Back

Deep Learning in Mobile and Wireless Networking: A Survey

Chaoyun ZhangPaul PatrasHamed Haddadi

article2018IEEE Communications Surveys and Tutorials1,563 citations

Provides a comprehensive taxonomy and critical review of how deep learning architectures are applied across mobile and wireless networking domains, offering practical guidance on deploying neural models onto resource-constrained mobile systems.

Listen

Mobile and wireless networking infrastructures are experiencing an unprecedented surge in data traffic driven by the rapid growth of smart devices, Internet of Things (IoT) technologies, and demanding 5G applications. Traditional network management, optimization methods, and classical shallow machine learning tools struggle to scale, automate feature engineering, and handle the high dimensionality of these complex environments. Incorporating deep learning into mobile and wireless systems offers a viable path to automate data analytics, optimize network operations, and meet stringent low-latency and high-throughput requirements.

The article delivers a comprehensive survey evaluating the intersection of deep learning and mobile and wireless networking. Its primary objective is to evaluate how modern deep neural network architectures can be integrated into network analytics, control, and edge computing, while identifying existing deployment enablers, tailoring strategies, and open research challenges.

The authors synthesized extensive literature across deep learning advancements and mobile networking paradigms, categorizing contributions across domains such as network-level traffic forecasting, mobile pattern recognition, user mobility and localization, edge computing, network security, and physical layer signal processing. The review specifically assesses enabling platforms, specialized optimization algorithms, and model adaptation strategies suited for distributed and resource-constrained environments.

The article highlights several key findings regarding the deployment and utility of deep learning in mobile networks. First, deep learning models significantly outperform traditional statistical and shallow machine learning methods across core tasks: for example, image-inspired super-resolution techniques applied to mobile traffic data can enhance measurement granularity by up to 100 times, and specialized hardware accelerators can process roughly 2,000 inferences per second at low power. Second, convolutional and recurrent neural network variants excel at capturing complex spatial and temporal correlations, transforming raw traffic snapshots and sensor streams into accurate forecasts. Third, deep reinforcement learning proves capable of solving high-dimensional network control problems, such as proactive resource allocation and mobility management, which are computationally intractable under traditional mathematical optimization. Finally, practical deployment relies on an evolving ecosystem of specialized hardware, distributed computing platforms, and optimization algorithms that enable millisecond-level inference.

These findings imply that mobile network operators can transition from reactive, manual network management to automated, proactive decision-making. Adopting deep learning reduces operational overhead and improves Quality of Experience (QoE), but also introduces critical tradeoffs: deep models require substantial compute, can be computationally heavy for edge devices, remain vulnerable to adversarial security attacks, and suffer from low interpretability. Consequently, leaders should consider deep learning primarily as an intelligent decision-support layer rather than a completely autonomous system.

For future implementation and next steps, organizations should prioritize hybrid deployment strategies—leveraging edge devices for lightweight, latency-sensitive tasks via model compression and quantization, while utilizing distributed cloud platforms for large-scale training. Further research and development should focus on enhancing model interpretability, developing robust defenses against adversarial attacks, and refining unsupervised and federated learning mechanisms to preserve user privacy. Because the article synthesizes experimental and theoretical literature across diverse problem settings, decision-makers should maintain moderate confidence in cross-domain generalizability and conduct thorough pilots on real-world production networks before executing full-scale, automated deployments.

Imperial College LondonUniversity of Edinburghagile methodsdeep learningmobile applicationsdeep learning driven mobile networkswireless networkingcellular traffic prediction
arXiv: 1803.04311
Cover for Deep Learning in Mobile and Wireless Networking: A Survey

Abstract

The rapid uptake of mobile devices and the rising popularity of mobile applications and services pose unprecedented demands on mobile and wireless networking infrastructure. Upcoming 5G systems are evolving to support exploding mobile traffic volumes, agile management of network resource to maximize user experience, and extraction of fine-grained real-time analytics. Fulfilling these tasks is challenging, as mobile environments are increasingly complex, heterogeneous, and evolving. One potential solution is to resort to advanced machine learning techniques to help managing the rise in data volumes and algorithm-driven applications. The recent success of deep learning underpins new and powerful tools that tackle problems in this space.

In this paper we bridge the gap between deep learning and mobile and wireless networking research, by presenting a comprehensive survey of the crossovers between the two areas. We first briefly introduce essential background and state-of-the-art in deep learning techniques with potential applications to networking. We then discuss several techniques and platforms that facilitate the efficient deployment of deep learning onto mobile systems. Subsequently, we provide an encyclopedic review of mobile and wireless networking research based on deep learning, which we categorize by different domains. Drawing from our experience, we discuss how to tailor deep learning to mobile environments. We complete this survey by pinpointing current challenges and open future directions for research.

Topics

agile methodsdeep learningmobile applicationsmobile environmentsmobile networksmobile systemsuser experiencewirelessconvolutional neural networks

Keywords

deep learning driven mobile networkswireless networkingcellular traffic predictiontraffic classificationmobility predictiondeep autoencoderLSTM networkedge networks

