Learning the embeddings for urban regions from human mobility data can reveal the functionality of regions, and then enables the correlated but distinct tasks such as crime prediction. Human mobility data contains rich but abundant information, which yields to the comprehensive region embeddings for cross domain tasks. In this paper, we propose multi-graph fusion networks (MGFN) to enable the cross domain prediction tasks. First, we integrate the graphs with spatio-temporal similarity as mobility patterns through a mobility graph fusion module. Then, in the mobility pattern joint learning module, we design the multi-level cross-attention mechanism to learn the comprehensive embeddings from multiple mobility patterns based on intra-pattern and inter-pattern messages. Finally, we conduct extensive experiments on real-world urban datasets. Experimental results demonstrate that the proposed MGFN outperforms the state-of-the-art methods by up to 12.35% improvement. https://github.com/wushangbin/MGFN

The coronavirus disease 2019 (COVID-19) break-out in late December 2019 has spread rapidly worldwide. Existing studies have shown that there is a significant correlation between large-scale human movements and the spread of the epidemic. However, there is a lack of quantification of these correlations, and it is still challenging to predict the spread of the epidemic at early stage. In this paper, we address this issue by conducting a statistical analysis on the spatio-temporal relationship between human mobility and the epidemic spread. Specifically, we proposed an improved SEIR model to adapt to the COVID-19 epidemic, so that we can predict the spread of the epidemic at the early stage using human mobility data and the early confirmed cases. We evaluated our model in various provinces and cities in China, and the results are superior to various baselines, verifying the effectiveness of the method.

Increased network traffic demands have led to ex-tremely dense network deployments. This translates to significant growth in energy consumption at the radio access networks, resulting in high network operation costs (OPEX). In this work, we apply deep reinforcement learning to reduce the energy consumption at the base station in dense wireless networks, by allowing cells that overlap in geographical areas to be put in standby mode according to the changing network conditions. We start by formulating the problem of the cell on/off energy saving in dense wireless networks as a Markov decision process. Then, a deep reinforcement learning (DRL) solution is proposed. This DRL solution takes into account different key performance indicators (KPIs) of both the network and user equipment and aims to reduce the energy consumed by the network without significantly impacting the overall KPIs. The performance of the proposed solution is evaluated using a practical network simulator.

Ultra-reliable low-latency communication (URLLC) services are intrinsically challenging to deliver, with many 5G and future services, including mobile game streaming, adding further complexity by demanding zero service downtime in high-mobility scenarios. Solving these challenges is essential and must be addressed beyond mobile gaming to realise a multitude of current and future services like eX-tended/Virtual Reality(XR/VR) or holoportation in mobile scenarios. Multi-access Edge Computing (MEC) brings services “closer” to user consumption with evident advantages yet at the cost of maintaining a zero downtime guarantee when user handovers (HOs) are prevalent due to the decentralisation of services towards the network edge. In this work, we design and evaluate intelligent HO prediction models between radio 5G Base Stations. The motivation for timely user HO prediction lies in being a vital presupposition for path steering and other MANO control actions in contemporary programmable 5G networks to deliver a zero downtime perception during HO events. Our detailed simulation and actual testbed evaluation results show that effective HO prediction can be achieved using a combination of Long Short-Term Memory (LSTM) or gradient boost regression with classification models, with the latter filtering out any Reference Signal Received Power (RSRP) prediction input outliers for predicting the serving cell.

This paper addressed mobile edge computing (MEC) mobility management using machine learning methods. The mobility decision was based on a reference signal received power (RSRP) value and uncertainty predictor. Neural networks (NNs) were used to establish the predictor, which output the RSRP mean values and standard deviations of different neighbor cells. Closed-form expressions of handover probabilities were derived. With these probabilities, the MEC server was able to cache user services in advance in order to minimize disruptions during handover. Real field data were collected in a typical dense urban area and used to evaluate the performance of the proposed algorithm.

