ABSTRACT Millimetre wave (mmWave) multiple‐input multiple‐output (MIMO) has emerged as a promising physical layer solution to address traffic demands in beyond 5G wireless communication systems. However, modern signal processing techniques for mmWave hybrid MIMO systems fall short of addressing the performance degradations due to residual transceiver hardware impairments (HIs). This paper thus considers a mmWave hybrid MIMO system with residual HIs. Using Bussgang decomposition, the residual transceiver HIs are modelled as an additive non‐Gaussian noise that severely affects the received pilot and information signals, which makes channel state acquisition challenging. In this context of channel‐estimation over non‐Gaussian noise due to HIs, this work presents a hyperparameter‐free maximum Versoria criterion (MVC)‐based channel estimation technique. In details, the MVC‐based channel‐estimator is rendered hyperparameter‐free by proposing a sampling rule for its spread parameter and a gradient descent‐based optimisation for its shape parameter . Finally, simulations are presented to show the generalisation and scenario‐invariance of the proposed MVC‐based channel‐estimator. The analytical expression for steady‐state misadjustment is also derived and validated by simulations.

Multiple-input multiple-output (MIMO) millimeter wave (mmWave) is a promising technology for beyond 5 G communication systems, which provides high data rates, ultra-low latency, massive connectivity, and high spectral efficiency. However, a typical mmWave communication link faces major challenges due to signal loss over distance and imperfections in the hardware, which cause distortions that don't follow the usual Gaussian pattern. Furthermore, MIMO mmWave channel exhibits sparseness due to blockage and scattering. The conventional orthogonal matching pursuit (OMP), zero-attracting least mean square (ZA-LMS), and their variants deliver suboptimal performance for MIMO-mmWave systems impaired by impulsive noise. In this letter, we propose robust zero-attracting logarithmic hyperbolic cosine adaptive filtering (ZA-LHCAF) for MIMO-mmWave channel estimation. Furthermore, we derive a sampling rule for spread parameter of ZA-LHCAF, which renders the proposed estimator hyperparameter free. The simulation results show that the proposed algorithm outperforms existing state-of-the-art estimators. Lastly, convergence analysis is presented for the proposed ZA-LHCAF algorithm, which validated by realistic computer simulations.

Quaternion-based machine learning algorithms have emerged as a promising candidate for four dimensional (4D) signal processing, which utilize 4D signal dimensions’ correlations for estimation purposes. The existing quaternion least mean square (QLMS) and quaternion maximum correntropy criterion (QMCC) algorithms are based on stochastic gradient descent and gradient ascent methods, respectively. Due to their reliance on instantaneous gradient approximations, these approaches exhibit slower convergence characteristics. To circumvent this limitation, quaternion recursive least square (QRLS) algorithm is proposed in the literature, which exhibits fast convergence and achieves lower MSE. However, the QRLS algorithm is based on minimum MSE (MMSE) criterion which optimizes error-energy, and is thus not a sufficient statistic for impulsive noise. In this paper, we propose novel non-MMSE criterion based recursive algorithms, namely quaternion recursive maximum correntropy criterion (QRMCC) and quaternion recursive maximum Versoria criterion (QRMVC). The proposed QRMCC and QRMVC algorithms exhibit robustness against impulsive/non-Gaussian distortions due to the inclusion of higher order error statistics. Simulations are presented for quaternion-valued channel estimation, which indicate that the proposed QRMCC and QRMVC algorithms deliver improved convergence compared to existing approaches. Lastly, the convergence of the proposed algorithms is analyzed with performance-validation using relevant computer simulations.

Information theoretic learning (ITL) criteria have emerged useful for mitigating degradations caused by unknown non-Gaussian noise processes in future wireless communication systems. Specifically, the reproducing kernel Hilbert space (RKHS) based approaches relying on ITL based learning criteria are envisioned to provide near-optimal mitigation of unknown hardware impairments and non-Gaussian noises. Among several ITL criteria, the recent works find the minimum error entropy with fiducial points (MEEF) promising due to its guarantee of unbiased estimation and generalization over generic noise distributions. However, MEEF based learning approaches are known to depend on an accurate kernel-width initialization. Also, the optimal value of this kernel-width is well-known to vary temporally and across deployment scenarios. To remove the dependency on kernel-width, a hyperparameter-free MEEF based adaptive algorithm is derived using random-Fourier features with sampled kernel widths (RFF-SKW). In addition, a detailed convergence analysis is presented for the proposed hyperparameter-free MEEF, which promises a near-optimal error-floor independent of step-size and guarantees convergence for a wide range of step sizes. The promised hyperparameter-independence and improved convergence for the proposed hyperparameter-free MEEF are validated by computer simulations considering different case studies.

In the context of outdoor localization over systems impaired by non-line of sight (NLoS), the minimum error entropy with fiducial points (MEE-FP) based methods have emerged as promising due to their excellent generalization and independence to statistics of NLoS. However, the performance of these approaches are well-known to depend on hyperparameters, such as, the spread parameter of the MEE-FP criterion. To enable MEEFP based hyperparameter-free localization, we propose a modified Gauss-Newton based localization algorithm based on sampled kernel-widths. Next, analytical results are derived to demonstrate the asymptotic equivalence of the proposed kernel-width sampling based localization algorithm to its ideal fixed kernel width based counterpart. This equivalence is validated through computer simulations assuming typical non-Gaussian NLoS distributions, which motivates the hyperparameter-independence of the proposed localization algorithm and its generalization to different NLoS statistics.

