In the last decade, there has been a surge of research interest in feature extraction using random sampling. These techniques are fast and scalable and, at the same time, have practical favorability in low-sample size and high-dimensional training data. Convolutional Kitchen Sinks-based methods are promising random feature extractors for time series data. Since these methods are data-independent, many of the extracted features are redundant. To address this problem, we propose a simple and efficient feature selection method based on knee/elbow detection in the curve of ordered coefficients in linear regression. Our empirical studies show that without significant loss in accuracy, the proposed feature selector, on average, prunes more than 84 percent of randomly generated features.

Localization in wireless networks is possible by measuring some characteristics of the propagating signal related to the position of the user, which is always corrupted by noise components. In this paper, a correntropy based algorithm is proposed for localization and tracking of a mobile station in wireless networks. The performance of the proposed algorithm is compared with the Least Mean Square (LMS) and Least Mean P-norm (LMP) algorithms, and its preference aspects are discussed. The results show that, using correntropy can bring robustness to localization and improve performance in many realistic scenarios such as fat-tail noise distributions, one of the serious bottlenecks of the next-generation 5G wireless communications systems. In addition, it is shown that, the Gaussian kernel of the correntropy function reduces the sensitivity of the algorithm to the learning rate.

In this letter, using a general functional model for the stochastic systems we show that, the gradient of the mutual information with respect to any arbitrary system parameters can be computed in general. Relying on this viewpoint, the validity condition of one of the main available results on the gradient of the mutual information is investigated. Also, the result on the general additive noise system model in the literature is recovered as a special case.

The authors propose a new technique to model the arriving energy's knowledge (AEK) in the energy harvesting (EH) communication systems. They consider a batteryless EH transmitter with the AEK non‐causally provided to it. They show that the capacity of the authors' model in general case, that is, discrete memoryless channel, can be derived through a binning scheme. This resembles a Gelfand–Pinsker type capacity formula which intuitively is the result of treating the energy as a channel state. Moreover, they investigate two especial cases of binary and Gaussian channels with the EH transmitters. In the binary symmetric channel with EH transmitter (BSCEH), their coding scheme leads to achieve the capacity of the classic binary symmetric channel (BSC). In the batteryless Gaussian channel with EH transmitter, the upper bound on the capacity is derived, and it is shown this bound is less than the capacity of the EH Gaussian channel with infinite battery.

This paper introduces a new applicable Gaussian wiretap channel with side information. In this channel, the state of the channel to the wiretapper is separated from the main channel's one. It means that the legitimate receiver and the wiretapper are listening to the transmitted signal through the different channels with different channel states which may have some correlation to each other. The state of the main channel, is assumed to be known at the transmitter, helps the sender to encrypt its messages. The achievable equivocation-rate region for this model is drived in the Gaussian case and the results are reduced to the previous works as special cases.

In this study the negative impacts of interference transmitters on automatic modulation classification (AMC) have been discussed. The authors proposed two approaches for AMC in the presence of interference: single user modulation classification (SUMC) and multiuser modulation classification (MUMC). When the received power of one transmitter is larger than the other transmitters, SUMC approach recognises the modulation type of the dominant transmitter and other transmitters are treated as interferences. Alternatively when the received powers of all the transmitters are close to each other the authors propose MUMC method to recognise the modulation type of all of the transmitted signals. The features being used to recognise the modulation types of transmitted signals for the both approaches, SUMC and MUMC are higher order cumulants. The superposition property of cumulants for independent random variables is utilised for SUMC and MUMC. The authors investigated the robustness of their classifier with respect to different powers of the received signals via analytical and simulation results and the authors have shown the analytical results will be confirmed by simulations. Also the authors studied the effects of signal synchronisation error, phase jitter and frequency offset on their proposed methods via simulations in the both condition for MUMC and SUMC.

In this study the authors provide an adaptive cross-layer design that maximises the spectral efficiency under prescribed quality of services (QoSs) such as error rate and delay. The considered scheme combines adaptive modulation at physical layer with a truncated automatic repeat request at data link layer in the presence of both channel estimation and prediction errors. Both the channel estimation and prediction are done based on adaptive pilot-symbol-assisted modulation in which both pilot spacing and power allocation on pilot symbols are optimised in order to maximise the spectral efficiency. We provide closed-form expression for average spectral efficiency and average packet error rate over Rayleigh fading channel. Numerical results demonstrate that our adaptive cross-layer design improves the spectral efficiency while meeting the prescribed QoS. We have shown that the channel prediction is more important than the channel estimation in order to improve the spectral efficiency in a practical situation. Also we have found that the increment of spectral efficiency decreases on increasing the maximum number of retransmissions, and thus in order to reduce the delay and buffer size at the data link layer, the maximum number of retransmissions are limited to a small number in practice.