Table of Contents

  • I Introduction
  • II Related High-level Articles and The Scope of This Survey
  • II-A Overviews of Deep Learning and its Applications
  • II-B Surveys on Future Mobile Networks
  • II-C Deep Learning Driven Networking Applications
  • II-D Our Scope
  • III Deep Learning 101
  • III-A The Evolution of Deep Learning
  • III-B Fundamental Principles of Deep Learning
  • III-C Forward and Backward Propagation
  • III-D Advantages of Deep Learning in Mobile and Wireless Networking
  • III-E Limitations of Deep Learning in Mobile and Wireless Networking
  • IV Enabling Deep Learning in Mobile Networking
  • IV-A Advanced Parallel Computing
  • IV-B Distributed Machine Learning Systems
  • IV-C Dedicated Deep Learning Libraries
  • IV-D Fast Optimization Algorithms
  • IV-E Fog Computing
  • V Deep Learning: State-of-the-Art
  • V-A Multilayer Perceptron
  • V-B Boltzmann Machine
  • V-C Auto-Encoders
  • V-D Convolutional Neural Network
  • V-E Recurrent Neural Network
  • V-F Generative Adversarial Network
  • V-G Deep Reinforcement Learning
  • VI Deep Learning Driven Mobile and Wireless Networks
  • VI-A Mobile Big Data as a Prerequisite
  • VI-B Deep Learning Driven Network-level Mobile Data Analysis
  • VI-C Deep Learning Driven App-level Mobile Data Analysis
  • VI-D Deep Learning Driven Mobility Analysis
  • VI-E Deep Learning Driven User Localization
  • VI-F Deep Learning Driven Wireless Sensor Networks
  • VI-G Deep Learning Driven Network Control
  • VI-H Deep Learning Driven Network Security
  • VI-I Deep Learning Driven Signal Processing
  • VI-J Emerging Deep Learning Applications in Mobile Networks
  • VII Tailoring Deep Learning to Mobile Networks
  • VII-A Tailoring Deep Learning to Mobile Devices and Systems
  • VII-B Tailoring Deep Learning to Distributed Data Containers
  • VII-C Tailoring Deep Learning to Changing Mobile Network Environments
  • VIII Future Research Perspectives
  • VIII-A Serving Deep Learning with Massive High-Quality Data
  • VIII-B Deep Learning for Spatio-Temporal Mobile Data Mining
  • VIII-C Deep learning for Geometric Mobile Data Mining
  • VIII-D Deep Unsupervised Learning in Mobile Networks
  • VIII-E Deep Reinforcement Learning for Mobile Network Control
  • VIII-F Summary
  • References

Knowls

  1. Knowl 1 — Taxonomy of Mobile Big Data for Deep Learning

    definition

    Mobile big data in cellular and wireless networking is categorized into two major classes based on the collection point and data characteristics:

    1. Network-Level Data: Data recorded across the cellular network infrastructure, comprising:

      • Infrastructure Metadata: Base station locations, radio access technology capabilities, antenna topologies, and equipment configurations.
      • Key Performance Indicators (KPIs): Time-series metrics including aggregated traffic volume, end-to-end packet delay, packet loss rates, jitter, and Quality of Experience (QoE) estimates.
      • Call Detail Records (CDRs): Telecommunication transaction logs recording session start and end timestamps, connected cell identifiers, service types (SMS, voice, data), and sender-receiver metadata.
      • Radio Information: Channel State Information (CSI), received signal strength indicators (RSSI), signal-to-noise ratios (SNR), carrier frequencies, spectrum occupancy, and modulation schemes.

    2. App-Level Data: Data gathered directly from user equipment, edge sensors, and applications, comprising:

      • Device Profiles: Device make/model, operating system version, hardware capabilities, and Media Access Control (MAC) addresses.
      • Sensor Streams: Temporal signals from physical sensors including Global Positioning System (GPS), accelerometers, gyroscopes, magnetometers, barometers, and Passive Infra-Red (PIR) motion detectors.
      • Application and System Logs: Multimodal content (voice, video, images), text entries, user interaction dynamics, crash reports, and operating system diagnostic traces.

  2. Knowl 2 — Comparative Characteristics of Deep Learning Architectures in Mobile Networking

    model/method

    Different deep learning model families present specific operational characteristics, strengths, limitations, and suitable mobile networking applications:

    Architecture Primary Paradigm Strengths Weaknesses Target Networking Tasks
    Multilayer Perceptron (MLP) Supervised/Unsupervised Simple implementation Dense connections; poor scaling Baseline modeling; multi-attribute classification
    Restricted Boltzmann Machine (RBM) Unsupervised Robust latent representations Computationally heavy training Unsupervised pretraining; feature extraction
    Auto-Encoder (AE / DAE / VAE) Unsupervised Nonlinear dimension reduction High pretraining overhead Anomaly detection; network compression; denoising
    Convolutional Neural Network (CNN) Supervised/RL Spatial locality; weight sharing High compute cost for large filters Spatio-temporal traffic snapshots; RF modulation
    Recurrent Neural Network (RNN / LSTM) Supervised/RL Captures temporal dependencies Vanishing/exploding gradients Sequential traffic forecasting; mobility tracking
    Generative Adversarial Network (GAN) Unsupervised Learns target data distributions Unstable minimax convergence Synthetic dataset generation; data augmentation
    Deep Reinforcement Learning (DRL) Reinforcement Learning Handles high-dimensional states Sample inefficiency; slow convergence Dynamic resource allocation; routing; BS sleep control

  3. Knowl 3 — Paradigms for Deep Learning Driven Mobile Network Control

    model/method

    Deep learning driven network control operates across three distinct architectural paradigms:

    1. Deep Reinforcement Learning (DRL): An agent parameterized by a deep neural network continuously interacts with a network environment (or a high-fidelity simulator). At time step tt, the agent receives a state representation sts_t, executes an action ata_t chosen according to its policy π(at∣st)\pi(a_t \mid s_t) or action-value function Q(st,at)Q(s_t, a_t), and receives a scalar reward rtr_t. The neural network is optimized via policy gradient or Q-learning variants to maximize the expected cumulative discounted return. It is primarily applied to dynamic spectrum access, power control, transmission scheduling, and adaptive bitrate streaming.

