arXiv subjects

Sumit Roy

Publications and source records attributed to Sumit Roy.

At least 37 records · Page 2Linked to original sources

arXiv2022

Experiments with mmWave Automotive Radar Test-bed

Millimeter-wave (mmW) radars are being increasingly integrated in commercial vehicles to support new Adaptive Driver Assisted Systems (ADAS) for its ability to provide high accuracy location, velocity, and angle estimates of objects, largely independent of environmental conditions. Such radar sensors not only perform basic functions such as detection and ranging/angular localization, but also provide critical inputs for environmental perception via object recognition and classification. To explore radar-based ADAS applications, we have assembled a lab-scale frequency modulated continuous wave (FMCW) radar test-bed (https://depts.washington.edu/funlab/research) based on Texas Instrument's (TI) automotive chipset family. In this work, we describe the test-bed components and provide a summary of FMCW radar operational principles. To date, we have created a large raw radar dataset for various objects under controlled scenarios. Thereafter, we apply some radar imaging algorithms to the collected dataset, and present some preliminary results that validate its capabilities in terms of object recognition. Our code is available at https://github.com/Xiangyu-Gao/mmWave-radar-signal-processing-and-microDoppler-classification.

eess.SP↗

arXiv2022

Architecture-Algorithmic Trade-offs in Multi-path Channel Estimation for mmWAVE Systems

5G mmWave massive MIMO systems are likely to be deployed in dense urban scenarios, where increasing network capacity is the primary objective. A key component in mmWave transceiver design is channel estimation which is challenging due to the very large signal bandwidths (order of GHz) implying significant resolved spatial multipath, coupled with large # of Tx/Rx antennas for large-scale MIMO. This results in significantly increased training overhead that in turn leads to unacceptably high computational complexity and power cost. Our work thus highlights the interplay of transceiver architecture and receiver signal processing algorithm choices that fundamentally address (mobile) handset power consumption, with minimal degradation in performance. We investigate trade-offs enabled by conjunction of hybrid beamforming mmWave receiver and channel estimation algorithms that exploit available sparsity in such wideband scenarios. A compressive sensing (CS) framework for sparse channel estimation -- Binary Iterative Hard Thresholding (BIHT) \cite{jacques2013robust} followed by linear reconstruction method with varying quantization (ADC) levels -- is explored to compare the trade-offs between bit-depth and sampling rate for a given ADC power budget. Performance analysis of the BIHT+ linear reconstruction method is conducted via simulation studies for 5G specified multi-path channel models and compared to oracle-assisted bounds for validation.

cs.IT↗

arXiv2022

Multi-Access Point Coordination for Next-Gen Wi-Fi Networks Aided by Deep Reinforcement Learning

Wi-Fi in the enterprise - characterized by overlapping Wi-Fi cells - constitutes the design challenge for next-generation networks. Standardization for recently started IEEE 802.11be (Wi-Fi 7) Working Groups has focused on significant medium access control layer changes that emphasize the role of the access point (AP) in radio resource management (RRM) for coordinating channel access due to the high collision probability with the distributed coordination function (DCF), especially in dense overlapping Wi-Fi networks. This paper proposes a novel multi-AP coordination system architecture aided by a centralized AP controller (APC). Meanwhile, a deep reinforcement learning channel access (DLCA) protocol is developed to replace the binary exponential backoff mechanism in DCF to enhance the network throughput by enabling the coordination of APs. First-Order Model-Agnostic Meta-Learning further enhances the network throughput. Subsequently, we also put forward a new greedy algorithm to maintain proportional fairness (PF) among multiple APs. Via the simulation, the performance of DLCA protocol in dense overlapping Wi-Fi networks is verified to have strong stability and outperform baselines such as Shared Transmission Opportunity (SH-TXOP) and Request-to-Send/Clear-to-Send (RTS/CTS) in terms of the network throughput by 10% and 3% as well as the network utility considering proportional fairness by 28.3% and 13.8%, respectively.

eess.SY↗

arXiv2022

RAMP-CNN: A Novel Neural Network for Enhanced Automotive Radar Object Recognition

Millimeter-wave radars are being increasingly integrated into commercial vehicles to support new advanced driver-assistance systems by enabling robust and high-performance object detection, localization, as well as recognition - a key component of new environmental perception. In this paper, we propose a novel radar multiple-perspectives convolutional neural network (RAMP-CNN) that extracts the location and class of objects based on further processing of the range-velocity-angle (RVA) heatmap sequences. To bypass the complexity of 4D convolutional neural networks (NN), we propose to combine several lower-dimension NN models within our RAMP-CNN model that nonetheless approaches the performance upper-bound with lower complexity. The extensive experiments show that the proposed RAMP-CNN model achieves better average recall and average precision than prior works in all testing scenarios. Besides, the RAMP-CNN model is validated to work robustly under nighttime, which enables low-cost radars as a potential substitute for pure optical sensing under severe conditions.

