Publications | Dr. Evangelos Vlachos

Geometry-Aware Regularization for Deep-Unfolded MIMO-OTFS Detection

Mon, 24 Aug 2026 00:00:00 +0000

Key Contributions

Gram matrix analysis: Analytical characterization of the non-uniform diagonal structure of the MIMO-OTFS Gram matrix, proving that all diagonal variation arises exclusively from inter-path coupling.
Geometry-aware regularizer: A coordinate-wise regularizer embedded into a deep-unfolded Conjugate Gradient detector, with per-component weights learned in a data-driven manner.
Performance gains: The proposed detector outperforms low-complexity baselines and approaches heavier neural architectures at medium-to-high SNR, with improved robustness in high-mobility scenarios.

Semantically-Informed Predictive Control for Autonomous UAV Connectivity in Urban Environments

Mon, 24 Aug 2026 00:00:00 +0000

Key Contributions

An Effective Capacity metric that extends the radio map paradigm by weighting raw throughput with a material-dependent reliability coefficient, decoupling signal strength from link stability and exposing permeable vegetation corridors that signal-strength planners overlook.
A two-layer planner comprising (i) a Semantically-Informed RRT backbone augmented with material-dependent reliability and capacity acceptance checks that prune expansion through multipath zones, and (ii) a gradient-based trajectory optimizer that deforms the path by balancing signal-seeking and reliability-seeking forces derived from the product-rule decomposition of ∇C_eff.
Hardware validation of the material ambiguity premise via characterization with a TI IWR6843AOP 60 GHz FMCW radar, confirming that radar alone cannot distinguish concrete from vegetation.

Hardware-Aware SE(3) Control Barrier Functions for Counter-UAS Interceptors with Directed Energy Payloads

Wed, 17 Jun 2026 00:00:00 +0000

Simulation Demo

The animation below compares two methods on the SE(3) CBF scenario: the unfiltered controller (red) and the proposed ADMM-CBF safety filter (blue), which maintains FoV pointing throughout the pursuit maneuver.

Key Contributions

A hardware-aware CBF safety filter on the SE(3) manifold that provides formal FoV pointing guarantees for directed-energy payloads under aggressive pursuit dynamics.
Formulation of phased-array electronic steering limits as a relative-degree-two safety constraint with analytically verified Lie derivatives.
An ADMM solver with constant-time core factorization enabling real-time onboard execution.
Validation in both a nonlinear SE(3) simulation and a PX4 SITL environment, demonstrating elimination of all FoV violations and recovery of antenna gain through beam tightening.

Biconvex ADMM for Sensing-Aided XL-MIMO Channel Estimation at the THz Frequency Band

Sat, 13 Jun 2026 00:00:00 +0000

Key Contributions

Biconvex ADMM with binary support enforcement: We show that the THz XL-MIMO dual-wideband channel concentrates onto a shared binary delay support of drastically reduced dimension, and develop a biconvex ADMM where iterative hard thresholding (IHT) exactly enforces this structure — outperforming continuous-valued sparse recovery and oracle-LS baselines in the barely-determined regime.
Sensing-aided warm start at no pilot cost: The base station’s monostatic radar output initialises the delay-support estimate, yielding provably exact support recovery beyond a pilot-length threshold, with no additional pilot overhead.
CRB proximity: The proposed estimator closely approaches the Cramér-Rao Bound and closes the ISAC sensing loop by delivering the recovered delay profile as a radar output.

Consistency-Gated Cooperative Localization for UAV Swarms Under Intermittent GPS

Tue, 02 Jun 2026 00:00:00 +0000

Key Contributions

Consistency-gated framework: A cooperative localization scheme that actively prevents harmful cooperative updates when inter-agent measurements are inconsistent under intermittent GPS.
Two-stage estimation: Decoupled architecture separating local observability recovery (EKF with GPS/IMU) from cooperative refinement (distributed ADMM over UWB range and AoA constraints).
Lightweight acceptance gate: A learning-based gate driven solely by runtime residual and consensus indicators — no ground truth required online.
Quantitative gains: Demonstrated moderate improvements in absolute accuracy and substantial gains in formation-relevant relative accuracy over GPS/IMU baselines.

Receiving RISs: Enabling Channel Estimation and Autonomous Configuration

Wed, 01 Apr 2026 00:00:00 +0000

Key Contributions

Proposes a semi-passive RIS architecture with RF chains that enables active channel estimation via tunable absorption phase profiles, eliminating the need for external pilot injection.
Introduces a channel estimation protocol exploiting beamspace sparsity and low-rank structure of extremely large MIMO channels, solved efficiently via ADMM.
Demonstrates how RIS-side channel estimates can be leveraged to autonomously optimize phase-quantized reflection coefficients, enabling self-configuring RIS operation.

Design and Evaluation of a Simulation Framework for FMCW Radar Sensors Deployed on UAV Platforms

Mon, 06 Oct 2025 00:00:00 +0000

ISAC for AAM: Integrated Sensing and Communication Technology for Advanced Air Mobility

Mon, 06 Oct 2025 00:00:00 +0000

Biconvex Optimization for Time-Domain Channel Estimation Under Dual-Wideband Fading Conditions

Wed, 01 Jan 2025 00:00:00 +0000

Key Contributions

Development of a unified framework for MIMO channel estimation that explicitly accounts for beam squint at both transmitter and receiver, as well as intersymbol interference (ISI) caused by propagation delays.
Proposal of a novel mixed-integer biconvex optimization model that handles diverse delay patterns using a binary vector and decomposes the channel into path-specific matrices, enabling smart environment integration.
Introduction of a semi-parametric approach that estimates delays and separates path gains/angles, offering advantages over conventional methods and enabling advanced optimization for dual-wideband fading conditions.

Improving Wideband Massive MIMO Channel Estimation With UAV State-Space Information

Wed, 01 Jan 2025 00:00:00 +0000

Key Contributions

Development of a wideband channel model incorporating beam squint effects relevant for high-frequency mMIMO-UAV communications.
Proposal of a state-assisted mMIMO channel estimation approach that leverages UAV motion information to improve estimation accuracy.
Design of an efficient estimation algorithm suitable for single-RF chain architectures, minimizing training overhead.
Demonstration via simulations that the proposed technique achieves high estimation accuracy with significantly reduced training data requirements.

Results & Insights

Simulated UAV flight scenarios and Kalman-based position estimation error across different trajectory profiles.

The state-space model accurately tracks UAV position across diverse scenarios, confirming that UAV motion data can be reliably incorporated into the channel estimation process.

