Applying Deep-Learning-Based Computer Vision to Wireless Communications: Methodologies, Opportunities, and Challenges

Tian, Yu; Pan, Gaofeng; Alouini, Mohamed‐Slim

doi:10.48550/arxiv.2006.05782

Cited by 2 publications

(4 citation statements)

References 7 publications

(12 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This is usually caused by either over-fitting or more likely, in the competition, the different data distribution of unseen test dataset from the training and validation datasets. The most recent work on resolving this competition task was shown by Tian et al in [6] where a new deep learning model based on ResNet and ResNeXt was proposed to embed the image. Leveraging the power of deeper models, they achieved significantly better scores than those in literature.…”

Section: Introductionmentioning

confidence: 99%

“…In practice, after a model is trained, the images in the validation dataset should not be seen before even they are taken from the same environment. Thus, for all the prior results in [3], [5], [6], whatever the performance scores have been achieved, it is not clear if the model performs well in validation because it is able to better learn image features or it just memorizes the same image in training. To get rid of the image repetition in both training and validation, we reformulate a new dataset where the images in training and validation are mutually exclusive.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Vision-Aided Beam Tracking: Explore the Proper Use of Camera Images with Deep Learning

Tian¹,

Wang²

2021

Preprint

Self Cite

View full text Add to dashboard Cite

We investigate the problem of wireless beam tracking on mmWave bands with the assistance of camera images. In particular, based on the user's beam indices used and camera images taken in the trajectory, we predict the optimal beam indices in the next few time spots. To resolve this problem, we first reformulate the "ViWi" dataset in [1] to get rid of the image repetition problem. Then we develop a deep learning approach and investigate various model components to achieve the best performance. Finally, we explore whether, when, and how to use the image for better beam prediction. To answer this question, we split the dataset into three clusters -(LOS, light NLOS, serious NLOS)-like -based on the standard deviation of the beam sequence. With experiments we demonstrate that using the image indeed helps beam tracking especially when the user is in serious NLOS, and the solution relies on carefully-designed dataset for training a model. Generally speaking, including NLOS-like data for training a model does not benefit beam tracking of the user in LOS, but including light NLOS-like data for training a model benefits beam tracking of the user in serious NLOS.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Vision-Aided Beam Tracking: Explore the Proper Use of Camera Images with Deep Learning

Tian¹,

Wang²

2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…The major objective of this novel taxonomy is to recognize the visual data aided HO management schemes-which has been long overlooked in the literature-by giving it a special place along with the traditional wireless data driven HO schemes. The visual aided wireless communications is an emerging research area in wireless communications where visual information (pictures/videos) captured from cameras, light detection and ranging (LIDAR), etc., are combined with wireless sensory data for wireless network optimization such as channel prediction, HO optimization, etc (Nishio et al, 2020;Tian et al, 2020). This is necessary because mm-wave communication networks possess unique challenges that would be difficult to handle using only wireless sensory data but with the assistance of visual data, some of these challenges can be handled properly.…”

Section: Machine Learning For Handover Managementmentioning

confidence: 99%

“…However, the use of mm-wave and THz frequencies in 5G and B5G networks would mean that BSs will have many antennas, communication will be through a large number of LOS beams, which would be subject to blockages of various types and would limit signal reception at the user end. In addition, much signaling overhead would be involved in the selection of the optimal beam for user connection in mm-wave networks if only wireless sensory data are exploited for optimal beam selection considering the massive number of beams that would be involved (Nishio et al, 2020;Tian et al, 2020).…”

Section: (I) Visual Data Aided Handover Optimizationmentioning

confidence: 99%

Improved handover decision scheme for 5g mm-wave communication: optimum base station selection using machine learning approach.

Mollel¹

View full text Add to dashboard Cite

The rapid growth in mobile and wireless devices has led to an exponential demand for data traf fic and exacerbated the burden on conventional wireless networks. Fifth generation (5G) and beyond networks are expected to not only accommodate this growth in data demand but also provide additional services beyond the capability of existing wireless networks, while main taining a high quality-of-experience (QoE) for users. The need for several orders of magnitude increase in system capacity has necessitated the use of millimetre wave (mm-wave) frequencies as well as the proliferation of low-power small cells overlaying the existing macro-cell layer. These approaches offer a potential increase in throughput in magnitudes of several gigabits per second and a reduction in transmission latency, but they also present new challenges. For exam ple, mm-wave frequencies have higher propagation losses and a limited coverage area, thereby escalating mobility challenges such as more frequent handovers (HOs). In addition, the ad vent of low-power small cells with smaller footprints also causes signal fluctuations across the network, resulting in repeated HOs (ping-pong) from one small cell (SC) to another. Therefore, efficient HO management is very critical in future cellular networks since frequent HOs pose multiple threats to the quality-of-service (QoS), such as a reduction in the system throughput as well as service interruptions, which results in a poor QoE for the user. How ever, HO management is a significant challenge in 5G networks due to the use of mm-wave frequencies which have much smaller footprints. To address these challenges, this work in vestigates the HO performance of 5G mm-wave networks and proposes a novel method for achieving seamless user mobility in dense networks. The proposed model is based on a double deep reinforcement learning (DDRL) algorithm. To test the performance of the model, a com parative study was made between the proposed approach and benchmark solutions, including a benchmark developed as part of this thesis. The evaluation metrics considered include system throughput, execution time, ping-pong, and the scalability of the solutions. The results reveal that the developed DDRL-based solution vastly outperforms not only conventional methods but also other machine-learning-based benchmark techniques. The main contribution of this thesis is to provide an intelligent framework for mobility man agement in the connected state (i.e HO management) in 5G. Though primarily developed for mm-wave links between UEs and BSs in ultra-dense heterogeneous networks (UDHNs), the proposed framework can also be applied to sub-6 GHz frequencies.

show abstract

Applying Deep-Learning-Based Computer Vision to Wireless Communications: Methodologies, Opportunities, and Challenges

Cited by 2 publications

References 7 publications

Vision-Aided Beam Tracking: Explore the Proper Use of Camera Images with Deep Learning

Vision-Aided Beam Tracking: Explore the Proper Use of Camera Images with Deep Learning

Improved handover decision scheme for 5g mm-wave communication: optimum base station selection using machine learning approach.

Contact Info

Product

Resources

About