This paper proposes a novel method of constructing a Q-learning framework for link adaptation (LA) in the fifth generation (5G) mobile network. The state-action function is approximated via a neural network (NN). The state space relies on the hybrid automatic repeat request (HARQ) and channel state information (CSI) reports from the user equipment (UE). The output of the Q-learning based LA (QLA) approach consists of the assigned modulation and coding schemes (MCSs) and number of layers, which are used to construct the action space. Reward is calculated based on the HARQ information and the transmit block size (TBS). System level simulation in a typical indoor hotspot scenario has been performed, showing that the proposed QLA outperforms the ordinary LA approach in terms of user throughput.

The significant increase of Machine-Type Communication (MTC) devices has left cellular networks facing unprecedented challenges to support them. One of the major challenges is the Random Access (RA) process performed by idle or disconnected devices. As MTC devices can be highly synchronized, with numerous devices attempting to establish a connection at the same time, the limited available resources during RA allow only for a small number of them to connect to the network, while the rest are required to try again later, increasing their delay. In this paper, we tackle the challenge of the RA process and propose a novel algorithm based on the recent advancements of Non-Orthogonal Random Access (NORA) schemes and the addition of Reinforcement Learning (RL), that is able to increase the number of successful connections, and decrease the network access delay. Our results show that our approach significantly outperforms the currently used approaches.

This study addresses spectral efficiencies (SEs) in a massive multiple‐input multiple‐output cellular network with pilot contamination. Both uplink and downlink SEs are investigated with practical channel state information. A downlink precoder minimising signal detection error and power leakage with considerations on spatial correlation has been derived. Also, a low‐complexity covariance matrix estimation method has been proposed and compared with the existing estimation method in terms of normalised mean‐squared error. It has been demonstrated via simulations that both uplink and downlink SEs using the proposed covariance matrix estimation method can grow with the number of antennas even with a relatively small number of pilot symbols.

A novel unified framework of geometry-based stochastic models for the fifth generation (5G) wireless communication systems is proposed in this paper. The proposed general 5G channel model aims at capturing small-scale fading channel characteristics of key 5G communication scenarios, such as massive multiple-input multiple-output, high-speed train, vehicle-to-vehicle, and millimeter wave communications. It is a 3-D non-stationary channel model based on the WINNER II and Saleh-Valenzuela channel models considering array-time cluster evolution. Moreover, it can easily be reduced to various simplified channel models by properly adjusting model parameters. Statistical properties of the proposed general 5G small-scale fading channel model are investigated to demonstrate its capability of capturing channel characteristics of various scenarios, with excellent fitting to some corresponding channel measurements.

This paper first proposes a three dimensional (3-D) non-stationary wideband multi-confocal ellipsoid channel model with uniform planar antenna array (UPA) for massive multiple-input multiple-output (MIMO) wireless communication systems. The proposed 3-D geometry-based stochastic model (GBSM) not only considers the non-stationary channel characteristics in time domain but also describes the non-stationary channel characteristics in array domain by adopting a birth-death (BD) process and seed algorithm on UPA for the first time. At the meanwhile, selective cluster evolution and cluster evolution areas (CEAs) are first proposed in this paper. Only clusters in CEAs go through cluster evolution because near-field effect and other clusters can be observed by all antennas. Channel parameters, including delay, Doppler frequency, angle of departure (AoD), and angle of arrive (AoA), are considered in both the azimuth direction and elevation direction. This paper also considers rotations of UPAs. Based on the proposed theoretical reference model, the relevant simulation model is also obtained. The statistical properties of theoretical reference model and simulation model can match well with numerical results.