Channel estimation NMSE comparison between the proposed state-assisted method and benchmark schemes across different UAV velocities.

The proposed estimator outperforms benchmark methods, particularly at higher UAV speeds, highlighting the effectiveness of incorporating state-space information for accurate channel estimation.

Intelligent UAV Path Planning for Ergodic Rate Maximization of MIMO Multipath Channels

Wed, 01 Jan 2025 00:00:00 +0000

Key Contributions

Proposes a decentralized MPC framework that jointly optimizes UAV trajectory planning and high-frequency MIMO communication performance under realistic multipath channel conditions.
Integrates ergodic spectral efficiency derived from location-dependent channel models into the control cost function, enabling communication-aware path planning.
Balances three competing objectives — mission time, probabilistic collision safety under stochastic disturbances, and uplink ergodic rate — within a unified control framework.

Flexible Hybrid Precoding for Integrated Sensing and Communications

Sat, 09 Nov 2024 00:00:00 +0000

Key Contributions

Presents flexible hybrid precoding architectures for Integrated Sensing and Communications (ISAC), enabling simultaneous radar sensing and multi-user data transmission from a shared antenna array.
Develops optimization frameworks that balance sensing beampattern design with communication spectral efficiency under practical hardware constraints including finite-resolution phase shifters.
Provides a comprehensive treatment of ISAC precoding in a book chapter context, covering design principles, algorithmic solutions, and trade-off analysis for next-generation wireless systems.

Millimeter Wave Channel Estimation for Lens Based Hybrid MIMO with Low Resolution ADCs

Sun, 28 May 2023 00:00:00 +0000

Presentation

Presenting the paper at IEEE ICC 2023, Rome, 1 June 2023.

Covariance-Based Hybrid Beamforming for Spectrally Efficient Joint Radar-Communications

Sun, 01 Jan 2023 00:00:00 +0000

Key Contributions

Proposes a robust hybrid beamformer based on covariance matrices to minimize interference in joint radar-communications systems.
Introduces fractional programming for optimal RF chain selection, enabling flexibility in hybrid beamformer design.
Maximizes mutual information using a weighted dual function approach, considering the trade-off between radar and communication performance.
Incorporates hardware constraints with low-resolution DACs, ensuring practical implementation feasibility.

Ideal communication capacity and its covariance-based approximation for the JRC system.

This figure demonstrates the effectiveness of the proposed covariance-based approach in approximating the ideal communication capacity, highlighting its accuracy and practicality for JRC systems.

Mutual information performance versus SNR for different weighting factors in the JRC system.

The results show that the proposed method consistently outperforms baseline cases, except for the interference-free baseline, across various SNR levels and weighting factors.

Mutual information performance versus SNR for different numbers of transmit antennas in the JRC system.

The proposed method exhibits superior performance compared to baseline approaches, particularly at higher SNR values, indicating its scalability and robustness.

Multi-user interference (MUI) performance versus the number of users for the proposed hybrid beamformer.

The proposed method achieves lower MUI levels compared to baseline approaches, demonstrating its effectiveness in managing interference in multi-user scenarios.

Subset Selection Based RIS-Aided Beamforming for Joint Radar-Communications

Sun, 01 Jan 2023 00:00:00 +0000

Key Contributions

Proposing a subset selection framework for optimizing beamforming weights in RIS-aided joint radar-communication systems.
Developing an efficient algorithm to select the optimal reflecting element subset while considering both radar and communication objectives.
Demonstrating significant performance improvements in terms of radar detection probability and communication throughput compared to conventional methods.

Results & Insights

The proposed algorithm consistently outperforms baseline methods across all SNR regimes, achieving higher detection probability and communication rate.

The algorithm exhibits rapid convergence even for large subset sizes, validating its computational efficiency.

Increasing the reflection coefficient magnitude improves both radar detection probability and communication rate, with diminishing returns beyond a certain threshold.

System performance improves with more reflecting elements, but the rate of improvement diminishes, indicating an optimal configuration size for the given setup.

Velocity-aided Channel Estimation for Spatially Selective mmWave Massive MIMO Communications

Sun, 01 Jan 2023 00:00:00 +0000

Key Contributions

Addresses the challenge of channel estimation in time-varying mmWave massive MIMO channels for moving UAVs, where conventional training-based methods fail due to beam squint and channel dynamics.
Proposes a novel channel estimation technique that leverages the UAV’s velocity information, obtained from its control module, to compensate for the beam squint effect and improve parameter recovery.
Introduces a time-varying channel model specifically designed for mmWave frequencies that incorporates the spatial selectivity (beam squint) effect, enabling more accurate and robust channel state information (CSI) estimation.
Demonstrates the effectiveness of the proposed method in providing reliable channel estimates even with short training sequences, essential for real-time UAV communications.

Results & Insights

This figure shows that the proposed velocity-aided method significantly outperforms conventional training-based approaches in terms of channel estimation accuracy (lower NMSE), especially at higher UAV velocities and moderate to high SNR levels, validating the effectiveness of incorporating velocity information.

The results indicate that the proposed method achieves lower BER for higher data rates and faster UAV speeds compared to baseline methods, demonstrating its practical benefit for reliable mmWave massive MIMO communication with moving platforms.

ADMM-based Cooperative Control for Platooning of Connected and Autonomous Vehicles

Sat, 01 Jan 2022 00:00:00 +0000

Key Contributions

Development of a distributed model-predictive control (DMPC) framework for vehicle platooning, incorporating consensus constraints for safety and coordination.
Formulation and solution of the DMPC problem using the Alternating Direction Method of Multipliers (ADMM), enabling low communication overhead and robust performance.
Evaluation of the proposed ADMM-based controller through extensive simulations, demonstrating its effectiveness in maintaining safe inter-vehicle distances and tracking the lead vehicle’s trajectory.

Results & Insights

Trajectories of all vehicles plotted against time, showing the following vehicles’ ability to closely track the leading vehicle’s path.

The simulation results in Fig. 3 demonstrate that the ADMM-based controller enables the platoon to maintain close proximity to the lead vehicle, with minimal tracking error across all vehicles, highlighting the controller’s effectiveness in achieving coordinated motion.

CDF plot of the distance between successive vehicles, illustrating the statistical distribution of inter-vehicle distances under the ADMM controller.

Fig. 5 reveals that the ADMM controller ensures a high probability (100%) of maintaining a safe inter-vehicle distance exceeding 1 meter, whereas other methods like unconstrained MPC and the low-complexity ADMM-L implementation show a significant risk of violating this safety margin.

