Trade-Offs for Video-Providers in LTE Networks: Smartphone Energy Consumption Vs Wasted Traffic

14
www3.informatik.uni-wuerzburg.de Institute of Computer Science Chair of Communication Networks Prof. Dr.-Ing. P. Tran-Gia Trade-Offs for Video-Providers in LTE Networks: Smartphone Energy Consumption Vs Wasted Traffic Christian Schwartz, Marc Scheib, Tobias Hoßfeld, Phuoc Tran-Gia, Jose Manuel Gimenez-Guzman

Transcript of Trade-Offs for Video-Providers in LTE Networks: Smartphone Energy Consumption Vs Wasted Traffic

www3.informatik.uni-wuerzburg.de

Institute of Computer ScienceChair of Communication Networks

Prof. Dr.-Ing. P. Tran-Gia

Trade-Offs for Video-Providers in LTE Networks:

Smartphone Energy Consumption Vs Wasted Traffic

Christian Schwartz, Marc Scheib, Tobias Hoßfeld, Phuoc Tran-Gia, Jose Manuel Gimenez-Guzman

Christian Schwartz2

2

Motivation

Video transmission is a mayor use case for LTE, and LTE deployments are going to be increased in the next years.

Video platform operators are interested in providing a good Quality of Experience to the user reducing the amount of traffic wasted if a user aborts watching a

video before completion.

[1] Cisco Virtual Networking Index: Global Mobile Data Traffic Forecast Update, 2012-2017

Christian Schwartz3

3

Video and Network Model

Christian Schwartz4

4

Assumptions

High QoE is the most important factor for the video platform operator:

Only mechanisms where no stalling occurs will be considered in the evaluation.

Constant video bitrate and constant bandwidth:

Not the most realistic assumptions, however useful to obtain a baseline for more realistic scenarios

Christian Schwartz5

5

Video Model

Download: Get complete Video with maximal bandwidth bW

Live: Video can only be downloaded with video bitrate bR

Provisioning: Download with 1.2 bR in order to prevent stalling

Streaming: Download until unplayed video length is larger θ + Θ, resume if video length is smaller than θ

Christian Schwartz6

6

Mobile Network Model

Network state depends on state machine, transitions depending on (in)activity.

RRC Idle state: No connection, but low energy consumption

RRC Connected state: Connected to network, higher energy consumption

State Continious Reception Short DRX Long DRX Idle DRX

Active Time 100 ms 1 ms 1 ms 43 ms

Inactive Time 0 ms 19 ms 39 ms 1.28 s

11.57 s

Christian Schwartz7

7

Energy Consumption

Energy consumption [2] depends on the currently used bandwidth

the current network state

[2] J. Huang, F. Qian, A. Gerber, Z. M. Mao, S. Sen, and O. Spatscheck, “A Close Examination of Performance and Power Characteristics of 4G LTE Networks,“

Christian Schwartz8

8

Wasted Traffic

Wasted traffic is the amount of data already downloaded in advance if a user stops watching.

User behaviour is modelled as a random variable A, giving the time of the abort.

Then, the wasted traffic is given as

Christian Schwartz9

9

Numerical Evaluation

Christian Schwartz10

10

Considered Scenario

We consider a video of 1600 seconds length.

We consider a bandwidth of 12.74 Mbps, the median bandwidth in current LTE networks. [2]

While video bit rates between 1 and 50 Mbit/s are in use, we consider bitrates between 1 and 10 Mbit/s to prevent stalling.

Due to our assumptions, no random variables are used, thus the results are exact under the given assumptions, i.e. no confidence intervals are shown.

[2] J. Huang, F. Qian, A. Gerber, Z. M. Mao, S. Sen, and O. Spatscheck, “A Close Examination of Performance and Power Characteristics of 4G LTE Networks,“

Christian Schwartz11

11

Energy Consumption

Energy consumption increases for higher bitrate independent of transmission mechanism.

Live streaming consumes most energy, video is transferred on demand with very low bandwidth.

Download consumes the least amount of energy, device is offline after initial transfer with high bandwidth.

Streaming consumes second least amount of energy.

Christian Schwartz12

12

Wasted Traffic

User behaviour has no significant impact on amount of wasted traffic.

Download mechanism causes most wasted traffic, because data is downloaded at the beginning with full bandwidth.

Live causes least amount of wasted traffic, data is only downloaded on demand.

Again, streaming provides second best result.

Christian Schwartz13

13

Influence of Buffer Threshold for Streaming Mechanism

If energy and wasted data are to be minimized, the smallest possible lower buffer threshold θ should be considered.

Changing the buffer size provides allows the provider to select the trade-off between wasted data and energy consumption.

The streaming mechanism provides a good trade-off between energy consumption and wasted data.

Christian Schwartz14

14

Summary and Conclusion

Using different video transmission mechanisms allows to minimise energy consumption and wasted traffic Streaming offers best trade-off between energy consumption

and wasted data Select smallest lower buffer size so that no stalling is generated Buffer size allows to weight the trade-off to the operators needs

User behaviour showed no impact on wasted traffic

Summary

Selecting appropriate transmission mechanisms and parameters allows the video provider to: decrease the energy consumption for the end user, thus

increasing the QoE reduce the amount of wasted traffic in the datacenter, saving

energy and resources

Conclusion