arXiv subjects

Sundeep Rangan

Publications and source records attributed to Sundeep Rangan.

At least 73 records · Page 4Linked to original sources

arXiv2021

Millimeter-Wave UAV Coverage in Urban Environments

With growing interest in mmWave connectivity for UAVs, a basic question is whether networks intended for terrestrial users can provide sufficient aerial coverage as well. To assess this possibility, the paper proposes a novel evaluation methodology using generative models trained on detailed ray tracing data. These models capture complex propagation characteristics and can be readily combined with antenna and beamforming assumptions. Extensive simulation using these models indicate that standard (street-level and downtilted) base stations at typical microcellular densities can indeed provide satisfactory UAV coverage. Interestingly, the coverage is possible via a conjunction of antenna sidelobes and strong reflections. With sparser deployments, the coverage is only guaranteed at progressively higher altitudes. Additional dedicated (rooftop-mounted and uptilted) base stations strengthen the coverage provided that their density is comparable to that of the standard deployment, and would be instrumental for sparse deployments of the latter.

eess.SP↗

arXiv2021

What Will the Future of UAV Cellular Communications Be? A Flight from 5G to 6G

What will the future of UAV cellular communications be? In this tutorial article, we address such a compelling yet difficult question by embarking on a journey from 5G to 6G and sharing a large number of realistic case studies supported by original results. We start by overviewing the status quo on UAV communications from an industrial standpoint, providing fresh updates from the 3GPP and detailing new 5G NR features in support of aerial devices. We then show the potential and the limitations of such features. In particular, we demonstrate how sub-6 GHz massive MIMO can successfully tackle cell selection and interference challenges, we showcase encouraging mmWave coverage evaluations in both urban and suburban/rural settings, and we examine the peculiarities of direct device-to-device communications in the sky. Moving on, we sneak a peek at next-generation UAV communications, listing some of the use cases envisioned for the 2030s. We identify the most promising 6G enablers for UAV communication, those expected to take the performance and reliability to the next level. For each of these disruptive new paradigms (non-terrestrial networks, cell-free architectures, artificial intelligence, reconfigurable intelligent surfaces, and THz communications), we gauge the prospective benefits for UAVs and discuss the main technological hurdles that stand in the way. All along, we distil our numerous findings into essential takeaways, and we identify key open problems worthy of further study.

cs.IT↗

arXiv2021

Lightweight UAV-based Measurement System for Air-to-Ground Channels at 28 GHz

Wireless communication at millimeter wave frequencies has attracted considerable attention for the delivery of high-bit-rate connectivity to unmanned aerial vehicles (UAVs). However, conducting the channel measurements necessary to assess communication at these frequencies has been challenging due to the severe payload and power restrictions in commercial UAVs. This work presents a novel lightweight (approximately 1.3 kg) channel measurement system at 28 GHz installed on a commercially available UAV. A ground transmitter equipped with a horn antenna conveys sounding signals to a UAV equipped with a lightweight spectrum analyzer. We demonstrate that the measurements can be highly influenced by the antenna pattern as shaped by the UAV's frame. A calibration procedure is presented to correct for the resulting angular variations in antenna gain. The measurement setup is then validated on real flights from an airstrip at distances in excess of 300 m.

eess.SP↗

arXiv2021

Asymptotics of Ridge Regression in Convolutional Models

Understanding generalization and estimation error of estimators for simple models such as linear and generalized linear models has attracted a lot of attention recently. This is in part due to an interesting observation made in machine learning community that highly over-parameterized neural networks achieve zero training error, and yet they are able to generalize well over the test samples. This phenomenon is captured by the so called double descent curve, where the generalization error starts decreasing again after the interpolation threshold. A series of recent works tried to explain such phenomenon for simple models. In this work, we analyze the asymptotics of estimation error in ridge estimators for convolutional linear models. These convolutional inverse problems, also known as deconvolution, naturally arise in different fields such as seismology, imaging, and acoustics among others. Our results hold for a large class of input distributions that include i.i.d. features as a special case. We derive exact formulae for estimation error of ridge estimators that hold in a certain high-dimensional regime. We show the double descent phenomenon in our experiments for convolutional models and show that our theoretical results match the experiments.

stat.ML↗

arXiv2021

On Single-User Interactive Beam Alignment in Next Generation Systems: A Deep Learning Viewpoint

