game-theoretic network bandwidth distribution for … · game-theoretic network bandwidth...

34
Game-Theoretic Network Bandwidth Distribution for Self- Adaptive Cameras Gautham Nayak Seetanadi 1 Martina Maggio, Karl-Erik Årzen 1 Luis Almeida 2 Luis Oliveira 3 1 Department of Automatic Control, Lund University 2 Universidade do Porto / Faculdade de Engenharia 3 Department of Computer Science, University of Pittsburgh

Upload: tranthuan

Post on 30-Apr-2018

225 views

Category:

Documents


1 download

TRANSCRIPT

Game-Theoretic Network Bandwidth Distribution for Self-

Adaptive Cameras

Gautham Nayak Seetanadi1

Martina Maggio, Karl-Erik Årzen1 Luis Almeida2 Luis Oliveira3

1Department of Automatic Control, Lund University2Universidade do Porto / Faculdade de Engenharia

3Department of Computer Science, University of Pittsburgh

Introduction• Multiple adaptive camera network

• Game-theoretic resource manager

• Improved bandwidth utilisation

• Local PI Control

• Convergence guarantees on bandwidth

Motivation

• Problem of Allocation• Finite Resource

• Multiple Entities

• Fair Bandwidth Distribution

• Global Threshold Limit

• Successfully applied to Multicore systems [1]

[1] M.Maggio et al, “A game-theoretic resource manager for rt applications”, In Euromicro Conference on Real-Time Systems, 2013

Decentralised Resource Allocation

• Resource Manager Allocates Resources

• Cameras Choose Service Level• Change Quality

Data

ControlCamera 1

�1

Camera 2 Camera n

�2 �n Resource Manager

Ethernet

Decentralised Resource Allocation

• Resource Manager Allocates Resources

• Cameras Choose Service Level• Change Quality

• Determines Responsibility�i

Data

ControlCamera 1

�1

Camera 2 Camera n

�2 �n Resource Manager

Ethernet

Resource Manager• Game Theory Based

• Matching function determines fairness

fp,w =Bp,w � sp,w

Bp,w

Resource Manager

Resource Manager

Camera2

Camera1

Managert = 0 t = 1 t = 2

50%

50%

Resource Manager

Camera2

Camera1

Managert = 0 t = 1 t = 2

50% 75%

50% 25%

CameraOriginal Frames

Encoded Frames

Camera• Camera generates an image of size

• Adaptable quality

• Normalised error

sp,w = 0.01 · qp,w · sp,max

+ �sp,w,

Can be -ve or +ve

Camera LoopController

(kp, ki)Camera

Quality

(qp,w)

�1

P

Allocated

Bandwidth

(Bp,w)

Error

(ep,w)

Frame

Size

(sp,tw )

Network Allocation• SDN using Flexible Time Triggered - SE [2]

[2] P.Pedreiras and L.Almeida, “The flexible time triggered paradigm. An approach to QOS management in distributed real-time systems”, 2003

Camera2

Camera1

Managert = 0 t = 1 t = 2

NETWORK ALLOCATION• SDN using Flexible Time Triggered - SE [2]

[2] P.Pedreiras and L.Almeida, “The flexible time triggered paradigm. An approach to QOS management in distributed real-time systems”, 2003

Camera2

Camera1

Managert = 0 t = 1 t = 2

I1,1 I1,2

I2,1 I2,2

NETWORK ALLOCATION• SDN using Flexible Time Triggered - SE [2]

[2] P.Pedreiras and L.Almeida, “The flexible time triggered paradigm. An approach to QOS management in distributed real-time systems”, 2003

Camera2

Camera1

Managert = 0 t = 1 t = 2

I1,1 I1,2

I2,1 I2,2

I1,2 : q1,2 ! s1,2B1,t=0 ! B1,w=2

NETWORK ALLOCATION• SDN using Flexible Time Triggered - SE [2]

[2] P.Pedreiras and L.Almeida, “The flexible time triggered paradigm. An approach to QOS management in distributed real-time systems”, 2003

Camera2

Camera1

Managert = 0 t = 1 t = 2

I1,1 I1,2 I1,4 I1,5

I2,1 I2,2 I2,3 I2,4

I1,2 : q1,2 ! s1,2B1,t=0 ! B1,w=2

Implementation• Cameras

• Logitech c270. COTS

• OpenCV

• Network

• Deadlines enforced using Flexible Time Triggered Switched Ethernet (FTT-SE)

• i7-4790 8 core PC running Fedora

Experimental Setup Blue Camera Red Camera

Assessment• Criteria?

