Performance Evaluation of Machine Learning-Based Channel Equalization Techniques: New Trends and Challenges

This paper presents a new approach to equalization of communication channels using Functional Link Artificial Neural Networks (FLANNs). A novel method of training the FLANNs using PSO Algorithm is described. The performance of the proposed network has been compared with the conventional LMS based channel equalizer and FLANN trained with BP algorithm based equalizer. From the results it can be noted that the proposed algorithm improves the classification capability of the FLANNs in differentiating the received data.

This paper presents a specific structure of neural network as the functional link artificial neural network (FLANN). This technique has been employed for classification tasks of data mining. In fact, there are a few studies that used this tool for solving classification problems. In this present research, we propose a hybrid FLANN (HFLANN) model, where the optimization process is performed using 3 known population based techniques such as genetic algorithms, particle swarm and differential evolution. This model will be empirically compared to FLANN based back-propagation algorithm and to others classifiers as decision tree, multilayer perceptron based back-propagation algorithm, radical basic function, support vector machine, and K-nearest Neighbor. Our results proved that the proposed model outperforms the other single model. (Abstract)

This study shows how the channel estimation based Deep Learning (DL) and a power allocation method are together employed for multi-user detection in a Power domain Non-Orthogonal Multiple Access (PD-NOMA) network. Successive interference cancellation (SIC) procedure is typically employed at receiver side, where numerous users are decoded in a successive approach. Fading channels may scatter transferred signal and initiate dependencies between scattered components, this might influence the channel estimation technique and therefore impact the SIC procedure and signal recognition precision. In our proposed scheme, the influence of Deep Neural Network (DNN) in clearly approximating the channel parameters for users in NOMA cell is inspected. In our scenario, we incorporate the Long Short Term Memory (LSTM) layer with NOMA cell where the LSTM is employed for complex data management to perform training and predication. The DNN is trained online on basis of random channel models and then the trained network is used to approximate the channel taps that will be utilized by the receiver in recovering the desired symbols. Additionally, power factors for user’s devices are optimized to maximize the sum-rate of users where whole power and Quality of service (QoS) restrictions are considered. Simulation outcomes in terms of Bit Error Rate (BER), Outage probability, and sum rate have shown the dominance of the suggested channel estimation using DL over standard estimation approach. Moreover, both fixed power and optimized power schemes are also assessed when DNN is applied.

Genes that contribute to complex traits pose special challenges that make candidate disease-associated gene discovery more difficult. In this work, we investigated topological features derived from PPI network to identify the causing genes of four complex diseases: Cancer, Type 1 Diabetes, Type 2 Diabetes, and Ageing genes. We used 10-fold cross-validation to evaluate the predictive capacity of all possible combinations of these features and found the features with the best predictive ability. We assessed the performance of Multi-layer Perceptron (MLP), Functional Link Artificial Neural Network (FLANN), and Support Vector Machines (SVM). We found that SVM provides higher accuracy than MLP and FLANN. However, the FLANN has significantly low computation time while its accuracy is comparable to that of SVM and MLP.

The Multi user Multi Carrier Direct sequencing Code Division Multiple Access (MU-MC-DS-CDMA) strategy was developed in response to rising demand for system capacity, higher data rates, and multimedia services. It's essentially a combination of CDMA and OFDM schemes. As a result, MC-CDMA takes the advantages of both CDMA and OFDM technologies. CDMA improves bandwidth consumption performance, whereas OFDM helps to minimize Inter Symbol Interference (ISI) and multipath fading. This dissertation suggests a method for evaluating the performance of multi user MC-DS-CDMA system. BER and PAPR are used to evaluate the results. The Bit Error Rate (BER) output of the proposed method is then examined using MATLAB for various modulation techniques. By using Walsh-Hadamard (W-H) code is used because it eliminates Multiple Access Interference (MAI). In terms of BER, it can be shown that BPSK performs better than QPSK. Key Words: AWGN, BER, ISI, MC-CDMA, Rayleigh Fading Channel, SNR, QPSK, MAI

Development and performance evaluation of neural network classifiers for Indian internet shoppers