Green Joint Radar-Communications: RF Selection with Low Resolution DACs and Hybrid Precoding

Sat, 01 Jan 2022 00:00:00 +0000

Key Contributions

Proposed an energy-efficient radio frequency (RF) chain and DAC bit selection approach for sub-arrayed hybrid MIMO joint radar-communication (JRC) systems.
Formulated a fractional energy efficiency (EE) maximization problem to determine the optimal number of RF chains and associated DAC bit resolution for communication scenarios.
Introduced a weighting formulation to represent the combined radar-communications performance, achieving high EE gains and a desirable radar beampattern performance.

Results & Insights

The proposed hybrid precoding approach with low-resolution DACs achieves significantly higher energy efficiency (EE) compared to the fully digital baseline, particularly at lower SNR levels, while maintaining comparable communication rates.

The proposed method achieves a highly directive radar beampattern with minimal sidelobes, comparable to conventional fully-digital systems, even with low-resolution DACs, ensuring effective radar functionality.

The proposed approach exhibits a clear energy-rate trade-off, allowing operators to balance communication throughput and energy consumption by adjusting the radar-communication traffic ratio (ρ), demonstrating its flexibility in practical deployments.

Optimal MmWave Sensor Selection for Bearing-Only Localization in Smart Environments

Sat, 01 Jan 2022 00:00:00 +0000

Key Contributions

Development of an optimal sensor selection technique for mmWave bearing-only localization.
Addressing the challenge of low-cost deployment in smart environments through efficient sensor usage.
Emphasizing privacy-preserving localization methods compared to traditional RF-based approaches.
Providing performance analysis through simulations under various scenarios.

Performance comparison of the proposed optimal sensor selection method versus baseline approaches across different signal-to-noise ratio (SNR) values.

The results demonstrate that the proposed method achieves superior localization accuracy with a reduced number of active sensors, highlighting its efficiency in smart environment applications.

Quantum Computing-Assisted Channel Estimation for Massive MIMO mmWave Systems

Sat, 01 Jan 2022 00:00:00 +0000

Key Contributions

Proposes a quantum computing-assisted framework for channel estimation in massive MIMO mmWave systems, leveraging quantum algorithms to address classical challenges.
Introduces a hybrid approach combining quantum computing with classical signal processing techniques to enhance estimation accuracy and reduce computational complexity.
Demonstrates the potential of quantum algorithms, such as HHL, to solve linear systems arising in channel estimation more efficiently than classical methods.
Provides insights into the practical implementation of quantum-assisted channel estimation, highlighting key performance gains and scalability considerations.

Towards 6G: Spectrally efficient joint radar and communication with radio frequency selection, interference and hardware impairments

Sat, 01 Jan 2022 00:00:00 +0000

Key Contributions

Development of a spectral efficiency maximization approach for inclined joint radar-communication systems with hybrid beamforming.
Design of an efficient selection algorithm for optimizing the number of RF chains while considering interference between radar and communication operations.
Incorporation of hardware impairments in the system model to achieve a practical communication-radar trade-off.

Energy-Efficiency Maximization of Hybrid Massive MIMO Precoding With Random-Resolution DACs via RF Selection

Fri, 01 Jan 2021 00:00:00 +0000

Key Contributions

Introduces an architecture with low-end components (reduced-resolution DACs) for hybrid massive MIMO precoding to maximize energy efficiency (EE) while minimizing the impact on spectral efficiency (SE).
Proposes a novel analog precoder design where the number of active RF chains is dynamically selected (RF selection) based on a criterion to balance EE and SE.
Develops an optimization framework that jointly designs the analog precoder structure (via RF selection) and the baseband precoding matrix to achieve EE maximization.

Results & Insights

Spectral efficiency (bits/sec/Hz/UE) plotted against transmit power (dBm) for different numbers of active RF chains.

This figure demonstrates that increasing transmit power generally improves spectral efficiency, but the rate of improvement saturates as more RF chains are activated, highlighting the diminishing returns of additional chains.

Energy efficiency (bits/Joule/UE) plotted against the number of transmission streams (Ns).

This plot shows a clear trade-off between energy efficiency and the number of streams; higher energy efficiency is achieved with fewer streams, but the system can support more users or data rates at the cost of reduced EE.

Joint Bit Allocation and Hybrid Beamforming Optimization for Energy Efficient Millimeter Wave MIMO Systems

Fri, 01 Jan 2021 00:00:00 +0000

Key Contributions

Introduces a novel decomposition of hybrid precoders and combiners into three matrices representing the analog precoder/combiner, DAC/ADC bit resolution matrix, and baseband precoder/combiner.
Proposes a joint optimization framework for hybrid beamforming matrices and DAC/ADC bit resolutions to maximize energy efficiency (EE).
Develops an efficient solution based on the alternating direction method of multipliers (ADMM) to solve the optimization problem at both transmitter and receiver.

Results & Insights

System architecture of the mmWave hybrid MIMO system with low-resolution DACs/ADCs, illustrating the three-part decomposition of the hybrid precoder/combiner and the beam training phase.

The proposed three-part decomposition enables joint optimization of analog beamforming, bit resolution allocation, and digital combining, forming the foundation of the energy-efficient end-to-end design.

A Hardware Architecture For Reconfigurable Intelligent Surfaces with Minimal Active Elements for Explicit Channel Estimation

Wed, 01 Jan 2020 00:00:00 +0000

Key Contributions

Introduces a novel hardware architecture for reconfigurable intelligent surfaces (RISs) with minimal active elements, utilizing a single RF chain for baseband measurements.
Proposes an alternating optimization approach for explicit channel estimation in the beamspace domain, leveraging the sparsity of wireless channels.
Demonstrates that the proposed architecture enables efficient online RIS configuration for specific environments through reduced training periods.

Results & Insights

The figure shows that the proposed technique achieves superior channel estimation accuracy with fewer training symbols compared to OMP and LS methods, highlighting its efficiency.

The results indicate that while the system achieves a rate of 8.5bps/Hz with perfect channel knowledge, the channel estimation process with T=200 training symbols results in a rate of approximately 7.6bps/Hz, demonstrating a trade-off between estimation accuracy and achievable rate.

Joint Localization and Channel Estimation for UAV-Assisted Millimeter Wave Communications

Wed, 01 Jan 2020 00:00:00 +0000

Key Contributions

Exploiting the line-of-sight (LOS) propagation conditions and beamspace representation of the wireless channel to estimate the UAV position.
Utilizing the estimated UAV position to formulate a matrix completion-based approach for recovering the non-line-of-sight (NLOS) components of the MIMO channel.
Demonstrating superior performance compared to baseline methods, requiring fewer channel training resources.