• Fair Bandwidth Usage (Manager)• Complete Bandwidth Utilisation (Camera)• SSIM (Structural Similarity Index)

• SSIM

Experiments Resource Allocator Camera

Equal Bandwidth Distribution No Adaptation

Equal Bandwidth Distribution DARTES Model [2]

Equal Bandwidth Distribution PI Controller

Game-Theoretic RA PI Controller

[2] J. Silvestre-Blanes, L. Almeida, R. Marau, and P. Pedreiras. “Online qos management for multimedia real-time transmission in industrial networks.” IEEE Transactions on Industrial Electronics, 58(3), March 2011

Experiment 1

Resource Allocator CameraEqual Bandwidth Distribution No Adaptation

0 10 20 30 40 50 60 70 80

0.0

10.0

20.0

30.0

Time [s]

BW

[Mbps]

AllocBW Camera 1 AllocBW Camera 2

InstBW Camera 1 InstBW Camera 2

Experiment 1

Resource Allocator CameraEqual Bandwidth Distribution No Adaptation

0 10 20 30 40 50 60 70 80

0.0

10.0

20.0

30.0

Time [s]

BW

[Mbps]

AllocBW Camera 1 AllocBW Camera 2

InstBW Camera 1 InstBW Camera 2

0 10 20 30 40 50 60 70 80

0.0

10.0

20.0

30.0

Time [s]

BW

[Mbps]

AllocBW Camera 1 AllocBW Camera 2

InstBW Camera 1 InstBW Camera 2

Experiment 1

Resource Allocator CameraEqual Bandwidth Distribution No Adaptation

Experiment 2

0 10 20 30 40 50 60 70 80 90 100 110 120

0.0

1.0

2.0

3.0

4.0

Time [s]

BW

[Mbps]

AllocBW Camera 1 AllocBW Camera 2

InstBW Camera 1 InstBW Camera 2

Resource Allocator CameraEqual Bandwidth Distribution DARTES Model

Experiment 2

0 10 20 30 40 50 60 70 80 90 100 110 120

0.0

1.0

2.0

3.0

4.0

Time [s]

BW

[Mbps]

AllocBW Camera 1 AllocBW Camera 2

InstBW Camera 1 InstBW Camera 2

Resource Allocator CameraEqual Bandwidth Distribution DARTES Model

Experiment 3

0 10 20 30 40 50 60 70 80 90 100 110 120

0.0

1.0

2.0

3.0

4.0

Time [s]

BW

[Mbps]

AllocBW Camera 1 AllocBW Camera 2

InstBW Camera 1 InstBW Camera 2

Resource Allocator CameraEqual Bandwidth Distribution PI Controller

Experiment 3

0 10 20 30 40 50 60 70 80 90 100 110 120

0.0

1.0

2.0

3.0

4.0

Time [s]

BW

[Mbps]

AllocBW Camera 1 AllocBW Camera 2

InstBW Camera 1 InstBW Camera 2

Resource Allocator CameraEqual Bandwidth Distribution PI Controller

Experiment 4

0 5 10 15 20 25 30 35 40 45 50 55 60

0.0

1.0

2.0

3.0

4.0

Time [s]

BW

[Mbps]

AllocBW Camera 1 AllocBW Camera 2

InstBW Camera 1 InstBW Camera 2

Resource Allocator CameraGame-Theoretic RA PI Controller

Experiment 4

0 5 10 15 20 25 30 35 40 45 50 55 60

0.0

1.0

2.0

3.0

4.0

Time [s]

BW

[Mbps]

AllocBW Camera 1 AllocBW Camera 2

InstBW Camera 1 InstBW Camera 2

Resource Allocator CameraGame-Theoretic RA PI Controller

SSIM

0 200 400 600 800 1000 1200 1400 1600

0

0,05

0,1

Frame Number

SSIM with PI controller � SSIM with DARTES model

Camera 1

Camera 2

SSIM

0 200 400 600 800 1000 1200 1400 1600

0

0,05

0,1

Frame Number

SSIM with PI controller � SSIM with DARTES model

Camera 1

Camera 2

Conclusions

• CPU allocation strategy

• Need for adaptation

• PI control advantages

• Game-Theoretic resource manager efficiency

• Convergence guarantees

Future Work

• Time triggered network manager decisions have drawbacks

• Moving to event triggered

[email protected]

0 10 20 30 40 50 60 70 80 90 100 110 120

0.0

1.0

2.0

3.0

4.0

Time [s]

BW

[Mbps]

AllocBW Camera 1 AllocBW Camera 2

InstBW Camera 1 InstBW Camera 2

0 10 20 30 40 50 60 70 80 90 100 110 120

0.0

1.0

2.0

3.0

4.0

Time [s]

BW

[Mbps]

AllocBW Camera 1 AllocBW Camera 2

InstBW Camera 1 InstBW Camera 2

0 5 10 15 20 25 30 35 40 45 50 55 60

0.0

1.0

2.0

3.0

4.0

Time [s]

BW

[Mbps]

AllocBW Camera 1 AllocBW Camera 2

InstBW Camera 1 InstBW Camera 2

Resource Allocator Camera