Manav R. Bhatnagar

Background/Objectives: Interpretable ECG rhythm classification remains challenging because of variability in recording conditions, patient characteristics, and annotation practices across datasets. Although deep learning models can achieve high classification performance, their limited interpretability and sensitivity to domain shift can restrict their applicability in real-world clinical settings. This study investigates an interpretable and computationally efficient framework for ECG rhythm abnormality detection using statistical descriptors of RR interval variability. Methods: Mean RR, SDNN, RMSSD, pNN50, and heart rate features were extracted from ECG recordings to characterize beat-to-beat temporal variability. A Random Forest classifier was employed for binary rhythm classification using the PhysioNet/CinC 2017 dataset, followed by cross-dataset testing on the ECG-SDS dataset, which consists of synthetic ECG recordings. Same-dataset evaluation was additionally performed on ECG-SDS to establish a same-dataset performance benchmark. Feature ablation and feature-level sensitivity analyses using controlled Gaussian perturbations were conducted to examine feature contribution and classifier sensitivity to perturbations in the extracted feature representation. Multi-class experiments were also performed by incorporating atrial fibrillation and sinus tachy-brady rhythms. Results: In binary cross-dataset evaluation, the proposed framework achieved an accuracy of 78.6%, abnormal rhythm recall of 74.2%, and ROC-AUC of 0.836 on ECG-SDS. The same-dataset ECG-SDS evaluation achieved an accuracy of 93% and ROC-AUC of 0.89. Ten-fold stratified cross-validation yielded a mean accuracy of 92.20% ± 2.63% and mean AUC of 0.8880 ± 0.0477. Feature ablation showed only modest performance changes when individual descriptors were removed. The exploratory same-dataset multi-class evaluation on ECG-SDS achieved an accuracy of 91.67%. Conclusions: The results demonstrate that compact RR interval statistical descriptors can provide an interpretable and computationally efficient representation for ECG rhythm classification. The reduction in performance during cross-dataset testing may reflect dataset shift and differences in model configuration. These cross-dataset results should be interpreted as a technical assessment of transferability rather than as validation in an independent clinical population. Further validation using independent real-world clinical ECG datasets is required before conclusions regarding clinical generalizability or screening applications can be drawn.

We study pilot-aided channel estimation for free-space optical (FSO) links operating under atmospheric turbulence and intentional interference. Both the legitimate and jamming paths experience gamma-gamma fading, while the jammer follows an on–off (Bernoulli) activity model that yields non-Gaussian, impulsive disturbances at the receiver. Building on an exact likelihood for this setting, we derive a steepest-descent maximum-likelihood (ML) estimator for the main channel coefficient and extend it to a maximum a posteriori (MAP) form that incorporates gamma-gamma priors. Monte Carlo simulations across multiple turbulence conditions ( α , β ), signal-to-noise ratios (SNRs), signal-to-jamming ratios (SJRs), and jammer activity probabilities demonstrate that the proposed estimators achieve consistently lower mean square error (MSE) and bit error rate (BER) than conventional mean-based or Gaussian-assumed baselines, with graceful degradation as jamming becomes more frequent and clear gains as the SJR improves. The results highlight that respecting the true gamma-gamma statistics rather than relying on Gaussian surrogates materially improves estimation robustness in hostile FSO environments, while the gradient-based forms remain simple enough for practical implementation.

