performance analysis of mpeg-4 video stream with fec error recovery over ieee 802.11 dcf wlan...

42
Performance Analysis of MPEG-4 Video Stream with FEC Error Recovery over IEEE 802.11 DCF WLAN Cheng-Han Lin, Huai-Wen Zhang, Ce-Kuen Shieh Department of Electrical Engineering, National Cheng Kung University, Taiwan Wen-Shyang Hwang* Department of Electrical Engineering, National Kaohsiung University of Applied Sciences, Taiwan WiOpt - WiVid 2013

Upload: alaina-norton

Post on 29-Dec-2015

214 views

Category:

Documents


0 download

TRANSCRIPT

Performance Analysis of MPEG-4 Video Stream with FEC Error Recovery overIEEE 802.11 DCF WLAN

Cheng-Han Lin, Huai-Wen Zhang, Ce-Kuen ShiehDepartment of Electrical Engineering, National Cheng Kung University,

Taiwan Wen-Shyang Hwang*

Department of Electrical Engineering, National Kaohsiung University of Applied Sciences, Taiwan

WiOpt - WiVid 2013

Performance Analysis of MPEG-4 Video Stream with FEC Error Recovery overIEEE 802.11 DCF WLAN

Cheng-Han Lin, Huai-Wen Zhang, Ce-Kuen ShiehDepartment of Electrical Engineering, National Cheng Kung University,

Taiwan Wen-Shyang Hwang*

Department of Electrical Engineering, National Kaohsiung University of Applied Sciences, Taiwan

WiOpt - WiVid 2013

3

Outline

Introduction Proposed Analytical Model Numerical Results Conclusion

WiOpt - WiVid 2013

4

Introduction

Wireless Local Area Network (WLAN) Convenience of wireless access The use of mobile devices is increasing The data rates and bandwidth are increasing Internet-based video streaming applications is popular

The performance analysis of video streaming over wireless networks has emerged as an important issue in the multimedia communications field.

WiOpt - WiVid 2013

5

Introduction

Wireless Local Area Network (WLAN) Convenience of wireless access The use of mobile devices is increasing The data rates and bandwidth are increasing Internet-based video streaming applications is popular

The performance analysis of video streaming over wireless networks has emerged as an important issue in the multimedia communications field.

WiOpt - WiVid 2013

6

Introduction

The literature contains many models based on a two-dimensional Markov chain for analyzing the performance of IEEE 802.11 DCF networks [7-10].

[7] The model assumed that unlimited retransmissions and no wireless bit errors.

[8] The frame retransmission limit is taken into consideration. The effects of wireless bit errors on the frame loss are ignored.

[9] The wireless bit errors is taken into consideration. The frame retransmission limit is ignored.

[10] A model analyzes the effects of both the frame retransmission limit and wireless bit errors.

WiOpt - WiVid 2013

7

Introduction

The literature contains many models based on a two-dimensional Markov chain for analyzing the performance of IEEE 802.11 DCF networks [7-10].

[7] The model assumed that unlimited retransmissions and no wireless bit errors.

[8] The frame retransmission limit is taken into consideration. The effects of wireless bit errors on the frame loss are ignored.

[9] The wireless bit errors is taken into consideration. The frame retransmission limit is ignored.

[10] A model analyzes the effects of both the frame retransmission limit and wireless bit errors.

WiOpt - WiVid 2013

8

Introduction

The [7-10] focus on the system performance, but do not enable the video quality over IEEE 802.11 DCF WLANs to be directly assessed.

[13] Playable Frame Rate (PFR), for analyzing the video quality of MPEG-4 video streaming over WLANs. The assumptions regarding the wireless transmission were overly simple.

[14] A more realistic model in which the effects on the frame losses of wireless channel errors and transmission collisions were both taken into account.

However, the models in [13-14] did not consider the effects of error recovery on the MPEG video streaming quality.

WiOpt - WiVid 2013

9

Introduction

The [7-10] focus on the system performance, but do not enable the video quality over IEEE 802.11 DCF WLANs to be directly assessed.

[13] Playable Frame Rate (PFR), for analyzing the video quality of MPEG-4 video streaming over WLANs. The assumptions regarding the wireless transmission were overly simple.

[14] A more realistic model in which the effects on the frame losses of wireless channel errors and transmission collisions were both taken into account.

However, the models in [13-14] did not consider the effects of error recovery on the MPEG video streaming quality.

WiOpt - WiVid 2013

10

Introduction

This paper proposes an analytical model for evaluating the performance of MPEG-4 video streaming over IEEE 802.11 DCF WLANs with FEC error protection. The model considers both congestion losses and wireless

channel losses. The model enforces the frame retransmission constraint

prescribed in IEEE 802.11. The model takes account of the FEC error recovery

performance in improving the perceived video quality at the receiver end.

