Femtocell technology has been drawing considerable attention as a cost-effective means of improving cellular coverage and capacity. However, under co-channel deployment, femtocells may incur high uplink interference to existing macrocells, and vice versa. To alleviate this interference, we propose a distributed and self-organizing femtocell management architecture, called CTRL Complementary TRi-control Loops), that consists of three control loops. First, for protection of macrocell users' uplink communications, CTRL controls the maximum TX power of femtocell users based on the fedback macrocell's load margin so as to keep, on average, the macrocell load below a certain threshold. Second, CTRL determines the target SINRs of femtocell users, conditioned on the maximum TX power, to reach a Nash equilibrium based on their utility functions, thus achieving efficient coordination of uplink usage among femtocells. Third, for protection of femtocell users' uplink communications, the instantaneous TX power of each femtocell user is controlled to achieve the target SINR against bursty interference from nearby macrocell or femtocell users. Our in-depth evaluation has shown CTRL to successfully preserve the macrocell users' service quality from femtocells' interference and converge to an optimal point under highly dynamic user TX conditions. CTRL is also shown to limit the effects of the estimation errors of channel gains and feedback delay.

In this paper, we develop an optimization framework for the symbiotic operation of a multiuser cognitive radio network (CRN) consisting of a non-orthogonal multiple access (NOMA)-based primary network (PN) and a reconfigurable intelligent surface (RIS)-based secondary network (SN) sharing the same spectrum. In the symbiotic relationship formed in this system model, the RIS devices' PN signal backscattering not only provides multiple primary receivers with a spatial diversity gain but also supports the transmission of the RIS devices' own data to their designated receiver. First, through investigation of a simple network configuration, we obtain insights into the impacts of various factors on the system performance, and thus, into the importance of radio resource allocation for CRN symbiotic operation. Then, we develop a resource allocation framework that maximizes the sum rate of the PN and SN by jointly optimizing the system resources for cases of devices with hardware imperfections. To reduce the complexity of solving this problem, we convert it into a convex weighted minimum mean square error problem. An iterative algorithm for finding the optimal NOMA power allocation for the PN and the reflection coefficients of the RISs is developed, and its overhead is analyzed. Simulation results demonstrate that the proposed algorithm outperforms various benchmarks.

To stream 360-degree video bandwidth-efficiently to virtual reality users, only a subset of a video's spatial tiles corresponding to the current viewport of a user can be transferred, but tile selection is challenging since a user's viewport keeps moving over time. In this letter, we find that the error distribution of fixation point prediction is formally derived as a normal distribution. Based on this theoretical finding, the region where a user's future viewport is likely to reside with a given confidence level is obtained in a closed form and the proposed algorithm selects tiles covering the region.

Offloading heavy virtual reality (VR) computational operations to a network edge computation entity is receiving increasing attention as a tool to wirelessly and energy efficiently provide low-end client devices with high-quality and immersive interactive VR services anytime and anywhere across the globe. In this work, we aim to provide an understanding of various characteristics of VR computation offloading through comprehensive experiments conducted using a prototype testbed for edge-assisted VR processing and streaming. First, we investigate the benefits of VR offloading in terms of computational load and power consumption reduction for a client device compared to standalone operation. Next, we measure VR traffic patterns, including frame size and data and packet rates with various settings, such as different resolution and encoding options. We also measure several performance metrics associated with the quality of experience, namely, frame rate, packet loss rate, and image quality, with various configuration settings. Then, we present latency measurement studies and investigate per-component latency with various settings. Furthermore, we report the rigorous experiments performed to study the impacts of latency and motion patterns on the black borders formed due to image reprojection and the overfilling technique used to eliminate these black borders.

For cases in which a machine learning model needs to be adapted to a new task, various approaches have been developed, including model-agnostic meta-learning (MAML) and transfer learning. In this paper, we investigate how the differences in the data distributions between the old tasks and the new target task impact performance in regression problems. By performing experiments, we discover that these differences greatly affect the relative performance of different adaptation methods. Based on this observation, we develop ensemble schemes combining multiple adaptation methods that can handle a wide range of data distribution differences between the old and new tasks, thus offering more stable performance for a wide range of tasks. For evaluation, we consider three regression problems of sinusoidal fitting, virtual reality motion prediction, and temperature forecasting. The evaluation results demonstrate that the proposed ensemble schemes achieve the best performance among the considered methods in most cases.