    2. Deep Imitation Learning: Also termed "learning by demonstration," this paradigm trains a deep neural network in a supervised manner to mimic the decisions of an expert teacher or a computationally intensive optimal solver (such as the Weighted Minimum Mean Square Error algorithm or Open Shortest Path First routing). Once trained, the neural network acts as a fast function approximator, outputting control actions in millisecond timescales to satisfy strict latency constraints without solving NP-hard optimizations in real time.

    3. Analysis-Based Control: This paradigm decouples prediction from decision-making. A deep neural network is employed as an auxiliary analytical module to predict non-observable channel features, forecast traffic load, or identify traffic classes. These analytical inferences are subsequently passed to an external deterministic or heuristic network controller that executes the final resource allocation or routing policy.

  4. Knowl 4 — Model and Training Parallelism for Distributed Mobile Deep Learning

    model/method

    Deploying and training deep learning models across distributed mobile environments and edge-cloud hierarchies relies on two core parallelization schemes:

    1. Model Parallelism: The layers or functional sub-components of a single large neural network are split across different physical devices. In hierarchical mobile settings, early shallow layers that extract basic representations are deployed on resource-constrained mobile or IoT devices to perform rapid, local, low-latency coarse inference. If confidence is insufficient or deeper processing is required, intermediate activation tensors are forwarded to fog or cloud servers hosting the remaining deep layers for fine-grained inference.

    2. Training Parallelism: The training process is parallelized across multiple nodes, each holding local data partitions:

      • Asynchronous Stochastic Gradient Descent (Asynchronous SGD): Distributed worker nodes compute gradients over local mini-batches and asynchronously update a central parameter server without locking synchronization barriers, reducing idle waiting time across heterogeneous mobile hardware.
      • Federated Learning: Worker devices keep raw training datasets entirely on-device to preserve privacy. Each device trains a local copy of the model and transmits only gradient or weight updates to an aggregation server. The central server computes a global model update using secure aggregation protocols, which decrypts aggregated parameter updates only when a threshold number of participating devices contribute, preventing the recovery of individual device data.

  5. Knowl 5 — Techniques for Tailoring Deep Learning Models to Resource-Constrained Mobile Devices

    model/method

    To execute deep neural networks within the strict memory, computational, and energy bounds of mobile and embedded devices, several architectural and algorithmic compression techniques are employed:

    1. Efficient Operator Design: Replacing standard dense convolutions with depth-wise separable convolutions (which separate spatial filtering from channel mixing) and point-wise group convolutions with channel shuffling, reducing multiplication-and-addition operations and parameter counts while maintaining feature representability.

    2. Low-Rank Tensor Decomposition: Applying Tucker decomposition or singular value decomposition to high-dimensional weight tensors in fully connected and convolutional layers to approximate dense parameter matrices with factorized lower-rank matrices.

    3. Weight Quantization: Converting 32-bit floating-point parameters (extFP32 ext{FP32}) and activations into low-bit representations such as 8-bit integers (extINT8 ext{INT8}), 16-bit floats (extFP16 ext{FP16}), or ternary values {−1,0,+1}\{-1, 0, +1\}, thereby reducing memory bandwidth requirements and cache footprints.

    4. Structured and Unstructured Pruning: Removing redundant weights, channels, or entire filter layers that contribute negligibly to task accuracy, followed by sparse fine-tuning.

    5. Feature Map Caching and Region Reuse: For continuous mobile sensing and vision streams, caching intermediate layer activations of prior frames. By executing a lightweight region-matching operator on incoming data, static or repetitive spatial regions reuse cached feature maps rather than recomputing them across all CNN layers.

  6. Knowl 6 — Spatio-Temporal Mobile Traffic Modeling and Video/NLP Analogies

    model/method

    Mobile network traffic distributed over geographic cellular grids can be formulated through analogies to computer vision and natural language processing:

    • Spatial Domain as Video Frames: City-wide mobile traffic measurements aggregated into discrete spatial bins at time tt form a 2D matrix (traffic snapshot) analogous to an image frame. A temporal sequence of these snapshots {Xt−s,…,Xt}\{X_{t-s}, \dots, X_t\} forms a 3D tensor analogous to a video clip, making 2D/3D CNNs and ConvLSTMs well suited for spatial feature extraction and regional demand prediction.
    • Temporal Domain as Language Sequences: When observing a single base station cell over extended time periods, the scalar traffic series resembles a 1D sequence of tokens, allowing LSTMs and attention mechanisms to capture temporal dependencies.

    Mobile traffic data exhibits three domain-specific properties that distinguish it from visual and speech data:

    1. Spatial Smoothness: Adjacent spatial grid cells generally exhibit continuous variations in mobile traffic volume without the abrupt edge discontinuities common in natural optical images.
    2. Multi-Level Periodicity: Mobile traffic exhibits strong, predictable diurnal (24-hour) and weekly (7-day) cyclical patterns driven by human routine, which are absent in standard video pixels.
    3. Mobility-Driven Spatial Shift: As populations travel between residential, transit, and business districts throughout the day, traffic volume peaks physically migrate across neighboring and distant geographic cells over time.

  7. Knowl 7 — Geometric Deep Learning for Graph- and Point-Cloud-Structured Mobile Data

    model/method

    Mobile networking datasets often possess non-Euclidean geometric structures that are not naturally modeled by standard grid-based CNNs or 1D RNNs:

    1. Point Cloud Representations: Spatial distributions of mobile users, IoT devices, or base stations can be formulated as point clouds P={p1,p2,…,pN}P = \{p_1, p_2, \dots, p_N\}, where each point pi=(xi,yi,zi,fi)p_i = (x_i, y_i, z_i, f_i) contains spatial coordinates and associated feature vectors (e.g., traffic demand, device capability). Deep point-set architectures (e.g., PointNet and PointNet++) process these permutation-invariant point sets to perform user clustering, dynamic coverage optimization, and individual mobility trajectory tracking.