eess.SP↗

arXiv2022

Learning to Detect Open Carry and Concealed Object with 77GHz Radar

Detecting harmful carried objects plays a key role in intelligent surveillance systems and has widespread applications, for example, in airport security. In this paper, we focus on the relatively unexplored area of using low-cost 77GHz mmWave radar for the carried objects detection problem. The proposed system is capable of real-time detecting three classes of objects - laptop, phone, and knife - under open carry and concealed cases where objects are hidden with clothes or bags. This capability is achieved by the initial signal processing for localization and generating range-azimuth-elevation image cubes, followed by a deep learning-based prediction network and a multi-shot post-processing module for detecting objects. Extensive experiments for validating the system performance on detecting open carry and concealed objects have been presented with a self-built radar-camera testbed and collected dataset. Additionally, the influence of different input formats, factors, and parameters on system performance is analyzed, providing an intuitive understanding of the system. This system would be the very first baseline for other future works aiming to detect carried objects using 77GHz radar.

eess.SP↗

arXiv2022

On birational Torelli theorems

Let $G$ be a simple simply-connected connected linear algebraic group over $\mathbb{C}$. We proved a $2$-birational Torelli theorem for the moduli space of semistable principal $G$-bundles over a smooth curve of genus $\geq 3$, which says that if two such moduli spaces are $2$-birational then the curves are isomorphic. We also proved a $3$-birational Torelli theorem for the moduli space of stable symplectic parabolic bundles over a smooth curve of genus $\geq 4$.

math.AG↗

arXiv2021

MIMO-SAR: A Hierarchical High-resolution Imaging Algorithm for mmWave FMCW Radar in Autonomous Driving

Millimeter-wave radars are being increasingly integrated into commercial vehicles to support advanced driver-assistance system features. A key shortcoming for present-day vehicular radar imaging is poor azimuth resolution (for side-looking operation) due to the form factor limits on antenna size and placement. In this paper, we propose a solution via a new multiple-input and multiple-output synthetic aperture radar (MIMO-SAR) imaging technique, that applies coherent SAR principles to vehicular MIMO radar to improve the side-view (angular) resolution. The proposed 2-stage hierarchical MIMO-SAR processing workflow drastically reduces the computation load while preserving image resolution. To enable coherent processing over the synthetic aperture, we integrate a radar odometry algorithm that estimates the trajectory of ego-radar. The MIMO-SAR algorithm is validated by both simulations and real experiment data collected by a vehicle-mounted radar platform.

eess.SP↗

arXiv2021

Coexistence of Age and Throughput Optimizing Networks: A Spectrum Sharing Game

We investigate the coexistence of an age optimizing network (AON) and a throughput optimizing network (TON) that share a common spectrum band. We consider two modes of long run coexistence: (a) networks compete with each other for spectrum access, causing them to interfere and (b) networks cooperate to achieve non-interfering access. To model competition, we define a non-cooperative stage game parameterized by the average age of the AON at the beginning of the stage, derive its mixed strategy Nash equilibrium (MSNE), and analyze the evolution of age and throughput over an infinitely repeated game in which each network plays the MSNE at every stage. Cooperation uses a coordination device that performs a coin toss during each stage to select the network that must access the medium. Networks use the grim trigger punishment strategy, reverting to playing the MSNE every stage forever if the other disobeys the device. We determine if there exists a subgame perfect equilibrium, i.e., the networks obey the device forever as they find cooperation beneficial. We show that networks choose to cooperate only when they consist of a sufficiently small number of nodes, otherwise they prefer to disobey the device and compete.

cs.GT↗

arXiv2021

Enhancing WiFi Multiple Access Performance with Federated Deep Reinforcement Learning

Carrier sensing multiple access/collision avoidance (CSMA/CA) is the backbone MAC protocol for IEEE 802.11 networks. However, tuning the binary exponential back-off (BEB) mechanism of CSMA/CA in user-dense scenarios so as to maximize aggregate throughput still remains a practically essential and challenging problem. In this paper, we propose a new and enhanced multiple access mechanism based on the application of deep reinforcement learning (DRL) and Federated learning (FL). A new Monte Carlo (MC) reward updating method for DRL training is proposed and the access history of each station is used to derive a DRL-based MAC protocol that improves the network throughput vis-a-vis the traditional distributed coordination function (DCF). Further, federated learning (FL) is applied to achieve fairness among users. The simulation results showcase that the proposed federated reinforcement multiple access (FRMA) performs better than basic DCF by 20% and DCF with request-to-send/clear-to-send (RTS/CTS) by 5% while guaranteeing the fairness in user-dense scenarios.