Communication in high frequencies such as millimeter wave and terahertz suffer from high path-loss and intense shadowing which necessitates beamforming for reliable data transmission. On the other hand, at high frequencies the channels are sparse and consist of few spatial clusters. Therefore, beam alignment (BA) strategies are used to find the direction of these channel clusters and adjust the width of the beam used for data transmission. In this work, a single-user uplink scenario where the channel has one dominant cluster is considered. It is assumed that the user transmits a set of BA packets over a fixed duration. Meanwhile, the base-station (BS) uses different probing beams to scan different angular regions. Since the BS measurements are noisy, it is not possible to find a narrow beam that includes the angle of arrival (AoA) of the user with probability one. Therefore, the BS allocates a narrow beam to the user which includes the AoA of the user with a predetermined error probability while minimizing the expected beamwidth of the allocated beam. Due to intractability of this noisy BA problem, here this problem is posed as an end-to-end optimization of a deep neural network (DNN) and effects of different loss functions are discussed and investigated. It is observed that the proposed DNN based BA, at high SNRs, achieves a performance close to that of the optimal BA when there is no-noise and for all SNRs, outperforms state-of-the-art.

cs.LG↗

arXiv2021

Implicit Bias of Linear RNNs

Contemporary wisdom based on empirical studies suggests that standard recurrent neural networks (RNNs) do not perform well on tasks requiring long-term memory. However, precise reasoning for this behavior is still unknown. This paper provides a rigorous explanation of this property in the special case of linear RNNs. Although this work is limited to linear RNNs, even these systems have traditionally been difficult to analyze due to their non-linear parameterization. Using recently-developed kernel regime analysis, our main result shows that linear RNNs learned from random initializations are functionally equivalent to a certain weighted 1D-convolutional network. Importantly, the weightings in the equivalent model cause an implicit bias to elements with smaller time lags in the convolution and hence, shorter memory. The degree of this bias depends on the variance of the transition kernel matrix at initialization and is related to the classic exploding and vanishing gradients problem. The theory is validated in both synthetic and real data experiments.

cs.LG↗

arXiv2020

Millimeter Wave Channel Modeling via Generative Neural Networks

Statistical channel models are instrumental to design and evaluate wireless communication systems. In the millimeter wave bands, such models become acutely challenging; they must capture the delay, directions, and path gains, for each link and with high resolution. This paper presents a general modeling methodology based on training generative neural networks from data. The proposed generative model consists of a two-stage structure that first predicts the state of each link (line-of-sight, non-line-of-sight, or outage), and subsequently feeds this state into a conditional variational autoencoder that generates the path losses, delays, and angles of arrival and departure for all its propagation paths. Importantly, minimal prior assumptions are made, enabling the model to capture complex relationships within the data. The methodology is demonstrated for 28GHz air-to-ground channels in an urban environment, with training datasets produced by means of ray tracing.

eess.SP↗

arXiv2020

Enabling Remote Whole-Body Control with 5G Edge Computing

Real-world applications require light-weight, energy-efficient, fully autonomous robots. Yet, increasing autonomy is oftentimes synonymous with escalating computational requirements. It might thus be desirable to offload intensive computation--not only sensing and planning, but also low-level whole-body control--to remote servers in order to reduce on-board computational needs. Fifth Generation (5G) wireless cellular technology, with its low latency and high bandwidth capabilities, has the potential to unlock cloud-based high performance control of complex robots. However, state-of-the-art control algorithms for legged robots can only tolerate very low control delays, which even ultra-low latency 5G edge computing can sometimes fail to achieve. In this work, we investigate the problem of cloud-based whole-body control of legged robots over a 5G link. We propose a novel approach that consists of a standard optimization-based controller on the network edge and a local linear, approximately optimal controller that significantly reduces on-board computational needs while increasing robustness to delay and possible loss of communication. Simulation experiments on humanoid balancing and walking tasks that includes a realistic 5G communication model demonstrate significant improvement of the reliability of robot locomotion under jitter and delays likely to experienced in 5G wireless links.

cs.RO↗

arXiv2020

Capacity Bounds for Communication Systems with Quantization and Spectral Constraints

Low-resolution digital-to-analog and analog-to-digital converters (DACs and ADCs) have attracted considerable attention in efforts to reduce power consumption in millimeter wave (mmWave) and massive MIMO systems. This paper presents an information-theoretic analysis with capacity bounds for classes of linear transceivers with quantization. The transmitter modulates symbols via a unitary transform followed by a DAC and the receiver employs an ADC followed by the inverse unitary transform. If the unitary transform is set to an FFT matrix, the model naturally captures filtering and spectral constraints which are essential to model in any practical transceiver. In particular, this model allows studying the impact of quantization on out-of-band emission constraints. In the limit of a large random unitary transform, it is shown that the effect of quantization can be precisely described via an additive Gaussian noise model. This model in turn leads to simple and intuitive expressions for the power spectrum of the transmitted signal and a lower bound to the capacity with quantization. Comparison with non-quantized capacity and a capacity upper bound that does not make linearity assumptions suggests that while low resolution quantization has minimal impact on the achievable rate at typical parameters in 5G systems today, satisfying out-of-band emissions are potentially much more of a challenge.

