Neural networks (NNs) are pivotal in enhancing data processing tasks such as classification, generation, and restoration. A crucial consideration in these applications is the signal-to-noise ratio (SNR), which serves as a measure of the quality of the data. In this paper, we hypothesize that optimizing NNs in some tasks can be more effective when all samples in the dataset are clustered based on quantized SNR levels regarding the statistical similarity between training/test dataset. Hence, we introduce two novel techniques, i.e., 1) a linear algebraic method with a single-shot data sample and 2) an NN-based method with few-shot data samples, for estimating the SNR of a sparse signal. The proposed techniques are based on the mathematical fact that the dominant singular values contain the information of a signal space when signals are Hankelized in matrix form. Both algorithms achieve over 93% clustering accuracy, and almost 100% accuracy in high SNR scenarios and increased signal length. Furthermore, we provide an example of signal denoising as practical validation of the benefits of these clustering results for optimizing NN in a task. The proposed approaches show a superior denoising performance while requiring an extremely small training dataset compared to conventional methods, which can be interpreted as an improvement in the learnability of the NN.

Many estimation and reconstruction algorithms in signal processing fields can be improved themselves if the number of signals is known. However, this assumption of pre-knowledge is challenging in real environments. Additionally, it is often necessary to obtain the information of physical parameters of the signal through a short data acquisition time, i.e., a small number of samples, in systems requiring the low latency. Accordingly, an algorithm to effectively detect the number of signals through a small number of samples can be of great help to various estimation and reconstruction algorithms as the pre-processor for them. In this paper, we introduce a new algorithm which detects the number of signals with the efficiently designed neural network (NN), referred to as NsigNet. The proposed method is based on optimizing the NN by inputting the singular values of the re-shaped informative matrix from the sampled signal and outputting the one-hot encoding vectors indicating the number of signals. Simulation results show that NsigNet outperforms the conventional schemes in the various environments. Notably, the proposed scheme requires extremely small number of training dataset and network size. Finally, we provide two applications, i.e., (i) sparse signal recovery with compressive sensing and (ii) signal denoising with the iterative K-truncated singular value decomposition (SVD), to validate the benefit of NsigNet in the practical on-/off-grid problems, respectively.

In this article, we introduce a novel compressed sensing (CS) scheme for sparse signal recovery in an effective method, namely compressed geometric sequential sensing (CGSS). This comes from the fact that an observation vector in CS can be interpreted as a superposition of multiple geometric sequences if the sensing matrix is a partial discrete Fourier transform (DFT) matrix. The main idea is based on the mathematical property that the nonorthogonally superposed geometric sequences can be decomposed, without loss of information, into the original geometric sequences in specific patterned ways. With this method, a K-sparse vector can be perfectly reconstructed through only$2K$observations in the ideal case (i.e., noise-free observations) regardless of the length of the original K-sparse vector. To verify the robustness of our proposed scheme, it is compared with existing CS techniques under two environments with noisy observations, which are the additive white Gaussian noise (AWGN) and the impulsive noise. In the simulation part, we show that the performance of CGSS can be improved through an appropriate denoising technique in AWGN cases. Notably, in impulsive noisy cases, the proposed scheme enables the perfect reconstruction of the sparse signal within the given condition.

Feasibility of using unlicensed spectrum for ultra reliable low latency communications (URLLC) is still a question for beyond 5G wireless networks. Low latency access to the channel and efficiently sharing spectrum among the multiple users are the main requirements for exploiting unlicensed spectrum for URLLC. Listen before talk and back-off procedures implemented to avoid the collisions in channel access hinder the low latency communication. In this paper, we propose a novel low-latency medium access control (MAC) scheme based on the collision resolution for a pairwise random wireless network. We use geometric sequence decomposition for collision resolution among the competing users. This enables the system to tackle collisions and thus removing the need for carrier sensing and back-off procedures. This saves time in obtaining access to the channel and improves the efficiency of the system. We implement our approach in the synchronized time slotted system and show that it yields significant improvement over existing MAC schemes.