    2. Graph-Structured Networking Representations: Communication topologies consisting of heterogeneous entities (base stations, routers, gateways, mobile users) and wireless links are represented as directed graphs G=(V,E)G = (V, E), where vertices vi∈Vv_i \in V represent network nodes and edges eij∈Ee_{ij} \in E represent communication links or traffic flows. Graph Convolutional Networks (Graph CNNs) and Graph Neural Networks (GNNs) leverage neighborhood aggregation and message passing across graph vertices to solve distributed routing, predict base-station traffic demands under non-Euclidean topologies, and detect network-wide graph anomalies.

  8. Knowl 8 — Deep Lifelong Learning and Transfer Learning in Dynamic Mobile Environments

    model/method

    Mobile networking environments are non-stationary due to shifting human mobility, dynamic channel conditions, and evolving security threats. Deep learning adaptation relies on two mechanisms:

    1. Deep Lifelong Learning: Enables neural models to continuously incorporate new network behaviors without experiencing catastrophic forgetting of previously learned patterns. This is implemented via:

      • Dual-Memory Architectures: Employing a fast memory neural module updated instantaneously upon encountering new streaming samples, combined with a deep memory network that is updated only when data from an unseen statistical distribution accumulates past a defined threshold.
      • External Memory Augmentation: Coupling neural networks with differentiable read/write memory modules (e.g., Neural Turing Machines) to store and retrieve historical patterns dynamically.

    2. Deep Transfer Learning: Reuses representations learned on a source domain (e.g., historical traffic or known cyber attacks) to bootstrap learning on a target domain (e.g., a newly deployed cell site or emerging malware variants). Under extreme conditions with scarce target data, this enables:

      • One-Shot Learning: Rapidly classifying new patterns from a single or few reference samples using metric-based embedding matching.
      • Zero-Shot Learning: Classifying unobserved classes using semantic attribute descriptions and learned cross-modal transfer functions without any explicit target-class training instances.

  9. Knowl 9 — Deep Learning Driven Device-Based and Device-Free Localization

    model/method

    Wireless user positioning using deep learning is divided into two architectural paradigms:

    1. Device-Based Localization: The target user carries an active mobile terminal that receives or transmits wireless signals. Input representations—such as Channel State Information (CSI) amplitude and calibrated phase vectors, multi-carrier RSSI vectors, and ambient magnetic field measurements—are processed by deep architectures (including RBMs, stacked Auto-Encoders, CNNs, and LSTMs). The neural networks learn location fingerprints that are invariant to environmental noise and multipath fading, mapping raw radio metrics to 2D/3D physical spatial coordinates.

    2. Device-Free Localization: The target entity does not carry any dedicated receiver or transmitter. Positioning relies on specialized environmental transceivers that analyze variations in radio frequency (RF) signals, multipath reflections, and Channel Impulse Response (CIR) caused by human presence and physical movement. Deep learning models (e.g., 2D/3D CNNs, denoising Auto-Encoders, and reinforcement learning agents driven by Variational Auto-Encoders) infer the location, pose, and vital signs (such as respiration and sleep patterns) directly from raw RF reflection dynamics.

  10. Knowl 10 — End-to-End Communications as Deep Auto-Encoders and Signal Processing Optimization

    model/method

    Deep learning models can represent and optimize physical layer communication blocks:

    1. End-to-End Communications via Auto-Encoders: A physical communications system is formulated as a single deep auto-encoder. The transmitter consists of a multilayer feed-forward neural network that maps a discrete input message s∈{1,2,…,M}s \in \{1, 2, \dots, M\} (represented as a one-hot vector) to an encoded transmit signal x∈R2nx \in \mathbb{R}^{2n}, followed by an explicit normalization layer ensuring physical power constraints ∥x∥2≤n\|x\|^2 \le n. The transmission channel is modeled as a non-trainable stochastic layer (e.g., adding Gaussian noise or Rayleigh fading) producing received signal yy. The receiver is a decoder neural network with a softmax activation that outputs a probability vector p^∈RM\hat{p} \in \mathbb{R}^M over the possible transmitted messages, trained end-to-end using stochastic gradient descent without requiring handcrafted modulation or coding schemes.

    2. MIMO Detection and Channel Estimation: Deep neural networks (MLPs and CNNs) unfold iterative optimization algorithms (such as projected gradient descent and approximate message passing) to perform multi-user MIMO signal detection, nonlinear equalization, and direction-of-arrival estimation with near-optimal accuracy and lower computational runtime.

    3. Automatic Modulation Classification: Deep CNNs, LSTMs, and Radio Transformer Networks process raw in-phase and quadrature (I/Q) time-series samples to classify complex digital and analog modulation schemes, integrating learned spatial transformer modules to perform automatic phase, frequency, and timing synchronization.

  11. Knowl 11 — Deep Learning Optimization and Parallel Acceleration Ecosystem for Mobile Systems

    model/method

    Deploying deep learning in mobile networks is enabled by a hierarchical hardware, distributed systems, and software stack:

    Layer Representative Tools/Platforms Key Operational Function
    Parallel Hardware GPUs, Google TPU, Neural Processing Units (NPUs) Accelerated matrix multiplications and low-power parallel inference.
    Distributed Engines Gaia, Tux2, MLbase, GeePS, Ray Inter-node consistency, graph computation, and WAN communication mitigation.
    Deep Learning Libraries TensorFlow, PyTorch, Caffe2, MXNet Automatic differentiation, GPU acceleration, and cross-platform graph serving.
    Mobile Runtime Frameworks Core ML, TensorFlow Lite, ncnn, DeepSense Low-latency on-device inference using mobile CPUs/NPUs on Android and iOS.
    Optimization Algorithms Adam, RMSprop, Nadam, AdaGrad Adaptive parameter-wise learning rates for fast non-convex loss convergence.
    Gradient Compression TernGrad, AdaComp Quantizing gradients to {−1,0,1}\{-1, 0, 1\} to reduce distributed communication bandwidth.