Results & Insights

NMSE performance of the proposed LOS channel estimation technique versus baseline methods.

The proposed method achieves significantly lower estimation error for the line-of-sight channel component across various scenarios, highlighting its effectiveness in leveraging UAV position information. The results demonstrate that the proposed approach maintains low estimation error even with reduced channel training resources, offering a more efficient solution for joint localization and channel estimation in UAV-assisted mmWave systems.

Radio-frequency chain selection for energy and spectral efficiency maximization in hybrid beamforming under hardware imperfections

Wed, 01 Jan 2020 00:00:00 +0000

Key Contributions

Development of a framework for hybrid beamforming that explicitly accounts for hardware imperfections, including phase noise and amplitude distortion.
Formulation of an optimization problem to jointly maximize energy efficiency and spectral efficiency, considering the impact of hardware imperfections.
Provision of numerical results demonstrating the performance trade-offs between energy and spectral efficiency under various system configurations and imperfection levels.

Wideband Channel Tracking for Millimeter Wave Massive Mimo Systems with Hybrid Beamforming Reception

Wed, 01 Jan 2020 00:00:00 +0000

Key Contributions

Proposed a matrix-completion-based channel tracking technique for time-correlated wideband mmWave massive MIMO channels with hybrid beamforming reception.
Formulated a constrained multi-objective optimization problem incorporating low-rank and group-sparse properties of the mmWave channel.
Capitalized on a time correlation model to reduce beam training overhead while maintaining estimation accuracy.
Developed an efficient algorithm based on the alternating direction method of multipliers (ADMM) for solving the optimization problem.

Results & Insights

The figure demonstrates that the proposed technique achieves near-optimal performance with significantly reduced training overhead, converging faster and achieving lower NMSE than OMP and AdaOMP for time-varying channels.

The results show that the proposed technique maintains low NMSE and achieves high achievable spectral efficiency even with short training sequences, highlighting its effectiveness in reducing beam training overhead while preserving channel estimation accuracy.

Dynamic RF Chain Selection for Energy Efficient and Low Complexity Hybrid Beamforming in Millimeter Wave MIMO Systems

Tue, 01 Jan 2019 00:00:00 +0000

Key Contributions

Proposing a novel hybrid beamforming architecture with dynamic RF chain selection to optimize energy efficiency.
Introducing a fractional programming framework based on the Dinkelbach method to determine the optimal number of active RF chains.
Developing a low-complexity codebook-based gradient pursuit (GP) algorithm for hybrid precoder design, outperforming state-of-the-art methods like OMP.
Demonstrating significant improvements in energy efficiency and reduced computational complexity compared to brute-force and fully digital beamforming approaches.

Performance comparison of energy efficiency across different hybrid beamforming techniques under varying SNR conditions.

This figure demonstrates that the proposed dynamic RF chain selection framework achieves superior energy efficiency compared to conventional hybrid beamforming methods, particularly at higher SNR levels.

Convergence behavior and bit error rate performance of the proposed Dinkelbach method-based optimization framework.

The results show rapid convergence of the optimization algorithm and significantly lower bit error rates compared to benchmark methods, validating the effectiveness of the proposed approach.

Power consumption analysis of the proposed framework versus conventional hybrid beamforming approaches across different SNR values.

The proposed dynamic RF chain selection reduces power consumption by up to 30% compared to fully-connected hybrid beamforming, especially at moderate SNR levels.

Computational complexity comparison of the proposed gradient pursuit algorithm versus state-of-the-art methods.

The proposed framework achieves substantially lower computational complexity (reduced by 40-60%) while maintaining comparable performance to more complex optimization techniques.

Energy Efficiency Maximization of Millimeter Wave Hybrid MIMO Systems with Low Resolution DACs

Tue, 01 Jan 2019 00:00:00 +0000

Key Contributions

Development of a novel optimization technique combining the Dinkelbach method with subset selection to maximize energy efficiency (EE) in mmWave hybrid MIMO systems with low-resolution DACs.
Consideration of the transmitter-side impact of low-resolution DACs, distinguishing this work from literature primarily focused on receiver-side ADCs.
Formulation of the EE maximization problem under a predefined power budget, accounting for the power consumption of active RF chains and their associated DACs.
Demonstration that the proposed technique achieves near-optimal EE performance with significantly reduced computational complexity compared to an exhaustive search baseline.

Presentation

Presenting the paper at IEEE ICC 2019, Shanghai, May 2019.

Energy Efficient ADC Bit Allocation and Hybrid Combining for Millimeter Wave MIMO Systems

Tue, 01 Jan 2019 00:00:00 +0000

Key Contributions

Introduces a novel decomposition of the hybrid combiner matrix into three parts: analog combiner, bit resolution matrix, and baseband combiner matrix.
Formulates the joint optimization problem as a matrix factorization task to approximate the fully digital combiner matrix.
Proposes an efficient solution using the alternating direction method of multipliers (ADMM) to solve the optimization problem.
Demonstrates that the approach achieves high energy efficiency by optimizing ADC bit allocation and hybrid combining jointly.

Results & Insights

No figures containing numerical results were provided.

Energy Efficient Transmission of 3D Meshes Over MMWave-Based Massive MIMO Systems

Tue, 01 Jan 2019 00:00:00 +0000

Key Contributions

Proposes a complexity-aware perceptual coding scheme for 3D meshes that minimizes reconstruction losses while considering hardware limitations.
Introduces an optimal bit allocation strategy that assigns different quantization resolutions to different levels of detail in the 3D mesh.
Demonstrates significant energy efficiency gains and improved reconstruction quality compared to uniform transmission approaches.

Reconstruction quality comparison between uniform transmission and the proposed bit allocation approach for different 3D meshes and bit-per-vertex values.

This figure shows that the proposed scheme provides superior reconstruction quality, particularly at lower bit rates, by adaptively allocating bits based on perceptual importance.

Comparison of uniform bit allocation versus optimized bit allocation for 3D mesh transmission.

The optimized bit allocation strategy significantly outperforms the uniform approach in terms of reconstruction quality, demonstrating the effectiveness of assigning different bit rates to different mesh details.

Performance (metric 6) versus total power consumption for different transmission approaches.

The results indicate that the proposed scheme achieves substantially better reconstruction quality at lower power levels, highlighting its energy efficiency benefits.