This paper presents a computationally efficient technique for decomposing non-orthogonally superposed k geometric sequences. The method, which is named as geometric sequence decomposition with k-simplexes transform (GSD-ST), is based on the concept of transforming an observed sequence to multiple k-simplexes in a virtual k-dimensional space and correlating the volumes of the transformed simplexes. Hence, GSD-ST turns the problem of decomposing k geometric sequences into one of solving a k-th order polynomial equation. Our technique has significance for wireless communications because sampled points of a radio wave comprise a geometric sequence. This implies that GSD-ST is capable of demodulating randomly combined radio waves, thereby eliminating the effect of interference. To exemplify the potential of GSD-ST, we propose a new radio access scheme, namely non-orthogonal interference-free radio access (No-INFRA). Herein, GSD-ST enables the collision-free reception of uncoordinated access requests. Numerical results show that No-INFRA effectively resolves the colliding access requests when the interference is dominant.

Inter-cell interference coordination (ICIC) is a promising technique to improve the performance of frequency-domain packet scheduling (FDPS) in downlink LTE/LTE-A networks. However, it is difficult to maximize the performance of FDPS using static ICIC schemes because of insufficient consideration of signal-to-interference-and-noise ratio distribution and user fairness. On the other hand, dynamic ICIC schemes based on channel state information (CSI) also have difficulty presented in the excessive signaling overhead and X2 interface latency. In order to overcome these drawbacks, we introduce a new concept of ICIC problem based on geometric network information (GNI) and propose an adaptive sector coloring game (ASCG) as a decentralized solution of the GNI-based ICIC problem. Furthermore, we develop an ASCG with a dominant strategy space noted as ASCG-D to secure a stable solution through proving the existence of Nash equilibrium. The proposed scheme provides better performance in terms of system throughput gain of up to about 44.1%, and especially of up to about 221% for the worst 10% users than static ICIC schemes. Moreover, the performance of the CSI-based ICIC, which require too much computational load and signaling overhead, is only 13.0% and 5.6% higher than that of ASCG-D regarding the total user throughput and the worst 10% user throughput, respectively. The most interesting outcome is that the signaling overhead of ASCG-D is 1/144 of dynamic ICIC schemes' one.

In heterogeneous cellular networks, users are sometimes forcibly redirected into a low power base station (BS) for the purpose of data offloading. In order to guarantee an acceptable performance for such users, who suffer from severe inter-tier interference, enhanced inter-cell interference coordination (eICIC) was proposed. Using this technique, macro BSs periodically mute their data transmission by using an almost blank subframe (ABS). In this paper, we formulate a joint optimization problem incorporating user association and ABS portion tuning to increase network-wide utility. In the development of our algorithms, we particularly consider the time-varying characteristics of wireless propagation channels in order to reflect practical signal transmission environments, and we derive a throughput estimation equation that is compatible with the eICIC operation. Based on this analysis, we separately develop algorithms for user association and ABS tuning, and the performance enhancement achieved by our proposed methods is verified through extensive system-level simulations.

Small cells are considered an emerging technology for increasing the potential capacity of cellular networks. However, as the density of infrastructure increases, users have many choices for connection, and therefore, selecting an appropriate base station (BS) becomes an important issue. This study aims to provide an improved user association rule, where each user autonomously chooses the best among all the BSs in the vicinity, while considering their congestion levels. As the first step, an optimization problem is formulated, which emphasizes both the time-varying nature of the wireless channel and fairness among users. On the basis of this formulation, the influence of a specific handover event on system performance is investigated, and then, two versions of the handover frameworks are developed. Simulation results show that the proposed algorithms increase the throughput of every user in the network by 2.8–12% compared to the best conventional scheme. Furthermore, these schemes especially enhance the performance of users having low service quality by 5.0–85%, through efficient utilization of the pre-installed infrastructures.