cs.IT↗

arXiv2020

Millimeter Wave Remove UAV Control and Communications for Public Safety Scenarios

Communication and video capture from unmanned aerial vehicles (UAVs) offer significant potential for assisting first responders in remote public safety settings. In such uses, millimeter wave (mmWave) wireless links can provide high throughput and low latency connectivity between the UAV and a remote command center. However, maintaining reliable aerial communication in the mmWave bands is challenging due to the need to support high speed beam tracking and overcome blockage. This paper provides a simulation study aimed at assessing the feasibility of public safety UAV connectivity through a 5G link at 28 GHz. Real flight motion traces are captured during maneuvers similar to those expected in public safety settings. The motions traces are then incorporated into a detailed mmWave network simulator that models the channel, blockage, beamforming and full 3GPP protocol stack. We show that 5G mmWave communications can deliver throughput up to 1 Gbps with consistent sub ms latency when the base station is located near the mission area, enabling remote offloading of the UAV control and perception algorithms.

eess.SP↗

arXiv2020

Low-Rank Nonlinear Decoding of $μ$-ECoG from the Primary Auditory Cortex

This paper considers the problem of neural decoding from parallel neural measurements systems such as micro-electrocorticography ($μ$-ECoG). In systems with large numbers of array elements at very high sampling rates, the dimension of the raw measurement data may be large. Learning neural decoders for this high-dimensional data can be challenging, particularly when the number of training samples is limited. To address this challenge, this work presents a novel neural network decoder with a low-rank structure in the first hidden layer. The low-rank constraints dramatically reduce the number of parameters in the decoder while still enabling a rich class of nonlinear decoder maps. The low-rank decoder is illustrated on $μ$-ECoG data from the primary auditory cortex (A1) of awake rats. This decoding problem is particularly challenging due to the complexity of neural responses in the auditory cortex and the presence of confounding signals in awake animals. It is shown that the proposed low-rank decoder significantly outperforms models using standard dimensionality reduction techniques such as principal component analysis (PCA).

q-bio.NC↗

arXiv2020

Generalization Error of Generalized Linear Models in High Dimensions

At the heart of machine learning lies the question of generalizability of learned rules over previously unseen data. While over-parameterized models based on neural networks are now ubiquitous in machine learning applications, our understanding of their generalization capabilities is incomplete. This task is made harder by the non-convexity of the underlying learning problems. We provide a general framework to characterize the asymptotic generalization error for single-layer neural networks (i.e., generalized linear models) with arbitrary non-linearities, making it applicable to regression as well as classification problems. This framework enables analyzing the effect of (i) over-parameterization and non-linearity during modeling; and (ii) choices of loss function, initialization, and regularizer during learning. Our model also captures mismatch between training and test distributions. As examples, we analyze a few special cases, namely linear regression and logistic regression. We are also able to rigorously and analytically explain the \emph{double descent} phenomenon in generalized linear models.

cs.LG↗

arXiv2020

Beamformed mmWave System Propagation at 60GHz in an Office Environment

Millimeter wave wireless systems rely heavily on directional communication in narrow steerable beams. Tools to measure the spatial and temporal nature of the channel are necessary to evaluate beamforming and related algorithms. This paper presents a novel 60~GHz phased-array based directional channel sounder and data analysis procedure that can accurately extract paths and their transmit and receive directions. The gains along each path can also be measured for analyzing blocking scenarios. The sounder is validated in an indoor office environment.

eess.SY↗

arXiv2020

Capacity Scaling of Cellular Networks: Impact of Bandwidth, Infrastructure Density and Number of Antennas

The availability of very wide spectrum in millimeter wave bands combined with large antenna arrays and ultra dense networks raises two basic questions: What is the true value of overly abundant degrees of freedom and how can networks be designed to fully exploit them? This paper determines the capacity scaling of large cellular networks as a function of bandwidth, area, number of antennas and base station density. It is found that the network capacity has a fundamental bandwidth scaling limit, beyond which the network becomes power-limited. An infrastructure multi-hop protocol achieves the optimal network capacity scaling for all network parameters. In contrast, current protocols that use only single-hop direct transmissions can not achieve the capacity scaling in wideband regimes except in the special case when the density of base stations is taken to impractical extremes. This finding suggests that multi-hop communication will be important to fully realize the potential of next-generation cellular networks. Dedicated relays, if sufficiently dense, can also perform this task, relieving user nodes from the battery drain of cooperation. On the other hand, more sophisticated strategies such as hierarchical cooperation, that are essential for achieving capacity scaling in ad hoc networks, are unnecessary in the cellular context.