Hybrid Beamforming with Random Analog Sampling for Wideband Channel Estimation in Millimeter Wave Massive MIMO Systems

Tue, 01 Jan 2019 00:00:00 +0000

Key Contributions

Development of a novel wideband channel estimation technique for mmWave massive MIMO systems with hybrid beamforming reception.
Capitalization on matrix completion tools to exploit the sparsity and low-rank properties of wireless channels.
Proposal of an analog combining architecture featuring an extended combiner and random sub-sampler to enhance estimation accuracy.
Demonstration of improved performance in terms of MSE with reduced beam training length.

Results & Insights

This figure demonstrates that the proposed algorithm achieves significantly lower channel estimation error (NMSE) compared to benchmark methods, particularly at higher SNR levels, validating its effectiveness.

The results show that the proposed algorithm converges faster and achieves lower steady-state NMSE even with shorter training lengths, highlighting its efficiency in practical scenarios with limited beam training time.

This plot indicates that the proposed algorithm maintains superior performance even as the number of antenna elements increases, showcasing its scalability for massive MIMO systems.

Robust and Efficient Privacy Preservation in Industrial IoT via correlation completion and tracking

Tue, 01 Jan 2019 00:00:00 +0000

Key Contributions

Proposing a privacy-preserving method for reconstructing the PCA subspace from a subset of correlation values, ensuring that the correlation matrices cannot reveal sensitive information.
Introducing an adaptive matrix completion approach that solves a rank-one completion problem iteratively, addressing the issue of missing entries in distributed IIoT networks.
Demonstrating through extensive simulations that the proposed approach improves privacy preservation, computational efficiency, and robustness against node and link failures.

Results & Insights

The evolution of privacy discrepancy ratio with respect to the number of iterations for K = 20 nodes.

This figure shows that increasing the number of iterations during the reconstruction process significantly enhances privacy preservation, with a notable improvement observed even after a small number of iterations (e.g., 5 iterations). This figure indicates that the obfuscated data covariance matrix is accurately estimated even in sparse networks, demonstrating the effectiveness of the proposed method in maintaining data utility while preserving privacy.

Robust Estimator for Lens-based Hybrid MIMO with Low-Resolution Sampling

Tue, 01 Jan 2019 00:00:00 +0000

Key Contributions

Development of a robust maximum a posteriori (MAP) estimator based on the Expectation-Maximization (EM) algorithm for symbol estimation in lens-based hybrid MIMO systems with low-resolution ADCs.
The proposed estimator effectively mitigates the impairments introduced by low-resolution ADCs and the analog beamformer, improving performance in medium to high signal-to-noise ratio (SNR) regimes.
The work demonstrates that lower ADC resolution does not necessarily degrade system performance, offering a pathway to reduce hardware complexity and power consumption in mmWave MIMO systems.

Results & Insights

Performance comparison of the proposed MAP-EM estimator versus conventional methods in terms of NMSE for different ADC resolutions.

This figure shows that the proposed MAP-EM estimator significantly outperforms conventional methods, particularly at medium to high SNR, across various ADC resolutions, demonstrating its robustness and effectiveness.

Signal Processing on Static and Dynamic 3D Meshes: Sparse Representations and Applications

Tue, 01 Jan 2019 00:00:00 +0000

Key Contributions

Comprehensive review of fundamental signal processing models (RPCA, CS, MC) for geometry processing.
Analysis of scalable architectures and optimization algorithms for sparse representation tasks.
Demonstration of sparse modeling impact on 3D processing tasks (noise removal, outlier rejection, data completion).

Performance comparison of different sparse recovery methods on static 3D meshes.

This figure shows that model-based methods outperform traditional approaches in terms of reconstruction accuracy for static meshes.

Comparison of denoising results using PCA-based sparse representation versus conventional methods.

The results demonstrate that the proposed sparse representation method achieves superior feature preservation and noise reduction compared to standard techniques.

Wideband MIMO Channel Estimation for Hybrid Beamforming Millimeter Wave Systems via Random Spatial Sampling

Tue, 01 Jan 2019 00:00:00 +0000

Key Contributions

Development of a novel wideband channel estimation framework for mmWave massive MIMO systems employing hybrid beamforming, addressing the challenges of hardware limitations and short coherence time.
Joint exploitation of the low-rank property of the channel and available angular information to enhance channel recovery accuracy, particularly in short beam training scenarios.
Formulation of the estimation problem as a convex optimization task, leading to a computationally efficient solution via singular value thresholding and leveraging the inherent structure of the mmWave channel.

Adaptive Windowing for ICI Mitigation in Vehicular Communications

Mon, 01 Jan 2018 00:00:00 +0000

Key Contributions

Proposed an adaptive windowing technique for ICI mitigation in OFDM systems operating in vehicular environments.
The technique tracks time-varying channel statistics to maximize average SINR with linear computational complexity.
Demonstrated enhanced robustness to channel estimation errors, which are common in vehicular communication scenarios.

Results & Insights

The proposed technique shows superior ICI mitigation performance, particularly at high SNR, compared to existing fixed and adaptive windowing methods.

The adaptive windowing technique achieves lower bit error rates across various vehicular speeds, confirming its effectiveness in non-stationary channel conditions.

The algorithm demonstrates stable convergence with appropriate step-size selection, maintaining low computational complexity while adapting to channel variations.

The technique exhibits significantly better performance stability in the presence of channel estimation errors compared to conventional methods.

Distributed consolidation of highly-incomplete dynamic point clouds based on rank minimization

Mon, 01 Jan 2018 00:00:00 +0000

Key Contributions

Introduces a novel reconstruction technique based on rank minimization theory to address the challenge of highly incomplete dynamic point clouds, ensuring a unique solution to an otherwise ill-posed problem.
Extends the rank minimization approach to incorporate spatial coherence, which is essential for accurately reconstructing soft-body animations.
Proposes a distributed consolidation method that allows multiple nodes to collaboratively reconstruct point clouds by enforcing temporal coherence between frame clusters, improving scalability and efficiency.

Results & Insights

NMSE comparison of the proposed method against PCA-based reconstruction for different levels of incompleteness.

The results demonstrate that the proposed rank minimization-based method significantly outperforms the PCA-based approach in terms of reconstruction accuracy, particularly for highly incomplete point clouds, achieving lower NMSE across all tested configurations.

Dithered Beamforming for Channel Estimation in Mmwave-Based Massive Mimo

Mon, 01 Jan 2018 00:00:00 +0000

Key Contributions

Proposed a dithered beamforming architecture by injecting random control signals into the analog beamformer and ADCs to combat quantization effects.
Developed a low-complexity Expectation-Maximization algorithm for sparse channel estimation in hybrid beamforming systems with low-resolution ADCs.
Demonstrated that the technique effectively mitigates quantization non-linearities and improves channel estimation performance in mmWave massive MIMO systems.