Nonlinear device characteristics present a severe performance-bottleneck for several upcoming next-generation wireless communication systems and prevent them from delivering high data-rates to the end-users. In this context, reproducing kernel Hilbert space (RKHS) based signal processing methods have gained widespread deployment and have been found to outperform classical polynomial-filtering-based solutions significantly. Furthermore, recent RKHS based techniques that rely on explicit feature-maps called random Fourier features (RFF) have emerged. These techniques alleviate the dependence on learning a dictionary and avoid the computations and errors incurred in dictionary-based learning. However, the performance of existing RKHS based solutions depends on choosing a suitable kernel-width. For the widely-used Gaussian kernel, we propose a methodology of assigning kernel-bandwidths that capitalizes on a stochastic sampling of kernel-widths using an ensemble drawn from a pre-designed probability density function. The technique is found to deliver a comparable convergence/error-rate performance to the scenario when the kernel-width is chosen by brute-force trial and error for tuning it for best performance. The desirable properties of the proposed kernel-sampling technique are supported by analytical proofs and are further highlighted by computer-simulations presented in the form of case studies in the context of next-generation communication systems.

Visible light communication (VLC) is eco-friendly and low-cost supplement to the existing radio frequency (RF) based communication system. However, the performance of VLC based system is limited by dispersive characteristics of the VLC channel which leads to inter-symbol-interference (ISI) and inter-carrier-interference (ICI). To alleviate the ISI and ICI, orthogonal time frequency space (OTFS) modulation has been proposed in the literature for both RF and millimeter wave communication systems. However, the performance analysis of OTFS over mobile multipath VLC channels is still an open area to investigate. This paper investigates the performance of OTFS over mobile multipath channels for the VLC system. Simulations performed over the realistic mobile multipath VLC channel indicate that OTFS with message passing detector exhibit better bit error rate over the conventional orthogonal frequency division multiplexing scheme with minimum mean square error based detector.

Color-shift keying (CSK) has emerged as an efficient modulation scheme for visible light communication (VLC) to enhance the data rate, which transmits data imperceptibly by variation of the color emitted by red, green and blue (RGB) light emitting diodes (LEDs). However, performance of the CSK-VLC systems is limited by the crosstalk between the adjacent color channels, which arises due to mismatch between the spectra of the RGB-LEDs and the filters at the receiver. Additionally, the RGB-LEDs' nonlinearity also degrades the overall symbol error rate (SER). In this paper, we propose a novel multivariate kernel least mean square based adaptive nonlinear equalizer for joint mitigation of crosstalk and RGB-LEDs' nonlinearity for CSK modulation scheme based VLC systems. Simulation results indicate that the proposed algorithm exhibits better SER performance as compared to the existing techniques. Analytical expression for SER is also derived for the proposed algorithm and validated by simulations.

Visible light communications (VLC) has lately emerged as a viable supplement to existing radio frequency (RF) based communication systems, and is an integral component of future communication systems. The possibility of deploying massive arrays of light emitting diodes (LED) and photo-detectors (PD), which translates to a massive array of multiple input multiple output (MIMO) system for VLC, have been recently proposed. However, massive-MIMO in VLC has the following drawbacks which limit its practical deployment: a) high condition number of the channel matrix, which negatively affects the robustness of traditional channel-inversion based detection, and b) high interference in a multi-user setting due to correlated columns of the overall MIMO channel matrix. In this paper, the multi-user (MU) massive-MIMO in VLC is considered, and a precoding technique is proposed that induces diverse effective channel matrices for different users, and hence facilitates MU detection using ordered successive interference cancellation. Further, it is shown that the proposed precoding technique maximizes an upper bound of the overall sum-rate of the MU massive-MIMO system. Lastly, simulations have been performed for MU massive LED and PD arrays demonstrating that the proposed precoding technique outperforms existing precoding techniques, and hence is a viable solution for massive-MIMO in VLC.

Visible light communication (VLC) is one of the main technologies driving the future 5G communication systems due to its ability to support high data rates with low power consumption, thereby facilitating high speed green communications. To further increase the capacity of VLC systems, a technique called non-orthogonal multiple access (NOMA) has been suggested to cater to increasing demand for bandwidth, whereby users' signals are superimposed prior to transmission and detected at each user equipment using successive interference cancellation. Some recent results on NOMA exist which greatly enhance the achievable capacity as compared with the orthogonal multiple access techniques. However, one of the performance-limiting factors affecting VLC systems is the nonlinear characteristics of a light emitting diode (LED). This paper considers the nonlinear LED characteristics in the design of pre-distorter for cognitive radio inspired NOMA in VLC, and proposes singular value decomposition-based Chebyshev precoding to improve performance of nonlinear multiple-input multiple-output NOMA-VLC. A novel and generalized power allocation strategy is also derived in this paper, which is valid even in scenarios when users experience similar channels. Additionally, in this paper, analytical upper bounds for the bit error rate of the proposed detector are derived for square M-quadrature amplitude modulation.

A new blind equalisation algorithm is presented for multi-level quadrature amplitude modulation schemes. The presented technique provides faster convergence and improved performance in terms of mean square error, which is demonstrated through various simulations. A mathematical analysis is also presented for the steady-state mean square error, which verifies the better performance of the proposed scheme over the conventional ones. The performance is also compared on the basis of inter-symbol interference metric. A computationally efficient implementation of the proposed algorithm is also presented. Simulations are done with various types of fading channels ranging from highly degrading Hoyt fading channels to moderate Rayleigh fading channels.