  12. Knowl 12 — Deep Learning for Mobile Network Security and Privacy Preservation

    model/method

    Deep learning provides defense mechanisms across three tiers of mobile network security:

    1. Infrastructure-Level Intrusion and Anomaly Detection: Stacked and denoising Auto-Encoders learn baseline statistical representations of legitimate network traffic in an unsupervised manner. When anomalous events occur (e.g., IEEE 802.11 injection, flooding, impersonation attacks, distributed denial of service), they produce elevated reconstruction errors relative to the learned normal distribution, triggering automated alarms without hand-engineered signatures.

    2. Software-Level Malware and Botnet Classification: Deep neural networks (CNNs, DBNs, and LSTMs) analyze disassembled application bytecodes, Linux kernel system call graphs, opcode sequences, and API permission combinations from mobile application packages to detect obfuscated malware and categorize botnet command-and-control communication traffic across their lifecycles.

    3. Privacy-Preserving Mobile Analytics: Hybrid edge-cloud architectures partition neural networks such that feature extraction is executed locally on user devices while classification and aggregation occur in the cloud. Differentially private Stochastic Gradient Descent (DP-SGD) introduces calibrated Gaussian noise into calculated gradients, preventing the recovery or leakage of sensitive user data from shared neural representations.

Coverage note — Detailed literature summaries of specific individual empirical testbeds across the survey's 500+ references were consolidated into representative structural taxonomies, domain-specific paradigms, and tailoring methodologies to focus on the paper's overarching contributions.