Hybrid analog-digital receiver with phase dithering in the analog RF combiner and quantization dithering before the ADCs.

This figure shows that the proposed dithered beamforming technique significantly reduces estimation error compared to non-dithered approaches, achieving near-optimal performance with a 1dB improvement for 2- and 3-bit quantization cases. The results demonstrate rapid convergence of the proposed algorithm, approaching near-optimal performance even with an approximate solution for the sparse channel estimation problem.

Efficient Channel Estimation in Millimeter Wave Hybrid MIMO Systems with Low Resolution ADCs

Mon, 01 Jan 2018 00:00:00 +0000

Key Contributions

Proposes an EM-SURE-GAMP algorithm that combines compressed sensing with expectation-maximization and adaptive denoising to exploit channel sparsity.
Develops a solution that adapts to low-resolution ADCs by using an additive quantization noise model (AQNM) and minimizes estimation error through Stein’s Unbiased Risk Estimate (SURE).
Demonstrates significant performance improvements in MSE across various SNR regimes, ADC bit resolutions, and training lengths compared to existing methods like EM-GAMP.

Results & Insights

Mean square error (MSE) versus signal-to-noise ratio (SNR) for different channel estimation methods.

The proposed EM-SURE-GAMP algorithm achieves lower MSE across all SNR regimes compared to the EM-GAMP baseline, particularly at low and high SNR.

Mean square error (MSE) versus the number of ADC bits for different channel estimation methods.

EM-SURE-GAMP consistently outperforms EM-GAMP as the number of ADC bits decreases, demonstrating robustness to quantization effects.

Comparison of channel estimation performance across different training lengths.

The algorithm maintains superior performance even with shorter training sequences, highlighting its efficiency and adaptability to practical constraints.

Illustration of convergence behavior for the EM-SURE-GAMP algorithm.

The EM-SURE-GAMP algorithm converges faster and achieves lower MSE compared to EM-GAMP, indicating its practical suitability for real-time applications.

Energy Efficient Transmitter with Low Resolution DACs for Massive MIMO with Partially Connected Hybrid Architecture

Mon, 01 Jan 2018 00:00:00 +0000

Key Contributions

Formulated an optimization problem to minimize transmitter power consumption by determining the optimal number of RF chains for mmWave massive MIMO systems.
Ensured the system capacity meets a minimum rate requirement while reducing the number of RF chains through a partially connected hybrid architecture.
Demonstrated that the proposed transmitter with low-resolution DACs achieves higher spectral efficiency and up to 30% improved energy efficiency compared to conventional systems with the same number of active RF chains.

Massive MIMO Channel Estimation for Millimeter Wave Systems via Matrix Completion

Mon, 01 Jan 2018 00:00:00 +0000

Key Contributions

Proposes a novel channel estimation algorithm for mmWave massive MIMO systems that jointly exploits both channel sparsity and low-rank properties via matrix completion.
Introduces an ADMM-based iterative optimization approach that converges quickly even for very small training lengths.
Demonstrates superior performance compared to state-of-the-art techniques, achieving better accuracy with significantly shorter training periods.
Enables reliable channel recovery in mmWave systems, which are characterized by high variability and challenging conditions.

Results & Insights

This figure clearly demonstrates the superior performance of the proposed technique across all training lengths, with the proposed method consistently achieving better accuracy than existing methods like SVT and VAMP, even at the smallest training lengths.

The proposed algorithm converges faster and achieves lower NMSE for all training lengths, indicating its efficiency and effectiveness in channel estimation.

This result highlights the proposed technique’s ability to maintain good performance even with very short training sequences, outperforming other methods significantly for all considered training lengths.

Autonomous driving in 5G: Mitigating interference in OFDM-based vehicular communications

Sun, 01 Jan 2017 00:00:00 +0000

Key Contributions

Development of an OSIC-based interference cancellation solution for OFDM systems in vehicular environments.
Introduction of a CG-based mechanism for efficient equalizer filter estimation, accelerating algorithm convergence.
Demonstration of superior performance compared to existing non-banded ICI cancellation methods, achieving lower BER with reduced complexity.

Results & Insights

Bit Error Rate (BER) performance comparison of the proposed OSIC-based equalizer with existing methods for 4QAM modulation in doubly selective channels.

The results show that the proposed solution achieves significantly lower BER across various signal-to-noise ratio (SNR) conditions compared to existing non-banded ICI cancellation methods, validating its effectiveness in improving communication reliability for V2X systems.

Compressed Sensing for Efficient Encoding of Dense 3D Meshes Using Model-Based Bayesian Learning

Sun, 01 Jan 2017 00:00:00 +0000

Key Contributions

Propose a novel compressed sensing framework for 3D mesh geometry compression using Bayesian learning, enabling aggressive compression ratios without compromising visual quality.
Introduce a multivariate Gaussian model for the most significant graph Fourier transform coefficients of each submesh, derived via expectation-maximization.
Leverage local smoothness of partitioned surfaces to enhance performance in highly underdetermined problems, reducing encoding complexity while maintaining reconstruction accuracy.

Results & Insights

Performance comparison of VNMSE versus compression ratio for the chair model across different methods.

This plot demonstrates the effectiveness of the proposed method, showing competitive compression ratios with significantly lower encoding complexity compared to state-of-the-art approaches, particularly at higher compression ratios.

Efficient graph-based matrix completion on incomplete animated models

Sun, 01 Jan 2017 00:00:00 +0000

Key Contributions

Proposing a graph-based regularization technique for matrix completion in animated models, leveraging structural information for improved accuracy.
Introducing an efficient optimization algorithm that reduces computational complexity while maintaining high reconstruction quality.
Demonstrating the effectiveness of the approach on incomplete animated model data, outperforming baseline methods in terms of accuracy and robustness.

Results & Insights

The proposed method consistently achieves higher accuracy across different datasets, highlighting the effectiveness of the graph-based regularization.

The results indicate that the method maintains low error rates even with significant data missing, showcasing its robustness.

The proposed method converges faster and to a lower error, demonstrating the efficiency of the optimization algorithm.

The method achieves significantly lower computational time, making it suitable for large-scale applications.

The method exhibits superior performance in noisy conditions, maintaining lower error rates compared to alternatives.