cs.NI↗

arXiv2021

Torelli theorem for the moduli space of symplectic parabolic Higgs bundles

Let $(X,D)$ and $(X',D')$ be two compact Riemann surfaces of genus $g \geq 4$ with the set of marked points $D \subset X$ and $D' \subset X'$. Fix a parabolic line bundle $L$ with trivial parabolic structure. Let $\mathcal{N}_{\textnormal{Sp}}(2m,α,L)$ and $\mathcal{N}'_{\textnormal{Sp}}(2m,α,L)$ be the moduli spaces of stable symplectic parabolic Higgs bundles over $X$ and $X'$ respectively, with rank $2m$ and fixed parabolic structure $α$, with the symplectic form taking values in $L$. We prove that if $\mathcal{N}_{\textnormal{Sp}}(2m,α,L)$ is isomorphic to $\mathcal{N}'_{\textnormal{Sp}}(2m,α,L)$, then there exist an isomorphism between $X$ and $X'$ sending $D$ to $D'$.

math.AG↗

arXiv2020

Hitchin fibration on moduli of symplectic and orthogonal parabolic Higgs bundles

Let $X$ be a compact Riemann surface of genus $g \geq 2$, and let $D \subset X$ be a fixed finite subset. Let $\mathcal{M}(r,d,α)$ denote the moduli space of stable parabolic $G$-bundles (where $G$ is a complex orthogonal or symplectic group) of rank $r$, degree $d$ and weight type $α$ over $X$. Hitchin discovered that the cotangent bundle of the moduli space of stable bundles on an algebraic curve is an algebraically completely integrable system fibered, over a space of invariant polynomials, either by a Jacobian or a Prym variety of spectral curves. In this paper we study the Hitchin fibers for $\mathcal{M}(r,d,α)$.

math.AG↗

arXiv2020

A Non-Cooperative Multiple Access Game for Timely Updates

We consider a network of selfish nodes that would like to minimize the age of their updates at the other nodes. The nodes send their updates over a shared spectrum using a CSMA/CA based access mechanism. We model the resulting competition as a non-cooperative one-shot multiple access game and investigate equilibrium strategies for two distinct medium access settings (a) collisions are shorter than successful transmissions and (b) collisions are longer. We investigate competition in a CSMA/CA slot, where a node may choose to transmit or stay idle. We find that medium access settings exert strong incentive effects on the nodes. We show that when collisions are shorter, transmit is a weakly dominant strategy. This leads to all nodes transmitting in the CSMA/CA slot, therefore guaranteeing a collision. In contrast, when collisions are longer, no weakly dominant strategy exists and under certain conditions on the ages at the beginning of the slot, we derive the mixed strategy Nash equilibrium.

cs.GT↗

arXiv2020

Communications-Caching-Computing Tradeoff Analysis for Bidirectional Data Computation in Mobile Edge Networks

With the advent of the modern mobile traffic, e.g., online gaming, augmented reality delivery and etc., a novel bidirectional computation task model where the input data of each task consists of two parts, one generated at the mobile device in real-time and the other originated from the Internet proactively, is emerging as an important use case of 5G. In this paper, for ease of analytical analysis, we consider the homogeneous bidirectional computation task model in a mobile edge network which consists of one mobile edge computing (MEC) server and one mobile device, both enabled with computing and caching capabilities. Each task can be served via three mechanisms, i.e., local computing with local caching, local computing without local caching and computing at the MEC server. To minimize the average bandwidth, we formulate the joint caching and computing optimization problem under the latency, cache size and average power constraints. We derive the closed-form expressions for the optimal policy and the minimum bandwidth. The tradeoff among communications, computing and caching is illustrated both analytically and numerically, which provides insightful guideline for the network designers.

cs.IT↗

arXiv2019

Absolving the SSINS of Precision Interferometric Radio Data: A New Technique for Mitigating Faint Radio Frequency Interference

We introduce a new pipeline for analyzing and mitigating radio frequency interference (RFI), which we call Sky-Subtracted Incoherent Noise Spectra (SSINS). SSINS is designed to identify and remove faint RFI below the single baseline thermal noise by employing a frequency-matched detection algorithm on baseline-averaged amplitudes of time-differenced visibilities. We demonstrate the capabilities of SSINS using the Murchison Widefield Array (MWA) in Western Australia. We successfully image aircraft flying over the array via digital television (DTV) reflection detected using SSINS and summarize an RFI occupancy survey of MWA Epoch of Reionization data. We describe how to use SSINS with new data using a documented, publicly available implementation with comprehensive usage tutorials.