WiOpt - WiVid 2013

11

Introduction

This paper proposes an analytical model for evaluating the performance of MPEG-4 video streaming over IEEE 802.11 DCF WLANs with FEC error protection. The model considers both congestion losses and wireless

channel losses. The model enforces the frame retransmission constraint

prescribed in IEEE 802.11. The model takes account of the FEC error recovery

performance in improving the perceived video quality at the receiver end.

WiOpt - WiVid 2013

Outline

Introduction Proposed Analytical Model Numerical Results Conclusion

WiOpt - WiVid 2013

Proposed Analytical Model

1. Performance analysis of IEEE 802.11 DCF WLANs

The loss of a transmission frame can be caused by: Congestion loss (PC)

Wireless loss (PE)

The probability of frame transmission failure

ECECECf PPPPPPP 111

WiOpt - WiVid 2013

Proposed Analytical Model

1. Performance analysis of IEEE 802.11 DCF WLANs

The loss of a transmission frame can be caused by: Congestion loss (PC)

Wireless loss (PE)

The probability of frame transmission failure

ECECECf PPPPPPP 111

WiOpt - WiVid 2013

Proposed Analytical Model

Wireless loss (PE) Loss probability of data frame (PE_data)

Loss probability of ACK frame (PE_ACK)

ACKEdataEACKEdataEE PPPPP ____

lengthdatabitEdataE PP _

__ 11

lengthACKbitEACKE PP _

__ 11

WiOpt - WiVid 2013

Proposed Analytical Model

Wireless loss (PE) Loss probability of data frame (PE_data)

Loss probability of ACK frame (PE_ACK)

ACKEdataEACKEdataEE PPPPP ____

lengthdatabitEdataE PP _

__ 11

lengthACKbitEACKE PP _

__ 11

WiOpt - WiVid 2013

Proposed Analytical Model

Congestion loss (PC) The collision probability for any station competing for channel

access

The probability of an station transmits a frame [1]

m: maximum backoff stage

11

1

121211

1212

mff

mfminf

mff

PPPCWP

PP

111 nCP

[1] “Saturation throughput analysis of error-prone 802.11 wireless networks,” Wiley Journal of Wireless Communications and Mobile Computing 2005

Proposed Analytical Model

Congestion loss (PC) The collision probability for any station competing for channel

access

The probability of an station transmits a frame [1]

m: maximum backoff stage

11

1

121211

1212

mff

mfminf

mff

PPPCWP

PP

111 nCP

[1] “Saturation throughput analysis of error-prone 802.11 wireless networks,” Wiley Journal of Wireless Communications and Mobile Computing 2005

Proposed Analytical Model

The probability of a frame transmission failure

The effective failure probability of each frame

Tmax: maximum number of frame retransmission

1111 nEf PP

iTEC

iC

T

i

maxeffective

maxmax

PPPi

TP

1

0

WiOpt - WiVid 2013

Proposed Analytical Model

The probability of a frame transmission failure

The effective failure probability of each frame

Tmax: maximum number of frame retransmission

1111 nEf PP

iTEC

iC

T

i

maxeffective

maxmax

PPPi

TP

1

0

WiOpt - WiVid 2013

Proposed Analytical Model

2. Analytical model for MPEG-4 video streaming with FEC error recovery

The probability of a successful frame transmission

n: the total number of source frame (k) and redundant frame (h) k: the number of source frame

n

ki

ineffective

ieffectiveeffective PP

i

nPknB 1,,

WiOpt - WiVid 2013

Proposed Analytical Model

2. Analytical model for MPEG-4 video streaming with FEC error recovery

The probability of a successful frame transmission

n: the total number of source frame (k) and redundant frame (h) k: the number of source frame

n

ki

ineffective

ieffectiveeffective PP

i

nPknB 1,,

WiOpt - WiVid 2013

Proposed Analytical Model

The successful transmission probabilities of the I-, P- and B-frames in the GOP

effectiveBBRBB

effectivePPRPP

effectiveIIRII

PSSSBQ

PSSSBQ

PSSSBQ

,,

,,

,,

WiOpt - WiVid 2013

SI, SP, SBNumbers of MAC I-, P-, and B-frames.

SIR, SPR, SBRNumbers of FEC redundant I-, P-, and B- frames.

Proposed Analytical Model

The successful transmission probabilities of the I-, P- and B-frames in the GOP

effectiveBBRBB

effectivePPRPP

effectiveIIRII

PSSSBQ

PSSSBQ

PSSSBQ

,,

,,

,,

WiOpt - WiVid 2013

SI, SP, SBNumbers of MAC I-, P-, and B-frames.

SIR, SPR, SBRNumbers of FEC redundant I-, P-, and B- frames.

