Fueled by the availability of more data and computing power, recent breakthroughs in cloud-based machine learning (ML) have transformed every aspect of our lives from face recognition and medical diagnosis to natural language processing. However, classical ML exerts severe demands in terms of energy, memory, and computing resources, limiting their adoption for resource-constrained edge devices. The new breed of intelligent devices and high-stake applications (drones, augmented/virtual reality, autonomous systems, and so on) requires a novel paradigm change calling for distributed, low-latency and reliable ML at the wireless network edge (referred to as edge ML). In edge ML, training data are unevenly distributed over a large number of edge nodes, which have access to a tiny fraction of the data. Moreover, training and inference are carried out collectively over wireless links, where edge devices communicate and exchange their learned models (not their private data). In a first of its kind, this article explores the key building blocks of edge ML, different neural network architectural splits and their inherent tradeoffs, as well as theoretical and technical enablers stemming from a wide range of mathematical disciplines. Finally, several case studies pertaining to various high-stake applications are presented to demonstrate the effectiveness of edge ML in unlocking the full potential of 5G and beyond.

/pdf/wireless-network-intelligence-at-the-edge-36kakqk6fq.pdf

Wireless Network Intelligence at the Edge

Fueled by the availability of more data and computing power, recent breakthroughs in cloud-based machine learning (ML) have transformed every aspect of our lives from face recognition and medical diagnosis to natural language processing. However, classical ML exerts severe demands in terms of energy, memory and computing resources, limiting their adoption for resource constrained edge devices. The new breed of intelligent devices and high-stake applications (drones, augmented/virtual reality, autonomous systems, etc.), requires a novel paradigm change calling for distributed, low-latency and reliable ML at the wireless network edge (referred to as edge ML). In edge ML, training data is unevenly distributed over a large number of edge nodes, which have access to a tiny fraction of the data. Moreover training and inference is carried out collectively over wireless links, where edge devices communicate and exchange their learned models (not their private data). In a first of its kind, this article explores key building blocks of edge ML, different neural network architectural splits and their inherent tradeoffs, as well as theoretical and technical enablers stemming from a wide range of mathematical disciplines. Finally, several case studies pertaining to various high-stake applications are presented demonstrating the effectiveness of edge ML in unlocking the full potential of 5G and beyond.

Machine learning tools are finding interesting applications in millimeter wave (mmWave) and massive MIMO systems. This is mainly thanks to their powerful capabilities in learning unknown models and tackling hard optimization problems. To advance the machine learning research in mmWave/massive MIMO, however, there is a need for a common dataset. This dataset can be used to evaluate the developed algorithms, reproduce the results, set benchmarks, and compare the different solutions. In this work, we introduce the DeepMIMO dataset, which is a generic dataset for mmWave/massive MIMO channels. The DeepMIMO dataset generation framework has two important features. First, the DeepMIMO channels are constructed based on accurate ray-tracing data obtained from Remcom Wireless InSite. The DeepMIMO channels, therefore, capture the dependence on the environment geometry/materials and transmitter/receiver locations, which is essential for several machine learning applications. Second, the DeepMIMO dataset is generic/parameterized as the researcher can adjust a set of system and channel parameters to tailor the generated DeepMIMO dataset for the target machine learning application. The DeepMIMO dataset can then be completely defined by the (i) the adopted ray-tracing scenario and (ii) the set of parameters, which enables the accurate definition and reproduction of the dataset. In this paper, an example DeepMIMO dataset is described based on an outdoor ray-tracing scenario of 18 base stations and more than one million users. The paper also shows how this dataset can be used in an example deep learning application of mmWave beam prediction.

DeepMIMO: A Generic Deep Learning Dataset for Millimeter Wave and Massive MIMO Applications.

As wireless networks evolve towards high mobility and providing better support for connected vehicles, a number of new challenges arise due to the resulting high dynamics in vehicular environments and thus motive rethinking of traditional wireless design methodologies. Future intelligent vehicles, which are at the heart of high mobility networks, are increasingly equipped with multiple advanced onboard sensors and keep generating large volumes of data. Machine learning, as an effective approach to artificial intelligence, can provide a rich set of tools to exploit such data for the benefit of the networks. In this article, we first identify the distinctive characteristics of high mobility vehicular networks and motivate the use of machine learning to address the resulting challenges. After a brief introduction of the major concepts of machine learning, we discuss its applications to learn the dynamics of vehicular networks and make informed decisions to optimize network performance. In particular, we discuss in greater detail the application of reinforcement learning in managing network resources as an alternative to the prevalent optimization approach. Finally, some open issues worth further investigation are highlighted.