cs.IT↗

arXiv2020

Towards 6G Networks: Use Cases and Technologies

Reliable data connectivity is vital for the ever increasingly intelligent, automated and ubiquitous digital world. Mobile networks are the data highways and, in a fully connected, intelligent digital world, will need to connect everything, from people to vehicles, sensors, data, cloud resources and even robotic agents. Fifth generation (5G) wireless networks (that are being currently deployed) offer significant advances beyond LTE, but may be unable to meet the full connectivity demands of the future digital society. Therefore, this article discusses technologies that will evolve wireless networks towards a sixth generation (6G), and that we consider as enablers for several potential 6G use cases. We provide a full-stack, system-level perspective on 6G scenarios and requirements, and select 6G technologies that can satisfy them either by improving the 5G design, or by introducing completely new communication paradigms.

cs.NI↗

arXiv2020

Inference in Multi-Layer Networks with Matrix-Valued Unknowns

We consider the problem of inferring the input and hidden variables of a stochastic multi-layer neural network from an observation of the output. The hidden variables in each layer are represented as matrices. This problem applies to signal recovery via deep generative prior models, multi-task and mixed regression and learning certain classes of two-layer neural networks. A unified approximation algorithm for both MAP and MMSE inference is proposed by extending a recently-developed Multi-Layer Vector Approximate Message Passing (ML-VAMP) algorithm to handle matrix-valued unknowns. It is shown that the performance of the proposed Multi-Layer Matrix VAMP (ML-Mat-VAMP) algorithm can be exactly predicted in a certain random large-system limit, where the dimensions $N\times d$ of the unknown quantities grow as $N\rightarrow\infty$ with $d$ fixed. In the two-layer neural-network learning problem, this scaling corresponds to the case where the number of input features and training samples grow to infinity but the number of hidden nodes stays fixed. The analysis enables a precise prediction of the parameter and test error of the learning.

cs.LG↗

arXiv2020

Energy and Latency of Beamforming Architectures for Initial Access in mmWave Wireless Networks

Future millimeter-wave (mmWave) systems, 5G cellular or WiFi, must rely on highly directional links to overcome severe pathloss in these frequency bands. Establishing such links requires the mutual discovery of the transmitter and the receiver %in the angular domain potentially leading to a large latency and high energy consumption. In this work, we show that both the discovery latency and energy consumption can be significantly reduced by using fully digital front-ends. In fact, we establish that by reducing the resolution of the fully-digital front-ends we can achieve lower energy consumption compared to both analog and high-resolution digital beamformers. Since beamforming through analog front-ends allows sampling in only one direction at a time, the mobile device is ''on'' for a longer time compared to a digital beamformer which can get spatial samples from all directions in one shot. We show that the energy consumed by the analog front-end can be four to six times more than that of the digital front-ends, depending on the size of the employed antenna arrays. We recognize, however, that using fully digital beamforming post beam discovery, i.e., for data transmission, is not viable from a power consumption standpoint. To address this issue, we propose the use of digital beamformers with low-resolution analog to digital converters (4 bits). This reduction in resolution brings the power consumption to the same level as analog beamforming for data transmissions while benefiting from the spatial multiplexing capabilities of fully digital beamforming, thus reducing initial discovery latency and improving energy efficiency.

eess.SP↗

arXiv2019

Inference with Deep Generative Priors in High Dimensions

Deep generative priors offer powerful models for complex-structured data, such as images, audio, and text. Using these priors in inverse problems typically requires estimating the input and/or hidden signals in a multi-layer deep neural network from observation of its output. While these approaches have been successful in practice, rigorous performance analysis is complicated by the non-convex nature of the underlying optimization problems. This paper presents a novel algorithm, Multi-Layer Vector Approximate Message Passing (ML-VAMP), for inference in multi-layer stochastic neural networks. ML-VAMP can be configured to compute maximum a priori (MAP) or approximate minimum mean-squared error (MMSE) estimates for these networks. We show that the performance of ML-VAMP can be exactly predicted in a certain high-dimensional random limit. Furthermore, under certain conditions, ML-VAMP yields estimates that achieve the minimum (i.e., Bayes-optimal) MSE as predicted by the replica method. In this way, ML-VAMP provides a computationally efficient method for multi-layer inference with an exact performance characterization and testable conditions for optimality in the large-system limit.

cs.LG↗