This paper tackles the problem of the worst-class error rate, instead of the standard error rate averaged over all classes. For example, a three-class classification task with class-wise error rates of 10%, 10%, and 40% has a worst-class error rate of 40%, whereas the average is 20% under the class-balanced condition. The worst-class error is important in many applications. For example, in a medical image classification task, it would not be acceptable for the malignant tumor class to have a 40% error rate, while the benign and healthy classes have a 10% error rates. To avoid overfitting in worst-class error minimization using Deep Neural Networks (DNNs), we design a problem formulation for bounding the worst-class error instead of achieving zero worst-class error. Moreover, to correctly bound the worst-class error, we propose a boosting approach which ensembles DNNs. We give training and generalization worst-class-error bound. Experimental results show that the algorithm lowers worst-class test error rates while avoiding overfitting to the training set.

Massive multiple-input multiple-output (MIMO) has been considered as one of the key technologies in 5G communication, owing to its promising potential to increase the spectrum efficiency significantly. The capacity grows rapidly with the increasing number of antennas. However, in real-time environment, the more antennas are employed, the more computing resources are required, and this increase is even in a nonlinear progression. In this work, radio computing architecture (RCA) with multiple parallel general purpose processors (GPPs) and automatic distribution system (ADS) are utilized to overcome the strict real-time constraints and establish a highly flexible prototype platform for massive MIMO research. The feasibility and performance of the GPP-based prototype has been investigated through field trial. A collocated 64-RF-channel base station (BS) with each RF channel driving 3 antennas can simultaneously serve twelve 8-antenna user equipments (UEs). The maximum overall bandwidth is 200 MHz, and the maximum flow number is pre-defined as 24. In this field trial, a peak total user throughput of 11.29 Gbps and a cell spectrum efficiency of 79.82 bps/Hz are achieved. As far as we know, this is the world's highest throughput ever achieved in the sub 6 GHz band. Furthermore, apart from multi-user MIMO, single-user MIMO is also tested in this field trial.

The 5th generation of mobile communications system (5G) is nowadays expected to support very diverse applications, devices, and services such as enhanced mobile broadband (eMBB) and the Internet of things (IoTs). In order to meet the future requirements, new 5G candidate radio access technologies have been proposed and studied. For example, Massive multiple-input multiple-output (MIMO) is an attractive technology to further improve the spectrum efficiency and to boost system capacity. This paper presents a large scale field experimental trial of downlink time division duplex (TDD) Massive MIMO using the developed 5G test bed at the 4.5 GHz band. Specifically, the system performance of multi-user (MU) Massive MIMO with different numbers of sets of user equipment (UEs) and spatial layers is investigated, including the impact of different inter-UE separations and UE deployments, which have not been well investigated compared to the previous studies. The experimental results show that the throughput performance for a wide UE separation linearly increases as the number of layers and UEs increases, while the performance improvement is diminished when inter-UE separation narrows. In addition, the maximum total user throughput of 11.29 Gbps, which corresponds to a spectrum efficiency of 79.82 bps/Hz/cell, was achieved with 24 layers. The results indicate that TDD Massive MIMO using the developed 5G test bed greatly improves the spectrum efficiency for large scale MU-MIMO.

Spectral efficiency is always a key factor to be improved and optimized along mobile communication networks evolving generation by generation. 5G enabling technologies must take spectral efficiency into consideration. In this paper, we show the performance of three key 5G technologies in sense of spectral efficiency improvement. Sparse code multiple access, polar codes, and filtered orthogonal frequency-division multiplexing are novel multiple access technology, channel coding scheme, and waveform, respectively. The combination of them is implemented in a 5G field trial testbed by NTT DOCOMO and Huawei for the first time. According to the field test results, we achieve over 100% spectral efficiency improvement comparison with baseline, where orthogonal frequency-division multiple access and turbo coding as LTE are used.