Toward Intelligent Vehicular Networks: A Machine Learning Framework

Channel estimation and hybrid precoding are considered for multi-user millimeter wave massive multi-input multi-output system. A deep learning compressed sensing (DLCS) channel estimation scheme is proposed. The channel estimation neural network for the DLCS scheme is trained offline using simulated environments to predict the beamspace channel amplitude. Then the channel is reconstructed based on the obtained indices of dominant beamspace channel entries. A deep learning quantized phase (DLQP) hybrid precoder design method is developed after channel estimation. The training hybrid precoding neural network for the DLQP method is obtained offline considering the approximate phase quantization. Then the deployment hybrid precoding neural network (DHPNN) is obtained by replacing the approximate phase quantization with ideal phase quantization and the output of the DHPNN is the analog precoding vector. Finally, the analog precoding matrix is obtained by stacking the analog precoding vectors and the digital precoding matrix is calculated by zero-forcing. Simulation results demonstrate that the DLCS channel estimation scheme outperforms the existing schemes in terms of the normalized mean-squared error and the spectral efficiency, while the DLQP hybrid precoder design method has better spectral efficiency performance than other methods with low phase shifter resolution.

/pdf/sparse-channel-estimation-and-hybrid-precoding-using-deep-1jyy5e9ddm.pdf

Sparse Channel Estimation and Hybrid Precoding Using Deep Learning for Millimeter Wave Massive MIMO

Millimeter-wave communication is a challenge in the highly mobile vehicular context. Traditional beam training is inadequate in satisfying low overheads and latency. In this paper, we propose to combine machine learning tools and situational awareness to learn the beam information (power, optimal beam index, etc) from past observations. We consider forms of situational awareness that are specific to the vehicular setting including the locations of the receiver and the surrounding vehicles. We leverage regression models to predict the received power with different beam power quantizations. The result shows that situational awareness can largely improve the prediction accuracy and the model can achieve throughput with little performance loss with almost zero overhead.

MmWave Beam Prediction with Situational Awareness: A Machine Learning Approach

High mobility poses challenges to millimeter wave (mmWave) vehicular beam tracking. In this paper, we propose to recommend a set of candidate beam pairs using machine learning classification and ranking, based on the target vehicle’s situational awareness. We observe that the optimal beam directions can be captured by the location of the receiver and the sizes and locations of the neighboring vehicles. This information is a natural part of the situational awareness exploited by connected and automated vehicles. The infrastructure conducts the beam search within the recommended beam set with the receiver. To further reduce overhead, we leverage a new beam search mechanism, where the search can automatically terminate when the current beam measurement exceeds a specific target power. We set the target power as the predicted average power of the top two beams. The numerical results show that the beam search overhead can be reduced by 70% when seven beams are recommended, at a comparable beam alignment probability.

Towards Robustness: Machine Learning for MmWave V2X with Situational Awareness

5G new radio (NR) requires a very dense deployment of cellular infrastructure, which poses a great economical challenge if traditional fiber backhaul links are used. To cope with this problem, 3GPP introduced integrated access and backhaul (IAB), aiming to use wireless backhaul to provide high capacity and great deployment flexibility. The multi-hop and multi-route topology for this new type of access networks challenges the existing retransmission-/repetition-based reliability-enhancing technique, such as ARQ/HARQ and packet duplication. However, it also opens up new opportunities for novel reliability enhancement techniques. In this paper, taking advantage of the more complex network topology of IAB, we propose to use linear network coding as a potentially better solution to improve end-to-end latency and reliability. We discuss its placement in the IAB protocol stack, and propose two novel schemes to improve the performance of network coding in the IAB network: the rate-proportional traffic splitting scheme in the multi-route scenario, and the adaptive coded-forwarding scheme in the multi-hop scenario. Simulation results show that in some typical IAB scenarios our network coding solution has considerable performance gains over the existing repetition-based technology.

Network Coding for Integrated Access and Backhaul Wireless Networks

With the increasing densification of cellular networks, it has become exceedingly difficult to provide traditional fiber backhaul access to each cell site, which is especially true for small cell base stations (SBSs). The increasing maturity of millimeter wave (mmWave) communication coupled with multiple-input-multiple-output (MIMO) and beamforming technologies has opened up the possibility of providing high-speed wireless backhaul to such cell sites. The third generation partnership project (3GPP) is defining an integrated access and backhaul (IAB) architecture for the fifth generation (5G) cellular networks, in which the same infrastructure and spectral resources are used for both, the access and the backhaul. In IAB networks, SBSs, so called IAB nodes, act either as relay nodes carrying the traffic through multiple hops from a macro cell to an end user and vice versa or as access points to serve user equipment (UEs) in their proximity. To this end, the topology of such IAB networks is essential to enable efficient traffic flow and minimize congestion or increase robustness to backhaul link failure. In this paper, we propose a topology formation algorithm together with methodologies to implement it in real networks and compare it with a standard random sequence approach as well as with an optimal topology obtained using dynamic programming. Our simulation results demonstrate that the proposed algorithm outperforms the random sequence approach by 26\% on average in terms of lower bound of the network capacity and is up to 99.7\% close to the optimal solution, while being significantly less complex.

/pdf/optimal-topology-formation-and-adaptation-of-integrated-2t68r8ymjl.pdf

Narasimha Murali

Papers

MmWave Beam Prediction with Situational Awareness: A Machine Learning Approach

MmWave Beam Prediction with Situational Awareness: A Machine Learning Approach

Towards Robustness: Machine Learning for MmWave V2X with Situational Awareness

Network Coding for Integrated Access and Backhaul Wireless Networks

Optimal Topology Formation and Adaptation of Integrated Access and Backhaul Networks