Vehicle clustering improves communication efficiency, safety, and comfort in connected vehicular networks. Cluster heads (CHs) manage critical tasks such as handoff (HO), coordination, and resource allocation. However, in overlapping regions, CH-based group HO handling is hindered by high traffic and limited channel access, causing delays, signaling overhead, and reduced bandwidth efficiency. These issues increase HO latency and risk congestion and HO failures. To address HO loading, latency in overlapping regions, and poor convergence in dynamic environments, this paper proposes an evolutionary game-theoretic framework that selects multiple HO agents (HOAs) in overlapping areas. These agents form HO clusters to distribute load and mitigate channel contention, achieving up to 72.21% lower delay compared to conventional CH-based HO. The buffering of HO data at HOAs, necessitated by the distinct variability of vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) links, is addressed through a queueing-theoretic model. This is followed by optimal resource block (RB) allocation, minimizing total waiting time to 2.063 ms at optimal RB values for two HO-clusters and enhancing network stability. At an average vehicle speed of 140 km/h, the proposed HOA-based HO achieves a total average HO delay of 0.91 ms, corresponding to an 83.8% reduction compared with direct V2I HO. Analytical proof of equilibrium and Lyapunov-based stability validation ensure robustness in dynamic environments. It is observed that, under high incentive, a 12.5% increase in vehicle velocity slows the convergence rate by approximately 60%, primarily due to increased channel variability at higher speeds.

The co-existence of communication and radar systems has emerged as a promising approach to address spectrum scarcity in 5G and beyond networks. This paper investigates a power-domain non-orthogonal multiple access (PD-NOMA)-based joint communication-radar framework that utilizes the same resource block to enhance spectral efficiency, while simultaneously introducing mutual cross-interference challenges.We propose three co-existence strategies: a communication-centric approach that maximizes the individual user rates in a two-userNOMAscenario, a radar-centric approach that maximizes the radar signal-to-interference-plus-noise ratio (SINR), and a joint strategy that balances both radar and communication objectives. The joint success probability (JSP), defined as the probability of simultaneously satisfying both radar and communication constraints, is derived together with a high-SINR approximation that offers deeper analytical insights into system performance. The JSP formulation is further extended to account for correlated log-normal shadowing in the canonical two-user setting and, separately, to enable scalability to a multi-user downlink NOMA system. Furthermore, radar target detection under communication interference is formulated as a binary hypothesis testing problem. Two optimal detection thresholds are derived: the Neyman-Pearson (NP) threshold for a fixed false-alarm probability and the minimum-error threshold, which minimizes the total probability of detection error. A closed-form relationship between the two thresholds is established, revealing a scaling dependence on the effective radar SINR. The analysis assumes coordinated communication-radar co-existence, while uncoordinated operation is examined as a benchmark to quantify performance degradation due to unknown communication interference. Moreover, the optimal power allocation factor (PAF) that maximizes the JSP is obtained. The simulation results validate the theoretical analysis and demonstrate the effectiveness of the proposed framework in achieving a reliable communication-radar spectrum sharing solution to overcome the spectrum congestion problem.

This work presents a novel neural network-assisted multiple-input multiple-output (MIMO) transmission framework that extends the traditional Alamouti space-time block code by incorporating learned orthogonal waveforms. In the proposed system, two independent Alamouti blocks are transmitted over distinct neural network-generated waveforms instead of conventional sinusoidal carriers. These waveforms are adaptively learned during training to optimize end-to-end communication performance while explicitly enforcing mutual orthogonality. We propose two neural network–based MIMO transmission strategies using learned orthogonal waveforms. Strategy I employs two parallel Alamouti blocks transmission from four antennas in two time slots to achieve a code rate of two and diversity order two, while Strategy II uses a unified 4× 4 space-time block code, offering full diversity order four and a code rate of one. We analyze bit error rate (BER) performance under varying waveform angles, regularization weights, and signal-to-noise ratio levels. Results show that while Strategy I is relatively robust to imperfect orthogonality, Strategy II is significantly more sensitive to imperfect orthogonality but give better BER performance as compared to strategy I. The proposed methods offer improved efficiency and BER, highlighting the trade-offs between code rate, diversity, and waveform separation in neural MIMO systems. This hybrid of classical communication theory and deep learning introduces a new paradigm for adaptive, intelligent waveform design in modern wireless systems.

