It is important to understand the properties of spatial correlation matrices in multiple-input multiple-output (MIMO) systems. In this paper, we derive new bounds for the maximum and minimum eigenvalues of spatial correlation matrices characterized by the exponential model. The new upper bounds for the maximum and minimum eigenvalues are tighter than the previously known bounds. Moreover, the new lower bound for the minimum eigenvalue, which has not yet been derived in the literature as a function of the number of antennas, is also tight. In order to predict the behavior of these bounds, we investigate the gap between the lower and upper bounds. These bounds are directly applicable to the analysis of wireless communication scenarios, such as uniform planar arrays and ill-conditioned channels, which are expected to be widely used for massive MIMO systems.

In this paper, we propose a deep neural network (DNN)-based automatic modulation classification (AMC) for digital communications. While conventional AMC techniques perform well for additive white Gaussian noise (AWGN) channels, classification accuracy degrades for fading channels where the amplitude and phase of channel gain change in time. The key contributions of this paper are in two phases. First, we analyze the effectiveness of a variety of statistical features for AMC task in fading channels. We reveal that the features that are shown to be effective for fading channels are different from those known to be good for AWGN channels. Second, we introduce a new enhanced AMC technique based on DNN method. We use the extensive and diverse set of statistical features found in our study for the DNN-based classifier. The fully connected feedforward network with four hidden layers are trained to classify the modulation class for several fading scenarios. Numerical evaluation shows that the proposed technique offers significant performance gain over the existing AMC methods in fading channels.

While the performance analysis of linear zero-forcing (ZF) and minimum mean-square error (MMSE) receivers has received much attention, there has been no exact closed-form distribution of signal-to-interference-plus-noise ratio (SINR) for MMSE multiple-input multiple-output (MIMO) systems in the arbitrary fading environments for over the past 20 years. In this paper, an exact and general distribution of SINR for MMSE MIMO systems is presented under uncorrelated Rayleigh fading environments. We start our analysis by finding regular patterns from the probability density function (PDF) of SINR for small dimensions, and using the regular patterns, we extend the PDF to arbitrary dimensions. From the exact and general PDF of SINR for MMSE MIMO systems, the cumulative distribution function and the moment-generating function are derived in terms of the incomplete gamma function and the confluent hypergeometric function of the second kind, respectively. As applications, the error probability, outage probability, and achievable sum rate are investigated by using the distribution of SINR for MMSE MIMO systems. Most notably, the achievable sum rates for ZF and MMSE MIMO systems are defined by using the Meijer G-function, which are the first closed-form expressions in the literature.

This paper proposes a blind interleaver parameters estimation method enhanced by identifying relatively error-less partial symbols among intercepted streams. By exploiting the distribution of the ranks of the random matrices, we can choose partial symbols having relatively small errors. Calculating the rank of the matrix constructed using these symbols can improve estimation of the blind interleaver parameters. Experimental results show that the proposed algorithm performs better than the previous ones.

It is well known that the distribution of the ranks of random matrices follows a specific distribution. By partitioning ranks into two possible outcomes, we can transform the estimating interleaver parameters into the counting of success events. Using this probabilistic framework, we can efficiently determine interleaver parameters in a noncooperative context.

Automatic modulation classification (AMC) plays an important role in cooperative and non-cooperative contexts. Many studies on the application of deep learning (DL) to AMC have widely been reported. This paper deals with an AMC for OFDM signals based on convolutional neural network (CNN) among DL methods. For AMC, we propose a loss function which we refer to as α-softmax loss function and present a deep CNN model utilizing the proposed loss function. By optimizing the proposed loss function, we can further separate the features of one modulation scheme from those of the other modulation schemes for the classification performance improvement. Through computer simulations, we show that the proposed model with α-softmax loss function outperforms the conventional ones in terms of classification accuracy.