Recently, NTT DOCOMO and Huawei conducted a large-scale experimental trial of key technologies for the 5th generation (5G) mobile systems. Downlink multi-user (MU) transmission with massive MIMO in the time-division duplex (TDD) mode is one of the technologies evaluated in this trial. With linear and non-linear downlink precoding schemes, we investigated the practical performance of MU massive MIMO systems under different numbers of users, user distributions, and different radio frequency channel calibration settings with up to 24 users deployed. When linear precoding is used, the cell spectrum efficiency reaches 39 bit/s/Hz, and it reaches 43 bit/s/Hz when non-linear precoding is used. In term of the cell throughput, up to 1.35 Gbps cell throughput was observed with a linear precoder and 100 MHz bandwidth; and 343 Mbps cell throughput was observed with a non-linear precoder and 20 MHz bandwidth. Based on these investigations, we verified the feasibility and performance of a TDD massive MIMO system for 5G mobile systems, and studied the impact of key factors on the system performance.

Non-orthogonal multiple access (NOMA) is a promising multiple access scheme for further improving the spectrum efficiency compared to that for orthogonal multiple access (OMA) in the 5th Generation (5G) mobile communication systems. All of the existing evaluations for NOMA focus on the macrocell deployment since NOMA fully utilizes the power domain and the difference in channel gains, e.g., path loss, between users, which is typically sufficiently large in macrocells. Currently, small cells are becoming important and being studied for future Long-Term Evolution (LTE) enhancements in order to improve further the system performance. Thus, it is of great interest to study the performance of NOMA for small cell deployment under various environments. This paper investigates the system level performance of NOMA in small cells considering practical assumptions such as the single user multiple-input multiple-output (SU-MIMO) technique, adaptive modulation and coding (AMC), feedback channel quality indicator (CQI). Some of the key NOMA specific functionalities, including multi-user paring and transmit power allocation are also taken into account in the evaluation. Based on computer simulations, we show that for both macrocell and small cell deployments, NOMA can still provide a larger throughput performance gain compared to that for OMA.

We investigate the influence of channel estimation error on the achievable system-level throughput performance of non-orthogonal access with successive interference cancellation (SIC) in the cellular downlink. The channel estimation error in non-orthogonal access causes residual interference in the SIC process, which decreases the achievable user throughput. Furthermore, the channel estimation error causes error in the transmission rate control for the respective users, which may result in decoding error at not only the destination user terminal but also other user terminals for SIC. However, we show that by using a simple transmission rate back-off algorithm, the impact of the channel estimation error is effectively alleviated and non-orthogonal access with SIC achieves clear average and cell-edge user throughput gains relative to orthogonal access similar to the case with perfect channel estimation.

As a promising downlink multiple access scheme for further LTE enhancement and future radio access (FRA), this paper investigates the system-level performance of non-orthogonal multiple access (NOMA) with a successive interference canceller (SIC) on the receiver side. The goal is to clarify the potential gains of NOMA over orthogonal multiple access (OMA) such as OFDMA, taking into account key link adaptation functionalities of the LTE radio interface such as adaptive modulation and coding (AMC), hybrid automatic repeat request (HARQ), time/frequency-domain scheduling, and outer loop link adaptation (OLLA), in addition to NOMA specific functionalities such as dynamic multi-user power allocation. Based on computer simulations, we show under multiple configurations that the system-level performance achieved by NOMA is superior to that for OMA.

This paper presents a non-orthogonal multiple access (NOMA) concept for cellular future radio access (FRA) towards the 2020s information society. Different from the current LTE radio access scheme (until Release 11), NOMA superposes multiple users in the power domain although its basic signal waveform could be based on the orthogonal frequency division multiple access (OFDMA) or the discrete Fourier transform (DFT)-spread OFDM the same as LTE baseline. In our concept, NOMA adopts a successive interference cancellation (SIC) receiver as the baseline receiver scheme for robust multiple access, considering the expected evolution of device processing capabilities in the future. Based on system-level evaluations, we show that the downlink NOMA with SIC improves both the capacity and cell-edge user throughput performance irrespective of the availability of the frequency-selective channel quality indicator (CQI) on the base station side. Furthermore, we discuss possible extensions of NOMA by jointly applying multi-antenna/site technologies with a proposed NOMA/MIMO scheme using SIC and an interference rejection combining (IRC) receiver to achieve further capacity gains, e.g., a three-fold gain in the spectrum efficiency representing a challenging target for FRA.