In this article, we study the power control (PC) process in wireless cellular code division-multiple access systems under flat fading and propose a novel approach to find an optimum step size for closed-loop power control algorithms. In this approach, an optimum step size will be computed from a proposed function. This function depends on system parameters such as, the number of co-channel users, processing gain, the period of PC, Doppler frequency, channel attenuation and the order of diversity. Based on this computation, the mobile station (MS) adjusts its transmit power optimally to decrease interference for other co-channel users. Simulation results for different sets of system parameters show that the proposed algorithm decreases the bit error rate, the outage probability at the base station (BS), and increases the battery life of the MS compared with other values of the step size. The performance of the proposed algorithm is compared with the fixed-step-size power control algorithm and superiority of its performance is confirmed by simulation results. Moreover, the upper and lower bounds of the outage probability and the received signal-to-interference ratio for the proposed algorithm at the BS will be calculated.

In this paper we consider a cognitive interference channel with two confidential messages. In our scenario, although the cognitive transmitter cooperates with the primary sender, the primary and cognitive messages must be secure at unintended receivers. The level of secrecy is measured by the equivocation rate. Also an expression is obtained for the rate-equivocation region of the discrete memoryless cognitive interference channel with confidential primary and secondary messages.

We propose a technique for enhancing the achievable spectral efficiency of multiuser direct-sequence code-division multiple-access (DS-CDMA) fading channels in the presence of additive white Gaussian noise (AWGN) and multiple access interference (MAI). The proposed scheme involves the joint optimization of outer loop power control (OLPC) and rate control using variable spreading factors (VSFs). The optimality is in the sense of average spectral efficiency maximization. The optimum outer loop target signal-to-noise ratio (SNR-target) and the corresponding spreading factor are derived jointly, in terms of the number of active users. Along with transmit power adaptation in the inner loop, this leads to maximized average spectral efficiency. Total and truncated channel inversion strategies are used in the inner loop. The average spectral efficiency of the joint optimization scheme is derived for the conventional matched-filter and the multiuser decorrelating detectors. Average transmit power and instantaneous bit error rate (BER) constraints are considered and the performance is evaluated over Nakagami-m flat-fading channels. In low SNRs, the proposed scheme can provide a considerable gain in bits/s/Hz for either of the detectors, compared to a VSF-assisted system that does not exploit OLPC and thus the optimum SNR-target.

An iterative reconstruction method that was proposed for clipping noise cancellation in orthogonal frequency domain multiplexing (OFDM) systems is applied to multilevel quadrature amplitude modulation-based multi-carrier (MC) code division multiple access signals in downlink. The iterative method uses a known distortion function, maybe a nonlinear one such as a successive clipping and filtering process, in the iterations to give an approximation of inverse of the distortion process and then the iterative method removes distortion under a convergence condition. The authors show that MC signal with properly chosen clipping threshold satisfies convergence conditions of the iterative method. In contrast to some of the other reconstruction-based techniques, this method requires no extra bandwidth and side information and it can be implemented with reasonable complexity. Furthermore, the authors show that the proposed iterative scheme can be enhanced by using an extra frequency bandwidth. Exploiting extra bandwidth improves the performance of the reconstruction-based methods in case of using successive clipping and filtering. Simulation results will be used to demonstrate achievable bit-error-rate improvement by the proposed enhanced iterative scheme.

We propose a technique for enhancing the achievable spectral efficiency of multiuser direct-sequence code-division multiple-access (DS-CDMA) channels in the presence of additive white Gaussian noise (AWGN) and multiple access interference (MAI). The proposed scheme is the joint optimization of outer loop power control and variable spreading factors (VSFs). The optimum SNR-target, the SNR that results in optimal spectral efficiency, is set in the outer loop. Accordingly, appropriate spreading factor is selected, and along with the transmit power adaptation in the inner loop, it leads to maximized spectral efficiency. Total and truncated channel inversion strategies are used in the inner loop. Analytical expressions are derived for the spectral efficiency of the joint optimization scheme for the conventional matched-filter and the multiuser decorrelating detectors. Average transmit power and instantaneous BER constraints are considered and the performance is evaluated over Nakagami-m flat fading channels. We show that our scheme achieves higher spectral efficiency than a system that uses VSF transmission but without outer loop optimal SNR-target. The gain is realized at no transmit power cost and without sacrificing the BER.

In this letter, we propose a new robust MUD, called /spl alpha/ detector, for non-Gaussian noise. We consider the Gaussian-mixture model for non-Gaussian or impulsive noise. Our technique outperforms the decorrelator and the minimax detectors in highly impulsive noise. The proposed method uses a parametric cost function, where the parameter /spl alpha/ is selected using the difference between the asymptotic variance of estimation error of the /spl alpha/ detector and that of the minimax detector.

We present a novel nonparametric multiuser detector for non-Gaussian channels that is based on the signed-rank norm for linear regression. Analytical and simulation results show that the proposed detector offers similar or better performance as compared to the minimax robust detector, but without requiring any a priori information on the noise. The complexity of this detector is lower than that of the pseudo-norm nonparametric detector stated previously by the authors. This is due to the fact that in contrast to the latter, it is not necessary to compute the intercept parameter for the signed-rank detector proposed in this correspondence. We analyze the behavior of the blind version of this detector and show that it outperforms the blind minimax detector. We also show that this detector has a bounded influence function and hence it is robust.