References

  1. 1.Cisco. Cisco Visual Networking Index: Forecast and Methodology, 2016-2021, June 2017.
  2. 2.Ning Wang, Ekram Hossain, and Vijay K Bhargava. Backhauling 5G small cells: A radio resource management perspective. IEEE Wireless Communications, 22(5):41–49, 2015.
  3. 3.Fabio Giust, Luca Cominardi, and Carlos J Bernardos. Distributed mobility management for future 5G networks: overview and analysis of existing approaches. IEEE Communications Magazine, 53(1):142–149, 2015.
  4. 4.Mamta Agiwal, Abhishek Roy, and Navrati Saxena. Next generation 5G wireless networks: A comprehensive survey. IEEE Communications Surveys & Tutorials, 18(3):1617–1655, 2016.
  5. 5.Akhil Gupta and Rakesh Kumar Jha. A survey of 5G network: Architecture and emerging technologies. IEEE access, 3:1206–1232, 2015.
  6. 6.Kan Zheng, Zhe Yang, Kuan Zhang, Periklis Chatzimisios, Kan Yang, and Wei Xiang. Big data-driven optimization for mobile networks toward 5G. IEEE network, 30(1):44–51, 2016.
  7. 7.Chunxiao Jiang, Haijun Zhang, Yong Ren, Zhu Han, Kwang-Cheng Chen, and Lajos Hanzo. Machine learning paradigms for next-generation wireless networks. IEEE Wireless Communications, 24(2):98–105, 2017.
  8. 8.Duong D Nguyen, Hung X Nguyen, and Langford B White. Reinforcement learning with network-assisted feedback for heterogeneous rat selection. IEEE Transactions on Wireless Communications, 2017.
  9. 9.Fairuz Amalina Narudin, Ali Feizollah, Nor Badrul Anuar, and Abdullah Gani. Evaluation of machine learning classifiers for mobile malware detection. Soft Computing, 20(1):343–357, 2016.
  10. 10.Kevin Hsieh, Aaron Harlap, Nandita Vijaykumar, Dimitris Konomis, Gregory R Ganger, Phillip B Gibbons, and Onur Mutlu. Gaia: Geo-distributed machine learning approaching LAN speeds. In USENIX Symposium on Networked Systems Design and Implementation (NSDI), pages 629–647, 2017.
  11. 11.Wencong Xiao, Jilong Xue, Youshan Miao, Zhen Li, Cheng Chen, Ming Wu, Wei Li, and Lidong Zhou. Tux2: Distributed graph computation for machine learning. In USENIX Symposium on Networked Systems Design and Implementation (NSDI), pages 669–682, 2017.
  12. 12.Paolini, Monica and Fili, Senza . Mastering Analytics: How to benefit from big data and network complexity: An Analyst Report. RCR Wireless News, 2017.
  13. 13.Chaoyun Zhang, Pan Zhou, Chenghua Li, and Lijun Liu. A convolutional neural network for leaves recognition using data augmentation. In Proc. IEEE International Conference on Pervasive Intelligence and Computing (PICOM), pages 2143–2150, 2015.
  14. 14.Richard Socher, Yoshua Bengio, and Christopher D Manning. Deep learning for NLP (without magic). In Tutorial Abstracts of ACL 2012, pages 5–5. Association for Computational Linguistics.
  15. 15.IEEE Network special issue: Exploring Deep Learning for Efficient and Reliable Mobile Sensing. http://www.comsoc.org/netmag/cfp/exploring-deep-learning-efficient-and-reliable-mobile-sensing, 2017. [Online; accessed 14-July-2017].
  16. 16.Mowei Wang, Yong Cui, Xin Wang, Shihan Xiao, and Junchen Jiang. Machine learning for networking: Workflow, advances and opportunities. IEEE Network, 32(2):92–99, 2018.
  17. 17.Mohammad Abu Alsheikh, Dusit Niyato, Shaowei Lin, Hwee-Pink Tan, and Zhu Han. Mobile big data analytics using deep learning and Apache Spark. IEEE network, 30(3):22–29, 2016.
  18. 18.Ian Goodfellow, Yoshua Bengio, and Aaron Courville. Deep learning. MIT press, 2016.
  19. 19.Volodymyr Mnih, Koray Kavukcuoglu, David Silver, Andrei A. Rusu, Joel Veness, Marc G. Bellemare, Alex Graves, Martin Riedmiller, Andreas K. Fidjeland, Georg Ostrovski, Stig Petersen, Charles Beattie, Amir Sadik, Ioannis Antonoglou, Helen King, Dharshan Kumaran, Daan Wierstra, Shane Legg, and Demis Hassabis. Human-level control through deep reinforcement learning. Nature, 518(7540):529–533, 2015.
  20. 20.Yann LeCun, Yoshua Bengio, and Geoffrey Hinton. Deep learning. Nature, 521(7553):436–444, 2015.
  21. 21.Jürgen Schmidhuber. Deep learning in neural networks: An overview. Neural networks, 61:85–117, 2015.
  22. 22.Weibo Liu, Zidong Wang, Xiaohui Liu, Nianyin Zeng, Yurong Liu, and Fuad E Alsaadi. A survey of deep neural network architectures and their applications. Neurocomputing, 234:11–26, 2017.
  23. 23.Li Deng and Dong Yu. Deep learning: methods and applications. Foundations and Trends R in Signal Processing, 7(3–4):197–387, 2014.
  24. 24.Li Deng. A tutorial survey of architectures, algorithms, and applications for deep learning. APSIPA Transactions on Signal and Information Processing, 3, 2014.
  25. 25.Samira Pouyanfar, Saad Sadiq, Yilin Yan, Haiman Tian, Yudong Tao, Maria Presa Reyes, Mei-Ling Shyu, Shu-Ching Chen, and S. S. Iyengar. A Survey on Deep Learning: Algorithms, Techniques, and Applications. ACM Computing Surveys (CSUR), 51(5):92:1–92:36, 2018.
  26. 26.Kai Arulkumaran, Marc Peter Deisenroth, Miles Brundage, and Anil Anthony Bharath. Deep reinforcement learning: A brief survey. IEEE Signal Processing Magazine, 34(6):26–38, 2017.
  27. 27.Ahmed Hussein, Mohamed Medhat Gaber, Eyad Elyan, and Chrisina Jayne. Imitation learning: A survey of learning methods. ACM Computing Surveys (CSUR), 50(2):21:1–21:35, 2017.
  28. 28.Xue-Wen Chen and Xiaotong Lin. Big data deep learning: challenges and perspectives. IEEE access, 2:514–525, 2014.
  29. 29.Maryam M Najafabadi, Flavio Villanustre, Taghi M Khoshgoftaar, Naeem Seliya, Randall Wald, and Edin Muharemagic. Deep learning applications and challenges in big data analytics. Journal of Big Data, 2(1):1, 2015.
  30. 30.NF Hordri, A Samar, SS Yuhaniz, and SM Shamsuddin. A systematic literature review on features of deep learning in big data analytics. International Journal of Advances in Soft Computing & Its Applications, 9(1), 2017.
  31. 31.Mehdi Gheisari, Guojun Wang, and Md Zakirul Alam Bhuiyan. A survey on deep learning in big data. In Proc. IEEE International Conference on Computational Science and Engineering (CSE) and Embedded and Ubiquitous Computing (EUC), volume 2, pages 173–180, 2017.
  32. 32.Shuai Zhang, Lina Yao, Aixin Sun, and Yi Tay. Deep learning based recommender system: A survey and new perspectives. ACM Computing Surveys (CSUR), 52(1):5:1–5:38, 2019.
  33. 33.Shui Yu, Meng Liu, Wanchun Dou, Xiting Liu, and Sanming Zhou. Networking for big data: A survey. IEEE Communications Surveys & Tutorials, 19(1):531–549, 2017.
  34. 34.Mohammad Abu Alsheikh, Shaowei Lin, Dusit Niyato, and Hwee-Pink Tan. Machine learning in wireless sensor networks: Algorithms, strategies, and applications. IEEE Communications Surveys & Tutorials, 16(4):1996–2018, 2014.
  35. 35.Chun-Wei Tsai, Chin-Feng Lai, Ming-Chao Chiang, and Laurence T Yang. Data mining for Internet of things: A survey. IEEE Communications Surveys and Tutorials, 16(1):77–97, 2014.
  36. 36.Xiang Cheng, Luoyang Fang, Xuemin Hong, and Liuqing Yang. Exploiting mobile big data: Sources, features, and applications. IEEE Network, 31(1):72–79, 2017.
  37. 37.Mario Bkassiny, Yang Li, and Sudharman K Jayaweera. A survey on machine-learning techniques in cognitive radios. IEEE Communications Surveys & Tutorials, 15(3):1136–1159, 2013.
  38. 38.Jeffrey G Andrews, Stefano Buzzi, Wan Choi, Stephen V Hanly, Angel Lozano, Anthony CK Soong, and Jianzhong Charlie Zhang. What will 5G be? IEEE Journal on selected areas in communications, 32(6):1065–1082, 2014.
  39. 39.Nisha Panwar, Shantanu Sharma, and Awadhesh Kumar Singh. A survey on 5G: The next generation of mobile communication. Physical Communication, 18:64–84, 2016.
  40. 40.Olakunle Elijah, Chee Yen Leow, Tharek Abdul Rahman, Solomon Nunoo, and Solomon Zakwoi Iliya. A comprehensive survey of pilot contamination in massive MIMO–5G system. IEEE Communications Surveys & Tutorials, 18(2):905–923, 2016.
  41. 41.Stefano Buzzi, I Chih-Lin, Thierry E Klein, H Vincent Poor, Chenyang Yang, and Alessio Zappone. A survey of energy-efficient techniques for 5G networks and challenges ahead. IEEE Journal on Selected Areas in Communications, 34(4):697–709, 2016.