The 4G up-gradation was essential because of the exponential increment of the wireless applications. So, 5G was introduced with large capacity, small latency, improved reliability, high data rate, and better quality of services (QoS). 4G uses orthogonal frequency-division multiplexing (OFDM) which has a decent quality of long-distance communication, by eliminating intersymbol interference (ISI) and improving the signal-to-noise ratio (SNR), but it has a spectrum wastage problem. However, the improved data rate is possible by spectrum utilization. 5G adopted various techniques for the better utilization of spectrum in current scenarios. The new generation waveforms, i.e., filtered orthogonal frequency-division multiplexing (F-OFDM), generalized frequency-division multiplexing (GFDM), universal filtered multicarrier (UFMC), and filter bank multicarrier (FBMC), are one of them. This paper will discuss FBMC which is the alternative technology of OFDM with better features by building subcarrier waveform based on prototype filters. Here, FBMC is investigated in terms of BER (bit error rate) by varying several parameters such as the size of inverse fast Fourier transforms/fast Fourier transform (IFFT/FFT), length of guard band, and length of symbols. Results of simulation will show that the bit error rate (BER) of FBMC is not affected by the variation of symbol length, while advancement in the size of IFFT/FFT increases BER, on the given parameter while the BER reduces for large size of the guard band of each subcarrier.KeywordsFilter Bank MulticarrierIFFT/FFTOffset QAMOrthogonal Frequency-Division MultiplexingBit Error Rate

In this paper proposes a receiver for direct sequence-code division multiple access (DS-CDMA) multi-user communication system based on Laguerre filter and the recursive least squares (RLS) adaptive algorithm. Intersymbol interference (ISI) is usually caused by multipath propagation and the inherent non-linear frequency response of a channel. Therefore, the objective of this paper is to minimize the effects of ISI with a long impulse response environment based on the proposed receiver. Because of the stability of Laguerre structure is compromise between the finite impulse response (FIR) filter and infinite impulse response (IIR) filter. It can improve the bit error rate (BER) performance of the proposed receiver compared to the conventional FIR filter receiver. The simulation results show that the performances of the proposed receiver in terms of BER and the stability are better than does the conventional FIR filter receiver.

Performance of an orthogonal frequency division multiplexing (OFDM) transmission system is evaluated analytically taking into considerations the combined influence of receiver noise, Rayleigh fading and the timing error due to imperfect slot synchronization at the receiver. Performance results are presented in terms of bit error rate (BER) at a bit rate of 100 kbps in the presence of inter-channel interference (ICI) and intersymbol interference (ISI) and additive white Gaussian channel noise. The power penalty suffered by the system due to above system imperfections is also determined at a BER of 10-6. The results show that there is a significant degradation in system BER due to timing error. The computed results show that the penalty suffered by the system only due to timing synchronization error is about 2.0 dB corresponding to timing bit errors to the right of m = 10 and is 5.25 dB for m = 14.

The Air Quality Index (AQI) measures how air pollution affects human health. As pollution levels rise in Indian cities, we need reliable prediction models for better environmental management. This paper analyses different machine learning techniques for predicting AQI. We use Support Vector Regression (SVR), Random Forest Regression (RFR), and CatBoost Regression (CR) on data from New Delhi, Bangalore, Kolkata, and Hyderabad. We evaluate model performance using Root Mean Square Error (RMSE) and accuracy. Experimental results show that RFR performs best in most cities, while CR is most effective in New Delhi. To tackle dataset imbalance, we use the Synthetic Minority Oversampling Technique (SMOTE), which improves prediction accuracy for all models. Additionally, we assess other models, including SARIMA, Support Vector Machine (SVM), and Long Short-Term Memory (LSTM) for Ahmedabad. Among these, SVM with a radial basis function (RBF) kernel shows the best results. The findings emphasize how combining data balancing methods with machine learning models can improve AQI prediction. This approach can help with pollution control strategies and better decision-making.

This paper considers time-domain equalization of fading channels for multicarrier systems with channels whose delay spread exceeds the length of the cyclic prefix. We propose to apply Regalia's Finite Interval Constant Modulus Algorithm, which directly minimizes the constant modulus criterion for finite data sets, to the output of a single sub-carrier. This results in a blind adaptive channel shortening technique which is used to design a time-domain finite impulse response equalizer or channel shortener, whose goal is to compress the overall channel impulse response to within the length of the cyclic prefix. Simulation results are provided in terms of bit error rate (BER) in the context of wireless systems.