Adaptive completion of the correlation matrix in wireless sensor networks

Fri, 01 Jan 2016 00:00:00 +0000

Key Contributions

Development of a privacy-preserving framework for decentralized correlation matrix estimation in wireless sensor networks.
Proposal of a novel adaptive rank-one matrix completion technique to handle the constrained cooperation between sensor nodes.
Demonstration of significantly faster convergence and lower computational complexity compared to traditional singular value thresholding (SVT) methods.

Convergence of the proposed algorithm showing the estimation error decreasing rapidly over iterations, reaching the ground-truth value after a small number of steps, while the SVT-based method converges much slower.

The results demonstrate that the proposed adaptive completion technique achieves rapid convergence to the true correlation matrix with substantially lower computational overhead, making it particularly suitable for resource-constrained wireless sensor networks with privacy constraints.

Blind distributed beamforming via matrix completion

Fri, 01 Jan 2016 00:00:00 +0000

Key Contributions

Addresses the challenge of distributed beamforming with incomplete channel statistics by leveraging matrix completion techniques.
Proposes a low-complexity iterative method for estimating correlation matrices at the fusion center using partial measurements.
Demonstrates that the technique converges to the optimal beamforming solution with a limited number of relays providing direct communication.

Results & Insights

Convergence curves for matrix completion of correlation matrices R(t) and G(t) versus time, showing NMSE reduction for different numbers of known entries.

These results demonstrate that the proposed matrix completion technique achieves near-optimal performance with significantly fewer known entries (34% for K=10), highlighting its efficiency in recovering the full correlation matrices needed for optimal beamforming.

Separate NMSE convergence of correlation matrices R(t) (top) and G(t) (bottom), comparing full-entry recovery against logarithmic subsampling strategies.

The results show that both matrices converge reliably using only O(K log K) known entries, validating the scalability and practicality of the matrix completion approach for distributed beamforming.

Low-complexity OSIC equalization for OFDM-based Vehicular Communications

Fri, 01 Jan 2016 00:00:00 +0000

Key Contributions

Proposing an OSIC equalization technique tailored for doubly selective vehicular channels to mitigate ICI effectively.
Introducing an iterative Galerkin projection-based algorithm that reduces computational complexity while maintaining performance.
Demonstrating the benefits of the proposed scheme in terms of both performance and complexity through simulations in high Doppler environments.

Results & Insights

Complexity order of the proposed ICI equalization method versus standard vehicular communication standards (V2V, Wi-Fi, WiMAX, LTE, DVB-T) as a function of Doppler spread and OFDM size N.

The proposed Galerkin-based equalizer achieves O(N) complexity for typical vehicular channels, making it significantly more efficient than full ICI cancellation approaches that require O(N²) or O(N⁴) operations.

BER versus SNR comparison of the proposed OSIC equalizer against benchmark schemes for low Doppler spread and 4-QAM modulation.

The proposed method closely tracks the performance of full OSIC while operating at a fraction of the complexity, demonstrating its practical viability for vehicular OFDM systems.

BER versus SNR comparison for medium and high Doppler spread scenarios, illustrating robustness of the proposed equalizer under severe ICI conditions.

The results confirm that the Galerkin projection approach maintains performance gains over banded approximation benchmarks even as Doppler spread increases, validating its suitability for high-mobility environments.

Supervised energy disaggregation using dictionary-based modelling of appliance states

Fri, 01 Jan 2016 00:00:00 +0000

Key Contributions

Proposed a supervised energy disaggregation method using multi-state finite state machines to model appliance behavior.
Introduced a dictionary-based approach where appliance states are represented by power consumption vectors selected via clustering from training data.
Utilized state transition probabilities and a reduced trellis structure to efficiently perform disaggregation during operation.

Results & Insights

The NMSE performance improves with an increasing number of appliances, demonstrating the method’s scalability and effectiveness on real-world data.

The proposed method achieves lower NMSE compared to existing approaches, highlighting its superior accuracy in energy disaggregation.

A decentralized approach to generalized power system state estimation

Tue, 01 Jan 2013 00:00:00 +0000

Key Contributions

Development of a decentralized generalized state estimation (GSE) approach that verifies circuit breaker statuses.
Utilization of block-sparsity promoting regularizers to enhance breaker status verification capabilities.
Formulation of the GSE problem as a distributed optimization problem amenable to the alternating direction method of multipliers (ADMM).
Design of a communication-efficient framework that maintains compatibility with existing power system solvers.

The test power system partitioned into four control areas for decentralized state estimation.

This figure demonstrates that the decentralized algorithm achieves estimation accuracy comparable to the centralized solution within a few iterations. The results show that the proposed decentralized `2-regularized GSE significantly reduces the number of incorrect breaker status identifications compared to the ordinary GSE approach.

A scheme for X-ray medical image denoising using sparse representations

Tue, 01 Jan 2013 00:00:00 +0000

Key Contributions

Proposes a novel denoising scheme for X-ray images leveraging sparse and redundant representations over learned dictionaries.
Investigates the impact of different dictionary types (fixed, trained on natural images, trained on X-ray images, trained on noisy images) on denoising performance.
Evaluates the effect of dictionary size and sparsity level on denoising results.

Results & Insights

Comparison of denoising performance across different dictionary types (DCT, general K-SVD, specialized K-SVD, noisy K-SVD) versus noise standard deviation.

This plot demonstrates that specialized and noisy dictionaries generally outperform fixed DCT dictionaries, especially at higher noise levels.

Comparison of denoising performance across different dictionary types (DCT, general K-SVD, specialized K-SVD, noisy K-SVD) versus the number of nonzero coefficients in the representation.

This plot shows that increasing the sparsity level generally improves denoising performance, with the noisy dictionary consistently performing best.

Comparison of different dictionaries in terms of PSNR of the denoised image versus the size of image patch.

This plot indicates that larger patch sizes (e.g., 8x8) generally lead to better denoising performance across all dictionary types.

Comparison of different dictionaries in terms of PSNR of the denoised image versus the number of nonzero elements of the representation vector.

This plot reveals that increasing the number of nonzero elements (sparsity) improves denoising performance, with the noisy dictionary showing the highest PSNR.

Example denoising results showing the improvement in Contrast to Noise Ratio (CNR) and visual quality.

This figure visually demonstrates the effectiveness of the proposed method in enhancing image contrast and revealing finer details like calcifications. This figure provides a qualitative assessment of the denoising performance, clearly showing the enhanced clarity and reduced noise in the reconstructed image.

