CS 4472 Mobile ComputingResearch Paper Presentation

I.Suventhan 120649PLinganesan 120337H

DAB: Dynamic and Agile Buffer-control for Streaming Videos on Mobile Devices

Ying Mao, Jiayin Wang, Bo Sheng

Introduction Smart phones become one of the most evolutionary devices in the history of computing.

Streaming Video-on-Demand has become a popular application and a significant consumer of network bandwidth.

It is reported that YouTube and Netflix accounts for 17.3%, 5.1% of North American mobile data traffic.

Problem Designing mobile apps for video streaming faces new challenges that do not exist in traditional wired Internet.

Some of the Major Issues◦ Network connection◦ User mobility◦ Network cost

These factors affect the quality of playback

Buffer Mechanism In order to guarantee the quality of playback on mobile device, media players have to buffer data beforehand.

Currently, there are two commonly use buffer mechanisms.◦ MAX : Buffer size is set to be infinite, media player

downloads data as much as possible.◦ Flip-Flop : Buffer size is static and less than the size

of the video, media player always maintains this buffer size.

Important Notes Mobile users only consume part or a small portion of the video.

The optimal size of the video buffer on smart phones depends on,◦ Video quality◦ Wireless link quality

The situation where buffer control is ineffective,◦ Link quality is poor◦ Mobile network temporarily disconnected

◦ E.g. Handoff between AP

Dynamic and Agile Buffer Control

Dynamic and Agile Buffer-control mechanism that take both users' behavior and mobility into consideration.

Consider the link quality and predict the trend of change. For this implementation, DAB uses,

◦ Readings of RSSI – Received Signal Strength Indicator◦ Readings of AccelerometerTo indicate the link quality and user mobility.

Basic Idea

Measurement of RSSI RSSI is indicate the signal strengths from the associated AP. RSSI is an effective numeric indicator for the current link quality.

Here use indexes (1 to 5) to represents the RSSI values with span of 10dB

Assumption◦ The client is leaving the associated AP if indexes become

larger. ◦ Similarly, if the indexes get smaller, we suppose the user is

moving towards the AP

Measurement of Accelerometer

Accelerometer is used to estimates the moving speed to help predict the change of link quality.

Two Schemes:◦ Detect whether a user is moving

Set a threshold τ(0.35) for the average amplitude to distinguish if a user is static or moving. From experiments value range is [0.86, 2.44] with avg of 1.034

◦ Estimate a user’s moving speedS = F · L.

F be the stride frequency (steps/sec) L is the length of one step.

Average amplitudes of 10 users

Average amplitudes of moving user

Estimate User’s Moving Speed

We continuously detect the average amplitude A’ of accelerometer every tm seconds and add the value to a list A = { A’1,A’2, A’3, ..., A’N}.◦ A’N = N · tm

The time interval of two peaks or two roughs indicate the time interval for one step, represented as C.◦ C = ( j − i) · tm◦ F = 1/C◦ The level index (S L) indicates the number of steps per second.

S L = F

Algorithm 1Check the user is static or moving

Update highest/lowest values

calculate the average time interval of each two consecutive peaks/troughs

Accelerometer Statistics in different speed levels

Dynamic and Agile Buffer control

RSSI Accelerometer• Values are more fluctuating • More consistent• Measured at discrete time

points• Monitored continuously

• Directly reflect the link quality • Indirectly indicate the link change

• Represent the quality of downlink

• Not necessarily linked to the signal quality

Compare RSSI and Accelerometer readings

Algorithm

User starts moving Start to monitor RSSI and collect m measurements

The Link quality is getting better

The Link quality is degrading

Reduce the buffer size towards Bmin

The Link quality is gets worse, increase the buffer size

User is static

Bu be the size of the actual data stored in the buffer

α (0, 1) and ∈ β>1

Shrink the buffer size to 0.9B

Use 0.9 as a threshold for checking the utilization of the buffer

Implementation & Performance EvaluationWorkload & Setup

•YouTube video, 115 seconds and 9.38MB

•For DAB Algorithm, set α = 0.3 and β = 1.1

•Compare with Max and Flip-flop mechanisms

•Smartphone connects with route that running Tcpdump

•Move around the router to simulate mobility of a user

Key Performance Metrics• Total stalling duration: time duration that C_BS – C_PS = 0 ,

where C_BS is the current buffered size and C_PS the current played size

• Buffer efficiency at a given time (Ei)

Here CBSi and CPSi are the i-th current buffered size and current played size respectively.

C BS i -

C PS i

Total stalling duration (fast connection)

CBSi/CPSi

38

Buffer efficiency (fast connection)

C BS i -

C PS i

Total stalling duration (slow connection)

ConclusionIn this paper, Authors present DAB, a dynamic buffer-control scheme that adaptively adjust the video buffer size based on the measurement of RSSI and accelerometer. Their solution predicts the change of link quality and correspondingly change the buffer size to help maintain a smooth playback while minimizing the bandwidth consumption. The experimental results have show that their solution is superior to typical buffer schemes.