Deep learning approaches applied in diverse fields further the expectations and demands of 5G wireless networks and necessitates leveraging the intelligence to network with widespread renewed interest of big data. The integration of machine learning (ML) notion across core and edge infrastructure of wireless networks imparts intelligence. ML can enable mobile devices to intelligently monitor their environment by learning the features of a wireless channel—its dynamics, traffic pattern, requests, and user context, among other aspects. ML enables the infrastructure to learn from the wireless networking environment and adaptively makes decision, which provides optimized solutions for best connectivity, resource allocation, quality-of-service parameters, etc. ML can redefine how the physical layer functions, namely, coding/decoding, channel estimation, channel equalization, and modulation/demodulation at the transmitter and receiver sides. The majority of the communication system is modeled as a linear system, but practically, ML captures the nonlinear effects of hardware and quantization and provides a better solution. A combined approach is possible through ML for the solutions provided in the literature for individual blocks, say, coding, resource allocation block. Also parallel processing ability and distributed memory aid to complexity reduction at the respective block. The proliferation of graphics processing unit–enabled hardware provides speedy action. These are the benefits of deep learning in physical layer design. This chapter intends to provide a deep learning model addressing the physical layer of the network. The authors propose an intelligent ML detector at the multiple-input multiple-output receiver able to classify a signal constellation and the transmitted antenna.

Support vector machines (SVMs) are a good candidate for the solution of antenna array processing problems such as beamforming, detection of the angle of arrival, or sidelobe suppression, due to the fact that these algorithms exhibit superior performance in generalization ability and reduction of computational burden. Here, we introduce three new approaches for antenna array beamforming based on SVMs. The first one relies on the use of a linear support vector regressor to construct a linear beamformer. This algorithm outperforms the minimum variance distortionless method (MVDM) when the sample set used for training is small. It is also an advantageous approach when there is non-Gaussian noise present in the data. The second algorithm uses a nonlinear multiregressor to find the parameters of a linear beamformer. A multiregressor is trained off line to find the optimal parameters using a set of array snapshots. During the beamforming operation, the regressor works in the test mode, thus finding a set of parameters by interpolating among the solutions provided in the training phase. The motivation of this second algorithm is that the number of floating point operations needed is smaller than the number of operations needed by the MVDM since there is no need for matrix inversions. Only a vector-matrix product is needed to find the solution. Also, knowledge of the direction of arrival of the desired signal is not required during the beamforming operation, which leads to simpler and more flexible beamforming realizations. The third one is an implementation of a nonlinear beamformer using a non-linear SVM regressor. The operation of such a regressor is a generalization of the linear SVM one, and it yields better performance in terms of bit error rate (BER) than its linear counterparts. Simulations and comparisons with conventional beamforming strategies are provided, demonstrating the advantages of the SVM approach over the least-squares-based approach.

Sequence Classification is one of the on-demand research projects in the field of Natural Language Processing (NLP). Classifying a set of images or text into an appropriate category or class is a complex task that a lot of Machine Learning (ML) models fail to accomplish accurately and end up under-fitting the given dataset. Some of the ML algorithms used in text classification are KNN, Naïve Bayes, Support Vector Machines, Convolutional Neural Networks (CNNs), Recursive CNNs, Recurrent Neural Networks (RNNs), Long Short Term Memory (LSTM), etc. For this experimental study, LSTM and a few other algorithms were chosen for a more comparative study. The dataset used is the SMS Spam Collection Dataset from Kaggle and 150 more entries were additionally added from different sources. Two possible class labels for the data points are spam and ham. Each entry consists of the class label, a few sentences of text followed by a few useless features that are eliminated. After converting the text to the required format, the models are run and then evaluated using various metrics. In experimental studies, the LSTM gives much better classification accuracy than the other machine learning models. F1-Scores in the high nineties were achieved using LSTM for classifying the text. The other models showed very low F1-Scores and Cosine Similarities indicating that they had underperformed on the dataset. Another interesting observation is that the LSTM had reduced the number of false positives and false negatives than any other model.

This paper addresses the problem of channel tracking and equalization for multi-input multi-output (MIMO) time-varying frequency-selective channels. These channels model the effects of inter-symbol interference (ISI), co-channel interference (CCI), and noise. A low-order autoregressive model approximates the MIMO channel variation and facilitates tracking via a Kalman filter. Hard decisions to aid Kalman tracking come from a MIMO finite-length minimum-mean-squared-error decision-feedback equalizer (MMSE-DFE), which performs the equalization task. Since the optimum DFE for a wide range of channels produces decisions with a delay /spl Delta/ > 0, the Kalman filter tracks the channel with a delay. A channel prediction module bridges the time gap between the channel estimates produced by the Kalman filter and those needed for the DFE adaptation. The proposed algorithm offers good tracking behavior for multiuser fading ISI channels at the expense of higher complexity than conventional adaptive algorithms. Applications include synchronous multiuser detection of independent transmitters, as well as coordinated transmission through many transmitter/receiver antennas, for increased data rate.