Distributed blind adaptive computation of beamforming weights for relay networks

Tue, 01 Jan 2013 00:00:00 +0000

Key Contributions

Proposed a distributed blind adaptive algorithm for coordinated beamforming in multi-cell networks with multiple receive antennas.
Designed the algorithm to maximize the harmonic sum of SINR, ensuring fairness among users.
Achieved significant throughput gains over traditional single-cell beamforming schemes.

Results & Insights

The proposed algorithm demonstrates substantial performance improvements, achieving 70% and 47% throughput gains for the 10th percentile and median users, respectively, compared to single-cell beamforming.

The algorithm converges quickly, stabilizing SINR values within a few iterations, ensuring practical implementation in real-time systems.

The proposed algorithm maintains superior performance even as the number of receive antennas increases, unlike existing methods which degrade in large-scale MIMO systems.

Galerkin projections-based ICI cancellation in OFDM systems with doubly selective channels

Tue, 01 Jan 2013 00:00:00 +0000

Key Contributions

Proposing two low-complexity iterative successive interference cancellation (SIC) schemes based on Galerkin projections for ICI cancellation in doubly selective OFDM channels.
Deriving a reduced-rank CG-based SIC technique and improving it using Galerkin projections theory to achieve significant complexity reduction.
Demonstrating through simulations that the proposed schemes offer improved complexity-performance tradeoffs compared to existing methods, particularly for medium Doppler spread scenarios.

Results & Insights

This figure illustrates the effectiveness of the proposed Galerkin projection-based SIC schemes in reducing ICI power across different Doppler spreads, showing superior performance compared to benchmark methods.

The convergence plots demonstrate the fast convergence of the proposed iterative schemes, indicating their practical suitability for real-time OFDM receivers even under challenging channel conditions.

This plot highlights the significant complexity reduction achieved by the proposed Galerkin projection-based schemes, particularly evident when the Doppler spread is moderate compared to the OFDM symbol length.

The Bit Error Rate (BER) results confirm that the proposed schemes maintain high performance levels while drastically reducing computational complexity, outperforming traditional methods in the relevant operating regimes.

Regularized MMSE ICI equalization for OFDM systems over doubly selective channels

Tue, 01 Jan 2013 00:00:00 +0000

Key Contributions

Development of a regularized MMSE ICI equalization technique for doubly selective channels to mitigate the effects of Doppler-induced ICI in OFDM systems.
Proposing a banded approximation for the channel matrix to reduce computational complexity while maintaining performance.
Demonstrating that the regularized estimation can recover performance similar to exact MMSE equalization, offering a balance between performance and computational efficiency.

Results & Insights

Comparison of BER versus SNR for different equalization schemes, showing the performance of the proposed regularized MMSE equalizer.

This figure demonstrates that the proposed regularized MMSE equalizer achieves performance comparable to other advanced methods like iterative MMSE and decision-feedback equalizers, particularly at higher SNR levels.

Comparison of BER versus normalized Doppler frequency for the proposed regularized MMSE equalizer and other methods.

This result highlights the robustness of the proposed regularized MMSE equalizer against frequency-selective fading, showing lower BER across different Doppler conditions compared to the benchmark methods.

Comparison of BER versus the length K of the banded approximation matrix, with SNR fixed at 30 dB.

This figure confirms that increasing the length K improves the BER performance, approaching the performance of the exact MMSE equalizer, while the banded approximation effectively reduces computational complexity.

Compressed Sensing Techniques for Decision Feedback Equalization of Sparse Wireless Channels

Sun, 01 Jan 2012 00:00:00 +0000

Key Contributions

Development of compressed sensing-based decision feedback equalization (DFE) schemes for sparse wireless channels.
Exploitation of channel sparsity to improve the performance of non-sparsity-aware DFE.
Proposing new DFE implementations using basis pursuit and matching pursuit techniques.

Results & Insights

Performance comparison of indirect DFE techniques under different SNR conditions.

The results show that the proposed compressed sensing-based DFE schemes outperform traditional methods, especially at higher SNR levels.

SER performance of direct equalization schemes as a function of training sequence length.

The SER performance improves with longer training sequences, indicating the importance of training for accurate channel estimation.

SER performance of DFE schemes across different SNR values.

The proposed CS-based DFE schemes demonstrate superior performance across all tested SNR levels, confirming their effectiveness.

Stochastic Gradient Pursuit for Adaptive Equalization of Sparse Multipath Channels

Sun, 01 Jan 2012 00:00:00 +0000

Key Contributions

Proposed a novel heuristic algorithm, termed Stochastic Gradient Pursuit, for sparse adaptive equalization.
Utilized a decision-feedback equalization structure to effectively handle long multipath channels.
Exploited channel sparsity from a compressive sensing perspective, developing an extension for unknown sparsity order.
Offered reduced computational complexity while maintaining performance comparable to conventional methods.

Results & Insights

Mean-square error (MSE) learning curves comparing the proposed Stochastic Gradient Pursuit algorithm against several benchmark methods.

These curves demonstrate the fast convergence and improved tracking capabilities of the proposed algorithm compared to conventional and other sparse-aware equalization schemes.

Mean-square error (MSE) learning curves for different sparsity orders in the proposed Stochastic Gradient Pursuit algorithm.

These results show the algorithm’s effectiveness in tracking sparse multipath channels even when the exact sparsity order is unknown.

Bit error rate (BER) performance of the proposed Stochastic Gradient Pursuit algorithm compared to benchmark methods.

The proposed algorithm achieves BER performance comparable to benchmark methods while requiring significantly less computational complexity.

Bit error rate (BER) performance versus input SNR for the proposed Stochastic Gradient Pursuit algorithm.

The results indicate that the proposed algorithm maintains good performance across a range of SNR values, demonstrating its robustness.

Greedy algorithms for sparse adaptive decision feedback equalization

Sat, 01 Jan 2011 00:00:00 +0000

Key Contributions

Proposing two novel adaptive greedy algorithms for sparse decision feedback equalization (DFE) in communication channels.
Developing a channel estimation-based approach that first identifies significant channel taps using a greedy algorithm and then adaptively computes the equalizer filters.
Demonstrating significantly improved performance in convergence speed, tracking capabilities, and reduced complexity compared to non-sparsity-aware methods and prior sparse-aware algorithms.
Offering a direct adaptive sparse equalizer based on a steepest descent (SD)-based greedy algorithm.

Results & Insights

The results demonstrate that the proposed schemes exhibit significantly improved tracking capabilities when channel parameters change dynamically, as shown in Figure 1(c). The algorithms maintain better performance and adapt faster to the evolving channel conditions compared to existing methods like the one in [7].