astro-ph.IM↗

arXiv2018

Optimizing City-Wide White-Fi Networks in TV White Spaces

White-Fi refers to WiFi deployed in the TV white spaces. Unlike its ISM band counterparts, White-Fi must obey requirements that protect TV reception. As a result, optimization of citywide White-Fi networks faces the challenges of heterogeneous channel availability and link quality, over location. The former is because, at any location, channels in use by TV networks are not available for use by White-Fi. The latter is because the link quality achievable at a White-Fi receiver is determined by not only its link gain to its transmitter but also by its link gains to TV transmitters and its transmitter's link gains to TV receivers. In this work, we model the medium access control (MAC) throughput of a White-Fi network. We propose heuristic algorithms to optimize the throughput, given the described heterogeneity. The algorithms assign power, access probability, and channels to nodes in the network, under the constraint that reception at TV receivers is not compromised. We evaluate the efficacy of our approach over example city-wide White-Fi networks deployed over Denver and Columbus (respectively, low and high channel availability) in the USA, and compare with assignments cognizant of heterogeneity to a lesser degree, for example, akin to FCC regulations.

cs.NI↗

arXiv2018

SAS-Assisted Coexistence-Aware Dynamic Channel Assignment in CBRS Band

The paradigm of shared spectrum allows secondary devices to opportunistically access spectrum bands underutilized by primary owners. Recently, the FCC has targeted the sharing of the 3.5 GHz (3550-3700 MHz) federal spectrum with commercial systems such as small cells. The rules require a Spectrum Access System (SAS) to accommodate three service tiers: 1) Incumbent Access, 2) Priority Access (PA), and 3) Generalized Authorized Access (GAA). In this work, we study the SAS-assisted dynamic channel assignment (CA) for PA and GAA tiers.We introduce the node-channel-pair conflict graph to capture pairwise interference, channel and geographic contiguity constraints, spatially varying channel availability, and coexistence awareness. The proposed conflict graph allows us to formulate PA CA and GAA CA with binary conflicts as max-cardinality and max-reward CA, respectively. Approximate solutions can be found by a heuristic-based algorithm that search for the maximum weighted independent set. We further formulate GAA CA with non-binary conflicts as max-utility CA. We show that the utility function is submodular, and the problem is an instance of matroid-constrained submodular maximization. A polynomial-time algorithm based on local search is proposed that provides a provable performance guarantee. Extensive simulations using a real-world Wi-Fi hotspot location dataset are conducted to evaluate the proposed algorithms. Our results have demonstrated the advantages of the proposed graph representation and improved performance of the proposed algorithms over the baseline algorithms.

cs.NI↗

arXiv2018

Improving the Performance of WLANs by Reducing Unnecessary Active Scans

We consider the problem of excessive and unnecessary active scans in heavily utilized WLANs during which low rate probe requests and responses are broadcast. These management frames severely impact the goodput. Our analysis of two production WLANs reveals that lesser number of non-overlapping channels in $2.4$ GHz makes it more prone to the effects of increased probe frames than $5$ GHz. We find that not only up to $90$% of probe responses carry redundant information but the probe traffic can be as high as $60$\% of the management traffic. Furthermore, active scanning severely impacts real-time applications at a client as it increases the latency by $91$ times. We present a detailed analysis of the impact of active scans on an individual client and the whole network. We discuss three ways to control the probe traffic in production WLANs -- access point configurations, network planning, and client modification. Our proposals for access point configuration are in line with current WLAN deployments, better network planning is device agnostic in nature, and client modification reduces the average number of probe requests per client by up to $50$% without hampering the ongoing WiFi connection.

cs.NI↗

arXiv2018

Offline Extraction of Indic Regional Language from Natural Scene Image using Text Segmentation and Deep Convolutional Sequence

Regional language extraction from a natural scene image is always a challenging proposition due to its dependence on the text information extracted from Image. Text Extraction on the other hand varies on different lighting condition, arbitrary orientation, inadequate text information, heavy background influence over text and change of text appearance. This paper presents a novel unified method for tackling the above challenges. The proposed work uses an image correction and segmentation technique on the existing Text Detection Pipeline an Efficient and Accurate Scene Text Detector (EAST). EAST uses standard PVAnet architecture to select features and non maximal suppression to detect text from image. Text recognition is done using combined architecture of MaxOut convolution neural network (CNN) and Bidirectional long short term memory (LSTM) network. After recognizing text using the Deep Learning based approach, the native Languages are translated to English and tokenized using standard Text Tokenizers. The tokens that very likely represent a location is used to find the Global Positioning System (GPS) coordinates of the location and subsequently the regional languages spoken in that location is extracted. The proposed method is tested on a self generated dataset collected from Government of India dataset and experimented on Standard Dataset to evaluate the performance of the proposed technique. Comparative study with a few state-of-the-art methods on text detection, recognition and extraction of regional language from images shows that the proposed method outperforms the existing methods.

cs.CV↗