  42. 42.Mugen Peng, Yong Li, Zhongyuan Zhao, and Chonggang Wang. System architecture and key technologies for 5G heterogeneous cloud radio access networks. IEEE network, 29(2):6–14, 2015.
  43. 43.Yong Niu, Yong Li, Depeng Jin, Li Su, and Athanasios V Vasilakos. A survey of millimeter wave communications (mmwave) for 5G: opportunities and challenges. Wireless Networks, 21(8):2657–2676, 2015.
  44. 44.Xenofon Foukas, Georgios Patounas, Ahmed Elmokashfi, and Mahesh K Marina. Network slicing in 5G: Survey and challenges. IEEE Communications Magazine, 55(5):94–100, 2017.
  45. 45.Tarik Taleb, Konstantinos Samdanis, Badr Mada, Hannu Flinck, Sunny Dutta, and Dario Sabella. On multi-access edge computing: A survey of the emerging 5G network edge architecture & orchestration. IEEE Communications Surveys & Tutorials, 2017.
  46. 46.Pavel Mach and Zdenek Becvar. Mobile edge computing: A survey on architecture and computation offloading. IEEE Communications Surveys & Tutorials, 2017.
  47. 47.Yuyi Mao, Changsheng You, Jun Zhang, Kaibin Huang, and Khaled B Letaief. A survey on mobile edge computing: The communication perspective. IEEE Communications Surveys & Tutorials, 2017.
  48. 48.Ying Wang, Peilong Li, Lei Jiao, Zhou Su, Nan Cheng, Xuemin Sherman Shen, and Ping Zhang. A data-driven architecture for personalized QoE management in 5G wireless networks. IEEE Wireless Communications, 24(1):102–110, 2017.
  49. 49.Qilong Han, Shuang Liang, and Hongli Zhang. Mobile cloud sensing, big data, and 5G networks make an intelligent and smart world. IEEE Network, 29(2):40–45, 2015.
  50. 50.Sukhdeep Singh, Navrati Saxena, Abhishek Roy, and HanSeok Kim. A survey on 5G network technologies from social perspective. IETE Technical Review, 34(1):30–39, 2017.
  51. 51.Min Chen, Jun Yang, Yixue Hao, Shiwen Mao, and Kai Hwang. A 5G cognitive system for healthcare. Big Data and Cognitive Computing, 1(1):2, 2017.
  52. 52.Xianfu Chen, Jinsong Wu, Yueming Cai, Honggang Zhang, and Tao Chen. Energy-efficiency oriented traffic offloading in wireless networks: A brief survey and a learning approach for heterogeneous cellular networks. IEEE Journal on Selected Areas in Communications, 33(4):627–640, 2015.
  53. 53.Jinsong Wu, Song Guo, Jie Li, and Deze Zeng. Big data meet green challenges: big data toward green applications. IEEE Systems Journal, 10(3):888–900, 2016.
  54. 54.Teodora Sandra Buda, Haytham Assem, Lei Xu, Danny Raz, Udi Margolin, Elisha Rosensweig, Diego R Lopez, Marius-Iulian Corici, Mikhail Smirnov, Robert T Mullins, O. Uryupina, A. Mozo, B. Ordozgoiti, A. Martin, A. Alloush, P. O’Sullivan, and I. G. Ben Yahia. Can machine learning aid in delivering new use cases and scenarios in 5G? In IEEE/IFIP Network Operations and Management Symposium (NOMS), pages 1279–1284, 2016.
  55. 55.Ali Imran, Ahmed Zoha, and Adnan Abu-Dayya. Challenges in 5G: how to empower SON with big data for enabling 5G. IEEE Network, 28(6):27–33, 2014.
  56. 56.Bharath Keshavamurthy and Mohammad Ashraf. Conceptual design of proactive SONs based on the big data framework for 5G cellular networks: A novel machine learning perspective facilitating a shift in the son paradigm. In Proc. IEEE International Conference on System Modeling & Advancement in Research Trends (SMART), pages 298–304, 2016.
  57. 57.Paulo Valente Klaine, Muhammad Ali Imran, Oluwakayode Onireti, and Richard Demo Souza. A survey of machine learning techniques applied to self organizing cellular networks. IEEE Communications Surveys and Tutorials, 2017.
  58. 58.Rongpeng Li, Zhifeng Zhao, Xuan Zhou, Guoru Ding, Yan Chen, Zhongyao Wang, and Honggang Zhang. Intelligent 5G: When cellular networks meet artificial intelligence. IEEE Wireless communications, 24(5):175–183, 2017.
  59. 59.Nicola Bui, Matteo Cesana, S Amir Hosseini, Qi Liao, Ilaria Malanchini, and Joerg Widmer. A survey of anticipatory mobile networking: Context-based classification, prediction methodologies, and optimization techniques. IEEE Communications Surveys & Tutorials, 19(3):1790–1821, 2017.
  60. 60.Rachad Atat, Lingjia Liu, Jinsong Wu, Guangyu Li, Chunxuan Ye, and Yi Yang. Big data meet cyber-physical systems: A panoramic survey. IEEE Access, 6:73603–73636, 2018.
  61. 61.Xiang Cheng, Luoyang Fang, Liuqing Yang, and Shuguang Cui. Mobile big data: The fuel for data-driven wireless. IEEE Internet of Things Journal, 4(5):1489–1516, 2017.
  62. 62.Panagiotis Kasnesis, Charalampos Patrikakis, and Iakovos Venieris. Changing the game of mobile data analysis with deep learning. IT Professional, 2017.
  63. 63.Xiang Cheng, Luoyang Fang, Liuqing Yang, and Shuguang Cui. Mobile big data: the fuel for data-driven wireless. IEEE Internet of Things Journal, 4(5):1489–1516, 2017.
  64. 64.Lidong Wang and Randy Jones. Big data analytics for network intrusion detection: A survey. International Journal of Networks and Communications, 7(1):24–31, 2017.
  65. 65.Nei Kato, Zubair Md Fadlullah, Bomin Mao, Fengxiao Tang, Osamu Akashi, Takeru Inoue, and Kimihiro Mizutani. The deep learning vision for heterogeneous network traffic control: proposal, challenges, and future perspective. IEEE Wireless Communications, 24(3):146–153, 2017.
  66. 66.Michele Zorzi, Andrea Zanella, Alberto Testolin, Michele De Filippo De Grazia, and Marco Zorzi. Cognition-based networks: A new perspective on network optimization using learning and distributed intelligence. IEEE Access, 3:1512–1530, 2015.
  67. 67.Zubair Fadlullah, Fengxiao Tang, Bomin Mao, Nei Kato, Osamu Akashi, Takeru Inoue, and Kimihiro Mizutani. State-of-the-art deep learning: Evolving machine intelligence toward tomorrow’s intelligent network traffic control systems. IEEE Communications Surveys & Tutorials, 19(4):2432–2455, 2017.
  68. 68.Mehdi Mohammadi, Ala Al-Fuqaha, Sameh Sorour, and Mohsen Guizani. Deep Learning for IoT Big Data and Streaming Analytics: A Survey. IEEE Communications Surveys & Tutorials, 2018.
  69. 69.Nauman Ahad, Junaid Qadir, and Nasir Ahsan. Neural networks in wireless networks: Techniques, applications and guidelines. Journal of Network and Computer Applications, 68:1–27, 2016.
  70. 70.Qian Mao, Fei Hu, and Qi Hao. Deep learning for intelligent wireless networks: A comprehensive survey. IEEE Communications Surveys & Tutorials, 2018.
  71. 71.Nguyen Cong Luong, Dinh Thai Hoang, Shimin Gong, Dusit Niyato, Ping Wang, Ying-Chang Liang, and Dong In Kim. Applications of Deep Reinforcement Learning in Communications and Networking: A Survey. arXiv preprint arXiv:1810.07862, 2018.
  72. 72.Xiangwei Zhou, Mingxuan Sun, Ye Geoffrey Li, and Biing-Hwang Juang. Intelligent wireless communications enabled by cognitive radio and machine learning. China Communications, 15(12):16–48, 2018.
  73. 73.Mingzhe Chen, Ursula Challita, Walid Saad, Changchuan Yin, and Mérouane Debbah. Machine learning for wireless networks with artificial intelligence: A tutorial on neural networks. arXiv preprint arXiv:1710.02913, 2017.
  74. 74.Ammar Gharaibeh, Mohammad A Salahuddin, Sayed Jahed Hussini, Abdallah Khreishah, Issa Khalil, Mohsen Guizani, and Ala Al-Fuqaha. Smart cities: A survey on data management, security, and enabling technologies. IEEE Communications Surveys & Tutorials, 19(4):2456–2501, 2017.
  75. 75.Nicholas D Lane and Petko Georgiev. Can deep learning revolutionize mobile sensing? In Proc. 16th ACM International Workshop on Mobile Computing Systems and Applications, pages 117–122, 2015.
  76. 76.Kaoru Ota, Minh Son Dao, Vasileios Mezaris, and Francesco GB De Natale. Deep learning for mobile multimedia: A survey. ACM Transactions on Multimedia Computing, Communications, and Applications (TOMM), 13(3s):34, 2017.
  77. 77.Preeti Mishra, Vijay Varadharajan, Uday Tupakula, and Emmanuel S Pilli. A detailed investigation and analysis of using machine learning techniques for intrusion detection. IEEE Communications Surveys & Tutorials, 2018.
  78. 78.Yuxi Li. Deep reinforcement learning: An overview. arXiv preprint arXiv:1701.07274, 2017.