This paper addresses the problem of masking effect in satellite communication by incorporating the benefits of aerial intelligent reflecting surface (AIRS). In this paper a cooperative dual-AIRS aided satellite system has been considered where a geostationary (GEO) satellite transmits signal to a masked ground node through two cooperative AIRS nodes. The AIRSs are considered to move in a circular trajectory. Realistic physical environment and heterogeneity of space, air and ground are taken into consideration to model channel fading and large scale path loss of space-to-air, air-to-air and air-to-ground links of the proposed model. The space-to-air link is modeled using Shadowed Rician distribution, while LoS channel is considered between two AIRS nodes and Nakagami-$m$ fading between aerial and ground node. With these channel models, we derive the closed-form expressions for probability density function, cumulative distribution function and moment generating function of the end-to-end signal-to-noise ratio (SNR) in the form of generalized-$K$ ($K_{G}$) distribution. Using these expressions, we further derive the closed-form expressions for exact and asymptotic error probability, outage probability and ergodic capacity. The accuracy of $K_{G}$ distribution is validated through Kullback-Leibler (KL) divergence measure. Furthermore, simulation results are provided to validate the accuracy of derived expressions and investigate the effect of satellite elevation, number of reflecting elements, impact of imperfect CSI and mobility of AIRSs on the performance of the system. The performance is also compared with decode-and-forward (DF) UAV relay aided and single AIRS aided satellite systems and the comparison shows that despite of increased multiplicative pathloss the proposed dual AIRS aided system outperforms both the aforementioned schemes.

We study pilot-aided channel estimation for free-space optical (FSO) links operating under atmospheric turbulence and intentional interference. Both the legitimate and jamming paths experience Gamma-Gamma fading, while the jammer follows an on-off (Bernoulli) activity model that yields non-Gaussian, impulsive disturbances at the receiver. Building on an exact likelihood for this setting, we derive a steepest-descent maximum-likelihood (ML) estimator for the main channel coefficient and extend it to a maximum a posteriori (MAP) form that incorporates Gamma-Gamma priors. We also explore a lightweight hybrid scheme that refines the steepest-descent algorithm (SDA) output with a shallow neural corrector. Monte-Carlo simulations across multiple turbulence conditions (α,β), signal-to-noise ratios (SNRs), signal-to-jamming ratios (SJRs), and jammer activity probabilities demonstrate that the proposed estimators achieve consistently lower mean square error (MSE) and Bit Error Rate (BER) than conventional mean-based or Gaussian-assumed baselines, with graceful degradation as jamming becomes more frequent and clear gains as SJR improves. The results highlight that respecting the true Gamma-Gamma statistics rather than relying on Gaussian surrogates materially improves estimation robustness in hostile FSO environments, while the gradient-based forms remain simple enough for practical implementation.

We study pilot-aided channel estimation for free-space optical (FSO) links operating under atmospheric turbulence and intentional interference. Both the legitimate and jamming paths experience Gamma-Gamma fading, while the jammer follows an on-off (Bernoulli) activity model that yields non-Gaussian, impulsive disturbances at the receiver. Building on an exact likelihood for this setting, we derive a steepest-descent maximum-likelihood (ML) estimator for the main channel coefficient and extend it to a maximum a posteriori (MAP) form that incorporates Gamma-Gamma priors. We also explore a lightweight hybrid scheme that refines the steepest-descent algorithm (SDA) output with a shallow neural corrector. Monte-Carlo simulations across multiple turbulence conditions (α,β), signal-to-noise ratios (SNRs), signal-to-jamming ratios (SJRs), and jammer activity probabilities demonstrate that the proposed estimators achieve consistently lower mean square error (MSE) and Bit Error Rate (BER) than conventional mean-based or Gaussian-assumed baselines, with graceful degradation as jamming becomes more frequent and clear gains as SJR improves. The results highlight that respecting the true Gamma-Gamma statistics rather than relying on Gaussian surrogates materially improves estimation robustness in hostile FSO environments, while the gradient-based forms remain simple enough for practical implementation.