMSQE) quantization. For spatial demultiplexing with severely quantized observations, we introduce the novel concept of virtual quantization which, combined with linear detection, provides reliable demodulation at significantly reduced complexity compared to maximum likelihood detection.

  PublisherOA PDFDOI

• On the design of low-resolution sensing matrices for noncoherent compressive channel estimation

  2022-10-31

  articleSenior author

  Low-overhead channel estimation is a crucial prerequisite for mmWave massive MIMO communication. Compressive sensing techniques are well suited to this task, as they exploit the angular domain sparsity of multipath channels to accurately recover all significant paths in the channel with a small number of beacon measurements. In practice, however, frontend limitations such as loss of phase coherence across beacons and heavy quantization of beamforming weights necessitate the modification of conventional compressive sensing algorithms for agile channel estimation on practical frontends. In this paper, we propose a novel procedure for noncoherent compressive channel estimation on low-resolution arrays that relies on careful design of the projection weight vectors from a finite dictionary to facilitate iterative recovery of the measurement phases. In the noiseless case, the proposed method can match the performance of coherent compressive channel estimation with a factor of 3 increase in measurement complexity. Numerical results show that this factor should be increased to 4 in order to provide robustness to measurement noise.

  PublisherDOI

Recent grants

• TCHCS: COLLABORATIVE RESEARCH: Millimeter-wave MIMO: A New Architecture for Integrated 10-40 Gigabit Wireless/Optical Hybrid Networks

  NSF · $270k · 2006–2010

• ITR: Cross-Layer Optimization For 4G Wireless Networks: Heavy-Tailed Traffic, Multiuser Channels, and Pseudocells

  NSF · $462k · 2002–2007

• NeTS: Large: Collaborative Research: GigaNets: A Path to Experimental Research in Millimeter Wave Networking

  NSF · $1.6M · 2015–2020

• NeTS-NOSS: Imaging Sensor Nets: from Concept to Prototypes

  NSF · $915k · 2005–2011

• CIF: Medium: Collaborative Research: Distributed coherence: fundamental building blocks, system concepts, and experimental demonstration

  NSF · $345k · 2013–2018

Frequent coauthors

• M.J.W. Rodwell

  Gillette Children's Specialty Healthcare

  38 shared

• Raghuraman Mudumbai

  University of Iowa

  35 shared

• G. Barriac

  Market Matters

  27 shared

• Maryam Eslami Rasekh

  University of California, Santa Barbara

  27 shared

• João P. Hespanha

  25 shared

• B.S. Manjunath

  University of California, Santa Barbara

  19 shared

• Sriram Venkateswaran

  Roche (Switzerland)

  18 shared

• Zhinus Marzi

  17 shared