Huawei

About: Huawei is a company organization based out in Shenzhen, China. It is known for research contribution in the topics: Terminal (electronics) & Signal. The organization has 41417 authors who have published 44698 publications receiving 343496 citations. The organization is also known as: Huawei Technologies & Huawei Technologies Co., Ltd..

Machine-to-Machine Communications in Ultra-Dense Networks—A Survey

Abstract: To achieve 1000-fold capacity increase in 5G wireless communications, ultradense network (UDN) is believed to be one of the key enabling technologies. Most of the previous research activities on UDNs were based very much on human-to-human communications. However, to provide ubiquitous Internet of Things services, machine-to-machine (M2M) communications will play a critical role in 5G systems. As the number of machine-oriented connections increases, it is expected that supporting M2M communications is an essential requirement in all future UDNs. In this paper, we aim to bridge the gaps between M2M communications and UDNs, which were commonly considered as two separate issues in the literature. The paper begins with a brief introduction on M2M communications and UDNs, and then will discuss the issues on the roles of M2M communications in future UDNs. We will identify different ways to implement M2M communications in the UDNs from the perspectives of layered architecture, including physical, media access control, network, and application layers. Other two important issues, i.e., security and network virtualization, will also be addressed. Before the end of this paper, we will give a summary on identified research topics for future studies.

Downlink Cellular Network Analysis With LOS/NLOS Propagation and Elevated Base Stations

Abstract: In this paper, we investigate the downlink performance of dense cellular networks with elevated base stations (BSs) using a channel model that incorporates line-of-sight (LOS)/non-line-of-sight (NLOS) propagation into both small-scale and large-scale fading. Modeling LOS fading with Nakagami- $m$ fading, we provide a unified framework based on stochastic geometry that encompasses both closest and strongest BS association. This paper is particularized to two distance-dependent LOS/NLOS models of practical interest. Considering the effect of LOS propagation alone, we derive closed-form expressions for the coverage probability with Nakagami- $m$ fading, showing that the performance for strongest BS association is the same as in the case of Rayleigh fading, whereas for closest BS association it monotonically increases with the shape parameter $m$ . Then, focusing on the effect of elevated BSs, we show that network densification eventually leads to near-universal outage even for moderately low BS densities: in particular, the maximum area spectral efficiency is proportional to the inverse of the square of the BS height.

Next-generation PON-part II: Candidate systems for next-generation PON

Abstract: Given the requirements for a next-generation PON, the architecture of the system solution must be considered. There are many different systems that can provide the services and system-level features desired for a next generation PON; however, each has its own challenges and advantages. This article presents the set of possible solutions, and puts them into perspective of likely standardization. It also considers how the key requirement of coexistence could be accommodated.

JHU ASpIRE system: Robust LVCSR with TDNNS, iVector adaptation and RNN-LMS

Abstract: Multi-style training, using data which emulates a variety of possible test scenarios, is a popular approach towards robust acoustic modeling. However acoustic models capable of exploiting large amounts of training data in a comparatively short amount of training time are essential. In this paper we tackle the problem of reverberant speech recognition using 5500 hours of simulated reverberant data. We use time-delay neural network (TDNN) architecture, which is capable of tackling long-term interactions between speech and corrupting sources in reverberant environments. By sub-sampling the outputs at TDNN layers across time steps, training time is substantially reduced. Combining this with distributed-optimization we show that the TDNN can be trained in 3 days using up to 32 GPUs. Further, iVectors are used as an input to the neural network to perform instantaneous speaker and environment adaptation. Finally, recurrent neural network language models are applied to the lattices to further improve the performance. Our system is shown to provide state-of-the-art results in the IARPA ASpIRE challenge, with 26.5% WER on the dev Jest set.

System and scheme of scalable OFDM numerology

Abstract: For a wireless communications system, scalable orthogonal frequency division multiplexing (OFDM) numerology is incorporated in a manner that can apply to radio link transmissions in future wireless network for frequency division duplex (FDD) and time division duplex (TDD) communications.