1. Optimization of Wireless Station Time Slot Allocation with Consideration of Throughput and Delay Constraints
指導教授：林永松 博士
研究生：林岦毅

2. Outline Introduction Problem Description Model Description 2018/11/22
OPLab, IM, NTU

3. Introduction

4. Background
The growth of the Internet has given rise to demands for higher data rate and more multimedia service.
New wireless networks, such as 3G and WiMAX, were designed to provide higher capacity for data transmission
QoS is more important for multimedia service
2018/11/22
OPLab, IM, NTU

5. Motivation
Try to maximize the total system revenue by optimizing the time slot allocation at the wireless station with considering the throughput and delay constraints.
2018/11/22
OPLab, IM, NTU

6. Problem Description

7. Problem
Packets arrive
System Queue … …
2018/11/22
OPLab, IM, NTU

8. Problem Assumptions (1/2)
There are four service classes, (1,2,3,4), in the system and service class 1 has the highest occupancy priority
The data is divided into fixed size packets
Each packet can be completely transmitted in a slot time, and each slot can transmit only one packet
2018/11/22
OPLab, IM, NTU

9. Problem Assumptions (2/2)
The channel quality will remain the same during the transmission
Transmission error will not happen during transmission
Each service class will have its own traffic arrival process
2018/11/22
OPLab, IM, NTU

10. Problem Description (1/4)
Given parameters:
The size of system queue is B packets, which will be shared between the four service classes
The maximum number of packets that can be transmitted in a frame is N
We defined revenue matrix for the four service classes
2018/11/22
OPLab, IM, NTU

11. Problem Description (2/4)
Given parameters:
Each service class has its own throughput and delay constraints
State transition matrix
2018/11/22
OPLab, IM, NTU

12. Problem Description (3/4)
Objective:
To maximize the total system revenue
Subject to:
The throughput and delay requirement of each service classes
The number of packets that can be transmitted in a frame
2018/11/22
OPLab, IM, NTU

13. Problem Description (4/4)
To determine:
The best time slot allocation policy for the system
2018/11/22
OPLab, IM, NTU

14. Model Description

15. Given parameters (1/2)
2018/11/22
OPLab, IM, NTU

16. Given parameters (2/2)
2018/11/22
OPLab, IM, NTU

17. State transition probability (1/4)
The arrival process of each service class is Poisson arrival with different arrival rate.
Consider a discrete-time Markov decision processes
2018/11/22
OPLab, IM, NTU

18. State transition probability (2/4)
2018/11/22
OPLab, IM, NTU

19. State transition probability (3/4)
(9,1,2,0)
(1,1,1,0)
B=12
N=6
(5,1,0,0)
2018/11/22
OPLab, IM, NTU

20. State transition probability (4/4)
2018/11/22
OPLab, IM, NTU

21. Decision Variable (1/2)
2018/11/22
OPLab, IM, NTU

22. Decision Variable (2/2)
2018/11/22
OPLab, IM, NTU

23. Problem Formulation (1/3)
Objective:
Subject to:
(LP1)
(LP1.1)
(LP1.2)
(LP1.3)
2018/11/22
OPLab, IM, NTU

24. Problem Formulation (2/3)
(LP1.4)
(LP1.5)
(LP1.6)
(LP1.7)
2018/11/22
OPLab, IM, NTU

25. Problem Formulation (3/3)
(LP1.8)
(LP1.9)
(LP1.10)
(LP1.11)
2018/11/22
OPLab, IM, NTU

26. Problem Reformulation (1/3)
Objective:
Subject to:
(LP2)
(LP1.1)
(LP1.2)
(LP1.3)
2018/11/22
OPLab, IM, NTU

27. Problem Reformulation (2/3)
(LP1.4)
(LP1.5)
(LP1.6)
(LP1.7)
2018/11/22
OPLab, IM, NTU

28. Problem Reformulation (3/3)
(LP1.8)
(LP1.9)
(LP1.10)
(LP1.11)
2018/11/22
OPLab, IM, NTU

29. Lagrangean Relaxation (1/3)
Objective:
(LP3)
2018/11/22
OPLab, IM, NTU

30. Lagrangean Relaxation (2/3)
Subject to:
(LP3.1)
(LP1.2)
(LP1.3)
(LP1.4)
2018/11/22
OPLab, IM, NTU

31. Lagrangean Relaxation (3/3)
(LP1.5)
(LP1.6)
(LP1.7)
(LP1.8)
(LP1.9)
2018/11/22
OPLab, IM, NTU

32. Thanks for your listening