Citation

MLA
Zhang, C., et al. “Deep Learning in Mobile and Wireless Networking: A Survey”. arXiv, 2018, http://arxiv.org/abs/1803.04311v3.
APA
Zhang, C., Patras, P., & Haddadi, H. (2018). Deep Learning in Mobile and Wireless Networking: A Survey. arXiv. http://arxiv.org/abs/1803.04311v3
Chicago
Zhang, C., P. Patras, and H. Haddadi. 2018. “Deep Learning in Mobile and Wireless Networking: A Survey”. arXiv. http://arxiv.org/abs/1803.04311v3.
Harvard
Zhang, C., Patras, P. and Haddadi, H. (2018) “Deep Learning in Mobile and Wireless Networking: A Survey”, arXiv [Preprint]. Available at: http://arxiv.org/abs/1803.04311v3.
Vancouver
1. Zhang C, Patras P, Haddadi H (2018) Deep Learning in Mobile and Wireless Networking: A Survey. arXiv

BibTeX

@article{zhang2018deep,
  title = {Deep Learning in Mobile and Wireless Networking: A Survey},
  author = {Zhang, Chaoyun and Patras, Paul and Haddadi, Hamed},
  year = {2018},
  journal = {arXiv},
  url = {http://arxiv.org/abs/1803.04311v3},
  eprint = {1803.04311}
}
Metadata:arXiv

Access the Paper

This paper is available from its original source. Click below to access the PDF.

Open PDF