Distributed Heteromodal Split Learning for Vision Aided mmWave Received Power Prediction

The goal of this work is the accurate prediction of millimeter-wave received power leveraging both radio frequency (RF) signals and heterogeneous visual data from multiple distributed cameras, in a communication and energy-efficient manner while preserving data privacy. To this end, firstly focusing on data privacy, we propose heteromodal split learning with feature aggregation (HetSLAgg) that splits neural network (NN) models into camera-side and base station (BS)-side segments. The BS-side NN segment fuses RF signals and uploaded image features without collecting raw images. However, the usage of multiple visual data leads to an increase in NN input dimensions, which gives rise to additional communication and energy costs. To overcome additional communication and energy costs due to image interpolation to blend different frame rates, we propose a novel BS-side manifold mixup technique that offloads the interpolation operations from cameras to a BS. Subsequently, we confront energy costs for operating a larger size of the BS- side NN segment due to concatenating image features across cameras and propose an energy-efficient aggregation method. This is done via a linear combination of image features instead of concatenating them, where the NN size is independent of the number of cameras. Comprehensive test-bed experiments with measured channels demonstrate that HetSLAgg reduces the prediction error by 44% compared to a baseline leveraging only RF received power. Moreover, the experiments show that the designed HetSLAgg achieves over 20% gains in terms of communication and energy cost reduction compared to several baseline designs within at most 1% of accuracy loss.

[1]  Ramesh Raskar,et al.  Distributed learning of deep neural network over multiple agents , 2018, J. Netw. Comput. Appl..

[2]  Branka Vucetic,et al.  Managing Vertical Handovers in Millimeter Wave Heterogeneous Networks , 2019, IEEE Transactions on Communications.

[3]  Masahiro Morikura,et al.  Proactive Handover Based on Human Blockage Prediction Using RGB-D Cameras for mmWave Communications , 2016, IEICE Trans. Commun..

[4]  Husheng Li,et al.  Motion Aware Beam Tracking in Mobile Millimeter Wave Communications: A Data-Driven Approach , 2019, ICC 2019 - 2019 IEEE International Conference on Communications (ICC).

[5]  Guigang Zhang,et al.  Deep Learning , 2016, Int. J. Semantic Comput..

[6]  Masahiro Morikura,et al.  Reinforcement learning based predictive handover for pedestrian-aware mmWave networks , 2018, IEEE INFOCOM 2018 - IEEE Conference on Computer Communications Workshops (INFOCOM WKSHPS).

[7]  Chun-hui Zhao,et al.  Multi-Camera-Based Object Handoff Using Decision-Level Fusion , 2009, 2009 2nd International Congress on Image and Signal Processing.

[8]  Thi-Lan Le,et al.  Fusion of wifi and visual signals for person tracking , 2016, SoICT.

[9]  Song Han,et al.  Learning both Weights and Connections for Efficient Neural Network , 2015, NIPS.

[10]  Ossi Kaltiokallio,et al.  A Three-State Received Signal Strength Model for Device-Free Localization , 2014, IEEE Transactions on Vehicular Technology.

[11]  Ramesh Raskar,et al.  Split learning for health: Distributed deep learning without sharing raw patient data , 2018, ArXiv.

[12]  Supun Samarasekera,et al.  Multi-modal sensor fusion algorithm for ubiquitous infrastructure-free localization in vision-impaired environments , 2010, 2010 IEEE/RSJ International Conference on Intelligent Robots and Systems.

[13]  Masahiro Morikura,et al.  Handover Management for mmWave Networks With Proactive Performance Prediction Using Camera Images and Deep Reinforcement Learning , 2019, IEEE Transactions on Cognitive Communications and Networking.

[14]  Mehdi Bennis,et al.  Wireless Network Intelligence at the Edge , 2018, Proceedings of the IEEE.

[15]  Xiongwen Zhao,et al.  Millimeter-Wave Propagation Channel Characterization for Short-Range Wireless Communications , 2009, IEEE Transactions on Vehicular Technology.

[16]  Masahiro Morikura,et al.  Cooperative Sensing in Deep RL-Based Image-to-Decision Proactive Handover for mmWave Networks , 2020, 2020 IEEE 17th Annual Consumer Communications & Networking Conference (CCNC).

[17]  Sofie Pollin,et al.  Database-Assisted Spectrum Prediction in 5G Networks and Beyond: A Review and Future Challenges , 2019, IEEE Circuits and Systems Magazine.

[18]  William J. Dally,et al.  Deep Gradient Compression: Reducing the Communication Bandwidth for Distributed Training , 2017, ICLR.