This study addresses the achievable rate of single cell and sum rate of multi‐cell orthogonal frequency division multiplexing (OFDM) index modulation (IM). The single‐cell achievable rate of OFDM‐IM with Gaussian input is calculated using a multi‐ary symmetric channel. Then, the cumulative distribution function of multi‐cell OFDM‐IM is investigated by stochastic geometry. Furthermore, it is proved in this study that the probability density function of noise plus inter‐cell interference in multi‐cell OFDM‐IM with quadrature‐amplitude modulation follows a mixture of Gaussians (MoGs) distribution. Next, parameters of the MoG distribution are estimated using a simplified expectation maximisation algorithm. Upper bound of sum rates of multi‐cell OFDM‐IM is derived. Furthermore, analytic and simulated results are compared and discussed.

This paper investigates intra-cluster channel characteristics of non line-of-sight (NLOS) 28 GHz channels in street canyon scenarios. These channel characteristics include cluster numbers, number of subpaths within each cluster, intra-cluster delay spreads, and intra-cluster angular spreads. Both measurement and ray tracing results are presented and compared. Furthermore, distribution fittings are performed and models and parameters for different intra-cluster channel characteristics are proposed.

This paper first proposes a novel three dimensional (3-D) wideband multi-confocal ellipsoid model for wireless multiple-input multiple-output (MIMO) communication channels. The proposed 3-D geometry-based stochastic model (GBSM) describes the channel in both the azimuth direction and elevation direction, including delay, Doppler frequency, angle of departure (AoD), and angle of arrive (AoA). It can be shown that the ellipsoid model can capture the effect of the clusters more accurately than the previous elliptic-cylinder model. Using the proposed theoretical model as the reference model, the corresponding simulation model is also derived. Numerical results demonstrate that their statistical properties can match well.

This paper proposes a novel method for ergodic sum rate analysis of spatial modulation (SM) systems with maximum likelihood receiver. This method is developed based on the MT-ary symmetric channel, where MTis the number of transmit antennas. The probability of antenna detection error is approximated by the pair-wise error probability. Then, an approximation to the ergodic sum rate of information transmission via SM with maximum likelihood receiver is computed. It is demonstrated via simulation that the proposed analysis method is able to provide an excellent approximation to the ergodic sum rate of SM.

This paper proposes a novel non-stationary wideband multi-confocal ellipse two dimensional (2-D) channel model for massive multiple-input multiple-output (MIMO) communication systems. Spherical wavefront is assumed in the proposed channel model, instead of the plane wavefront assumption used in conventional MIMO channel models. In addition, the birth-death process is incorporated into the proposed model to capture the dynamic properties of clusters on both the array and time axes. Statistical properties of the channel model such as the space-time-frequency correlation function and power imbalance on the antenna array are studied. The impact of the spherical wavefront assumption on the statistical properties of the channel model is investigated. Furthermore, numerical analysis shows that the proposed channel model is able to capture specific characteristics of massive MIMO channel as observed in measurements.

This paper analyzes the capacity of the finite scatterer (FS) multiple-input multiple-output (MIMO) channel model with finite angular resolution. The rank of the channel is bounded by not only the number of antennas but also the finite angular resolution of antenna arrays. The probability mass function (PMF) of rank is studied and derived based on the Stirling number of the second kind. As a result, the closed-form expression of the capacity of a FS MIMO channel model is obtained. Furthermore, asymptotic analyses such as the large-scale system analysis, infinite scatterer analysis, and the multiplexing gain analysis are investigated.

This paper proposes a novel theoretical non-stationary three dimensional (3-D) wideband twin-cluster channel model for massive multiple-input multiple-output (MIMO) communication systems with carrier frequencies on the order of gigahertz (GHz). As the dimension of antenna arrays cannot be ignored for massive MIMO, near field effects instead of far field effects are considered in the proposed model. These include the spherical wavefront assumption and a birth-death process to model non-stationary properties of clusters such as cluster appearance and disappearance on both the array and time axes. Their impacts on massive MIMO channels are investigated via statistical properties including correlation functions, condition numbers, and angular power spectra. Additionally, the impact of elevation angles on correlation functions is discussed. A corresponding simulation model for the theoretical model is also proposed. Finally, numerical analysis shows that the proposed channel models are able to serve as a design framework for massive MIMO channel modeling.