Proposed Analytical Model

Playable Frame Rate (PFR) A performance metric for evaluating the quality of video

streaming over lossy network. The ratio of the expected number of playable video frames

at the receiver to the total number of video frames transmitted by the sender.

BPI RRRR

RI, RP, RBPlayable frame rates of I-, P-, and B-frames over entire video sequence.

Proposed Analytical Model

Playable Frame Rate (PFR) A performance metric for evaluating the quality of video

streaming over lossy network. The ratio of the expected number of playable video frames

at the receiver to the total number of video frames transmitted by the sender.

BPI RRRR

RI, RP, RBPlayable frame rates of I-, P-, and B-frames over entire video sequence.

Proposed Analytical Model

The PFR of I-frames

The effective GOP transmission rate

Note that the PFR is computed on a per-second basis RF : the encoding frame rate per second

NP and NB : the number of P- and B-frames in the GOP

II QGR

BP

F

NN

RG

1

Proposed Analytical Model

The PFR of I-frames

The effective GOP transmission rate

Note that the PFR is computed on a per-second basis RF : the encoding frame rate per second

NP and NB : the number of P- and B-frames in the GOP

II QGR

BP

F

NN

RG

1

Proposed Analytical Model

The playable frame rate for P-frame

P

PP

NPIPPNPN

PIPPP

PIP

QRQRR

QRQRR

QRR

1

212

1

...

P

NPP

I

N

iPiP Q

QQQGRR

PP

1

1

1

Proposed Analytical Model

The playable frame rate for P-frame

P

PP

NPIPPNPN

PIPPP

PIP

QRQRR

QRQRR

QRR

1

212

1

...

P

NPP

I

N

iPiP Q

QQQGRR

PP

1

1

1

Proposed Analytical Model

The playable frame rate for B-frame

IBPiBij

BPiBij

QQRR

QRR

1

P

P

Niwhen

Niwhen

10

P

PPN

PIP

NPP

B

N

iIBPBiBPB QQ

Q

QQQQGNRNR

1

1

00

Proposed Analytical Model

The playable frame rate for B-frame

IBPiBij

BPiBij

QQRR

QRR

1

P

P

Niwhen

Niwhen

10

P

PPN

PIP

NPP

B

N

iIBPBiBPB QQ

Q

QQQQGNRNR

1

1

00

Proposed Analytical Model

P

PPN

PIP

NPP

BBPP

NPP

I

BPI

QQQ

QQQN

Q

QQQG

RRRR

111

11

The overall PFR for a FEC-Protected MPEG video stream is equal to the sum of the PFRs of the I-, P- and B-frames, respectively

Proposed Analytical Model

P

PPN

PIP

NPP

BBPP

NPP

I

BPI

QQQ

QQQN

Q

QQQG

RRRR

111

11

The overall PFR for a FEC-Protected MPEG video stream is equal to the sum of the PFRs of the I-, P- and B-frames, respectively

Outline

Introduction Proposed Analytical Model Numerical Results Conclusion

WiOpt - WiVid 2013

Numerical Results

Simulation topology Parameter settings

WiOpt - WiVid 2013

Packet payload (Ldata) 8184 bits Slot time 50 μsACK (LACK) 240 bits DIFS 128μsMAC header 272 bits SIFS 28μsPHY header 128 bits CWmin 32

Channel Data Rate 1 Mbps Tmax 5BEP (PE_bit) 10-5 CI 3.91

NP 3 CP 2.05NB 8 CB 1.52

……

……

Numerical Results

Variation of Playable Frame Rate with number of active stations

Bit Error Probability (BEP) = 10-4 Bit Error Probability (BEP) = 10-6

WiOpt - WiVid 2013

Numerical Results

Variation of Playable Frame Rate with number of active stations

Bit Error Probability (BEP) = 10-4 Bit Error Probability (BEP) = 10-6

WiOpt - WiVid 2013

Numerical Results

Variation of Playable Frame Rate with number of active stations

The maximum backoff stage (m) = 4 The maximum backoff stage (m) = 6

WiOpt - WiVid 2013

Numerical Results

Variation of Playable Frame Rate with number of active stations

The maximum backoff stage (m) = 4 The maximum backoff stage (m) = 6

WiOpt - WiVid 2013

Outline

Introduction Proposed Analytical Model Numerical Results Conclusion

WiOpt - WiVid 2013

Conclusion

This paper has proposed an analytical model for evaluating the video quality of MPEG-4 video streaming over FEC-protected IEEE 802.11 DCF WLANs.

The proposed model considers the effects of congestion and wireless frame losses the performance of the FEC error recovery mechanism

The proposed model has been validated by comparing the predicted results the results obtained from NS-2 simulations two existing analytical models [8, 9].

WiOpt - WiVid 2013