Non-orthogonal multiple access (NOMA) has gained considerable interest from the 3GPP community as a potential radio access strategy for the future fifth-generation (5G) wireless networks. Compared to orthogonal multiple access (OMA), NOMA is more efficient from the perspective of throughput performance making it more favorable for 5G systems. Existing NOMA techniques merely offer a rigid user grouping without exploring channel heterogeneity and diversity to cluster users, resulting in a poor throughput performance. An adaptive user clustering (AUC) approach has been proposed to search through all possible combinations to obtain the best clusters with the highest throughput. This scheme exploits the channel diversity of users to maximize throughput, however, the brute-force search (B-FS) method to find the optimal combinations results in a prohibitive complexity. In this paper, a novel artificial neural network (ANN) approach is proposed for user clustering in the downlink of the 5G NOMA system in order to maximize throughput performance at an acceptable complexity. In the proposed strategy, ANN model is first trained with the historical dataset, which contains the transmitting powers and channel gains of the downlink NOMA users, along with the information of the corresponding clusters which maximize the throughput performance of the system. Next, validation is performed to tune the values of hyper-parameters such as learning rate, length of training data, and epoch learned during training to validate cluster formation and to avoid over-fitting of the model. Finally, the ANN model is tested with the learned parameters and tuned hyper-parameters, to predict the formation of clusters and to evaluate the accuracy of the model. Simulation results demonstrate that the proposed scheme is able to obtain a significant reduction in terms of complexity with a performance of 98% for throughput (near-optimal throughput performance) when compared with the optimal approaches.

Fingerprint-based indoor positioning systems (F-IPS) may provide inexpensive solutions to GPS-denied environments. Most F-IPSs adopt traditional machine learning for position prediction, resulting in low accuracy. Deep neural networks (DNN) were recently employed for F-IPSs to minimize positioning errors. Nevertheless, a DNN-IPS fails to guarantee high accuracy in dynamic environments as it is sensitive to changes in the input data. A convolutional neural network (CNN) is recommended to replace DNN due to its capability to learn the overall topology of fingerprinting images and capture highly abstract features. Due to the convolution process and image representation, CNN-IPS incurs prohibitive storage and computational requirement for implementation on resource-limited devices. This paper incorporates knowledge distillation (KD) into CNN-IPS to distil knowledge from large deep CNNs into small CNNs. The pre-trained teacher network uses the soft probability output where the score vector from the trained network is converted into a probability distribution, which is softened by the temperature hyperparameter, leading to a more simplified model. Based on the numerical results, KD-CNN-IPS manifests better localization performance where 79.84% of the positioning errors are within 2 meters while its testing time is only 79.68% of that of the teacher model. Compared to the CNN-IPS, KD-CNN-IPS with precisely the same architecture and size could achieve a performance improvement of 13.65% in terms of the average positioning error.

Indoor positioning has garnered significant interest over the last decade due to the rapidly growing demand for location-based services. As a result, a multitude of techniques has been proposed to localize objects and devices in indoor environments. Wireless fingerprinting, which leverages machine learning, has emerged as one of the most popular positioning approaches due to its low implementation cost. The prevailing fingerprinting-based positioning mainly utilizes wireless fidelity (Wi-Fi) and Bluetooth low energy (BLE) signals. However, the RSS of Wi-Fi and BLE signals are very sensitive to the layout of the indoor environment. Thus, any change in the indoor layout could potentially lead to severe degradation in terms of localization performance. To foster the development of new positioning methods, several open-source location fingerprinting datasets have been made available to the research community. Unfortunately, none of these public datasets provides the received signal strength (RSS) measurements for indoor environments with different layouts. To fill this gap, this paper presents a new hybrid Wi-Fi and BLE fingerprinting dataset for multi-floor indoor environments with different layouts to facilitate the future development of new fingerprinting-based positioning systems that can provide adaptive positioning performance in dynamic indoor environments. Additionally, the effects of indoor layout change on the location fingerprint and localization performance are also investigated.

Non-orthogonal multiple access (NOMA) emerges as a promising candidate for 5G, which radically alters the way users share the spectrum. In the NOMA system, user clustering (UC) becomes another research issue as grouping the users on different subcarriers with different power levels has a significant impact on spectral utilization. In previous literature, plenty of works have been devoted to solving the UC problem. Recently, the artificial neural network (ANN) has gained considerable attention due to the availability of UC datasets, obtained from the Brute-Force search (BF-S) method. In this paper, deep neural network-based UC (DNN-UC) is employed to effectively characterize the nonlinearity between the cluster formation with channel diversity and transmission powers. Compared to the ANN-UC, the DNN-UC is more competent as UC is a non-convex NP-complete problem, which cannot be entirely captured by the ANN model. In this work, the DNN-UC is first trained with the training samples and then validated with the testing samples to examine its mean square error (MSE) and throughput performance in an asymmetrical fading NOMA channel. Unlike the ANN-UC, the DNN-UC model offers greater room for hyper-parameter optimizations to maximize its learning capability. With the optimized hyper-parameters, the DNN-UC can achieve near-optimal throughput performance, approximately 97% of the throughput of the BF-S method.

The received signal strength (RSS) based Wi-Fi fingerprinting method is one of the most potential and easily deployed approaches for a reliable indoor positioning system. However, due to the labor intensive and time-consuming radio map construction process, interpolation is often incorporated. To ensure the interpolated radio map is robust against environmental noise and RSS fluctuations, we propose two novel interpolation methods, termed as DimRed and DimRedClust, for an improved radio map construction. The former performs dimensionality reduction prior to the interpolation while the latter employs both the dimensionality reduction and clustering before interpolating the radio map. For dimensionality reduction, principal component analysis (PCA) or truncated singular value decomposition (TSVD) is adopted to profoundly extract essential features from the RSS data while the K-means algorithm is used to partition the reference points (RPs) into several clusters. Subsequently, the RSS for all virtual points are interpolated via inverse distance weighting (IDW). Numerical results based on the real-world multi-floor multi-building dataset confirm the supremacy of the proposed schemes over the baseline IDW interpolation. Compared to the baseline IDW, the proposed PCA-K-means-IDW, TSVD-K-means-IDW, PCA-IDW, and TSVD-IDW could attain a performance gain in terms of average positioning error of up to 30.17%, 30.93%, 19.33%, and 21.61%, respectively.

Base station (BS) coordination with respect to data and energy cooperation has recently emerged as a potential solution for enhancing the energy efficiency (EE) of multi-cell multi-tier cellular network architecture. This work studies the EE maximization problem in a hybrid-powered (grid and renewable energy source) heterogeneous network (HetNet) where the data and energy are jointly coordinated among the BSs. We propose a combinatorial optimization algorithm to maximize the system EE with the aim to reduce grid power consumption (GPC). Due to the complexity of the formulation, Lagrange dual decomposition and metaheuristic method are incorporated to solve the problem. Furthermore, the non-fractional programming EE problem is solved using the Dinkelbach's method which converges faster with a lower complexity. Simulation results show that cooperation among the BSs to share the channel information and energy reduces the GPC by nearly 20% and increases EE around 10% during harvested energy scarcity among the BSs.