[19]  Peter Richtárik,et al.  Federated Learning: Strategies for Improving Communication Efficiency , 2016, ArXiv.

[20]  Mehdi Bennis,et al.  Communication-Efficient On-Device Machine Learning: Federated Distillation and Augmentation under Non-IID Private Data , 2018, ArXiv.

[21]  Kiyoharu Aizawa,et al.  Tracking Persons using Particle Filter Fusing Visual and Wi-Fi Localizations for Widely Distributed Camera , 2007, 2007 IEEE International Conference on Image Processing.

[22]  Ioannis Mitliagkas,et al.  Manifold Mixup: Better Representations by Interpolating Hidden States , 2018, ICML.

[23]  Guna Seetharaman,et al.  Heterogeneous Multi-View Information Fusion: Review of 3-D Reconstruction Methods and a New Registration with Uncertainty Modeling , 2016, IEEE Access.

[24]  Fei-Fei Li,et al.  RGB-W: When Vision Meets Wireless , 2015, 2015 IEEE International Conference on Computer Vision (ICCV).

[25]  Theodore S. Rappaport,et al.  A Flexible Millimeter-Wave Channel Sounder With Absolute Timing , 2017, IEEE Journal on Selected Areas in Communications.

[26]  Jaime Lloret,et al.  An Intelligent handover process algorithm in 5G networks: The use case of mobile cameras for environmental surveillance , 2017, 2017 IEEE International Conference on Communications Workshops (ICC Workshops).

[27]  Masahiro Morikura,et al.  Measurement Method of Temporal Attenuation by Human Body in Off-the-Shelf 60 GHz WLAN with HMM-Based Transmission State Estimation , 2018, Wirel. Commun. Mob. Comput..

[28]  Edward W. Knightly,et al.  IEEE 802.11ay: Next-Generation 60 GHz Communication for 100 Gb/s Wi-Fi , 2017, IEEE Communications Magazine.

[29]  Sander Oude Elberink,et al.  Accuracy and Resolution of Kinect Depth Data for Indoor Mapping Applications , 2012, Sensors.

[30]  Anit Kumar Sahu,et al.  Federated Optimization in Heterogeneous Networks , 2018, MLSys.

[31]  King Ngi Ngan,et al.  Segmentation and Tracking Multiple Objects Under Occlusion From Multiview Video , 2011, IEEE Transactions on Image Processing.

[32]  Sundeep Rangan,et al.  An MDP model for optimal handover decisions in mmWave cellular networks , 2015, 2016 European Conference on Networks and Communications (EuCNC).

[33]  Sanjiv Kumar,et al.  cpSGD: Communication-efficient and differentially-private distributed SGD , 2018, NeurIPS.

[34]  Masahiro Morikura,et al.  Communication-Efficient Multimodal Split Learning for mmWave Received Power Prediction , 2020, IEEE Communications Letters.

[35]  Furong Wang,et al.  A trajectory-aware handoff algorithm based on GPS information , 2010, Ann. des Télécommunications.

[36]  Lei Chen,et al.  Revolution of Self-Organizing Network for 5G MmWave Small Cell Management: From Reactive to Proactive , 2018, IEEE Wireless Communications.

[37]  Dit-Yan Yeung,et al.  Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcasting , 2015, NIPS.

[38]  G. E. Zein,et al.  Influence of the human activity on wide-band characteristics of the 60 GHz indoor radio channel , 2004, IEEE Transactions on Wireless Communications.

[39]  Mehdi Bennis,et al.  GADMM: Fast and Communication Efficient Framework for Distributed Machine Learning , 2019, J. Mach. Learn. Res..

[40]  Blaise Agüera y Arcas,et al.  Communication-Efficient Learning of Deep Networks from Decentralized Data , 2016, AISTATS.

[41]  Hongyi Zhang,et al.  mixup: Beyond Empirical Risk Minimization , 2017, ICLR.

[42]  Takayuki Nishio,et al.  Extreme URLLC: Vision, Challenges, and Key Enablers , 2020, ArXiv.

[43]  Richard Nock,et al.  Advances and Open Problems in Federated Learning , 2019, Found. Trends Mach. Learn..

[44]  Sergey Ioffe,et al.  Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift , 2015, ICML.

[45]  Masahiro Morikura,et al.  Proactive Received Power Prediction Using Machine Learning and Depth Images for mmWave Networks , 2018, IEEE Journal on Selected Areas in Communications.

[46]  Geoffrey E. Hinton,et al.  On the importance of initialization and momentum in deep learning , 2013, ICML.