Lumos5G

Narayanan, Arvind; Ramadan, Eman; Mehta, Rishabh; Hu, Xiaodong; Liu, Qingxu; Fezeu, Rostand A. K.; Dayalan, Udhaya Kumar; Verma, Saurabh; Ji, Peiqi; Li, Tao; Qian, Feng; Zhang, Zhili

doi:10.1145/3419394.3423629

Cited by 107 publications

(20 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, the 5G technology remains within the development phase, and, thus, the availability of commercially deployed 5G networks is limited, particularly in the middle and low-economy countries. Although some existing works present real network 5G mmWave datasets [3], [11], [12], these have been collected using one or two smartphones, i.e., users. As the user experienced datarate is a function of the network load, i.e., the number of users associated with a BS, hence, it is significant to test the efficacy of any new algorithm for different network loads.…”

Section: Pilot Studymentioning

confidence: 99%

“…In transfer learning, the knowledge gained from previously learned tasks is applied to a target domain of related tasks. Interestingly, the publicly available 5G datasets [3], [11], [12] provide enough throughput traces that can be used as a source domain for the transfer learning and can be applied to a target 5G domain with a simulated dataset. To the best of our knowledge, this is the first work that uses transfer learning to predict 5G throughput using simulated traces.…”

Section: ) Existing Work On Throughput Prediction and Theirmentioning

confidence: 99%

“…• Lumos-5G dataset [11] -The Lumos-5G dataset from [11] contains a measurement study of commercial 5G mmWave services in Minneapolis, USA, and has recorded the downlink throughput as perceived by applications running on UE. It contains information on the UE's geographical coordinates, moving speed, compass directions, downlink throughput (reported using iperf 3.7), radio type (4G/5G), signal strength (LTE -RSRP, RSRQ, RSSI & 5G -SSRSRP, SSRSRQ, SSRSSI), handover events, etc.…”

Section: Implementation and Training Of ML Modelsmentioning

confidence: 99%

“…We train the LSTM model of the throughput decision engine with the public-5G datasets [3], [11], [12] with a trainvalidation split of 80%-20%. The trained model is then retrained by means of Transfer Learning using the ns-3 simulated 5G dataset.…”

Section: Implementation and Training Of ML Modelsmentioning

confidence: 99%

“…Transfer learning is used to compensate for the limited availability of the 5G network throughput traces. For this, we first train a recurrent neural network with real-world 5G datasets [3], [11], [12] and then retrain the learned model with data collected from extensive simulations of a 5G mmWave network. The simulations help us capture every little nuance associated with the functioning of the 5G mmWave network while also allowing us to vary the number of active users in the network and observe the corresponding performance over the video streaming application.…”

mentioning

confidence: 99%

See 4 more Smart Citations

Improving UE Energy Efficiency Through Network-Aware Video Streaming Over 5G

Palit

Sen

Mondal³

et al. 2023

IEEE Trans. Netw. Serv. Manage.

View full text Add to dashboard Cite

Adaptive Bitrate (ABR) Streaming over the cellular networks has been well studied in the literature; however, existing ABR algorithms primarily focus on improving the end-users' Quality of Experience (QoE) while ignoring the resource consumption aspect of the underlying device. Consequently, proactive attempts to download video data to maintain the user's QoE often impact the battery life of the underlying device unless the download attempts are synchronized with the network's channel condition. In this work, we develop EnDASH-5G -a wrapper over the popular DASH-based ABR streaming algorithm, which establishes this synchronization by utilizing a network-aware video data download mechanism. EnDASH-5G utilizes a novel throughput prediction mechanism for 5G mmWave networks by upgrading the existing throughput prediction models with a transfer learning-based approach leveraging publicly available 5G datasets. It then exploits deep reinforcement learning to dynamically decide the playback buffer length and the video bitrate using the predicted throughput. This ensures that the data download attempts get synchronized with the underlying network condition, thus saving the device's battery power. From a thorough evaluation of EnDASH-5G, we observe that it achieves a near 30.5% decrease in the maximum energy consumption than the state-of-the-art Pensieve ABR algorithm while performing almost at par in QoE.

show abstract

Section: Pilot Studymentioning

confidence: 99%