Associate Prof. Yuh-Shyan Chen Dept. of Computer Science and Information Engineering

Chapter 11Code Placement and Replacement

Strategies for Wideband CDMA OVSF/ROVSF Code Tree Management

Associate Prof. Yuh-Shyan Chen

Dept. of Computer Science and Information Engineering

National Chung-Cheng University

2

1. Introduction

3

1. Introduction

4

2G GSM

2.5G GPRS

3GUMTS

1. Introduction

5

Structural network architecture 3G UMTS system architecture

1. Introduction

Uu

Cu Iub Iur

Iu

USIM

ME

UE

BS

BS

BS

BS

RNC

RNC

MSC/VLR

UTRAN

HLR

GGSN

GMSC

SGSN

CN

Iur-CS

Iur-PS

UE

6

OVSF code tree

(C)

(C,C)

(C,-C)

C8,8 = (-1,-1,-1,-1,-1,-1,-1,-1)

C8,7 = (1,-1,-1,1,1,-1,-1,1)

C8,6 = (1,-1,1,-1,-1,1,-1,1)

C8,5 = (1,-1,1,-1,1,-1,1,-1)

C8,4 = (1,1,-1,-1,-1,-1,1,1)

C8,3 = (1,1,-1,-1,1,1,-1,-1)

C8,2 = (1,1,1,1,-1,-1,-1,-1)

C8,1 = (1,1,1,1,1,1,1,1)

C4,4 = (1,-1,-1,1)

C4,3 = (1,-1,1,-1)

C4,2 = (1,1,-1,-1)

C4,1 = (1,1,1,1)

C2,2 = (1,-1)

C2,1 = (1,1)

C1,1 = (1) . . .

(a) (b)

SF = 1 SF = 2 SF = 4 SF = 8

1. Introduction

7

1. Introduction

8

2. Problem statement

Code placement problem Code blocking probability Internal fragmentation

Code replacement problem Code reassignment cost

9

Example of OVSF Code Tree

: used code : new request

Code blocking

:4R

10



Code blocking:2R

11



Internal fragmentation: 3R

12



Internal fragmentation: 3R

13

3. Code placement and replacement strategies

Y.-C. Tseng and C.-M. Chao, "Code Placement and Replacement Strategies for Wideband CDMA OVSF Code Tree Management", IEEE Trans. on Mobile Computing, Vol. 1, No. 4, Oct.-Dec. 2002, pp. 293-302.

Tseng’s Code placement schemes Random placement scheme Leftmost placement scheme Crowded-first placement scheme

14

A new call of rate 2R

15

3. Code replacement strategy

Tseng’s Code replacement schemes Find the minimum-cost branch

Based on DCA Relocate until done

Based on code placement schemes

16

A Code Replacement Example

A new call of rate 8R

17

Multi-Code Approach

C.-M. Chao, Y.-C. Tseng, and L.-C. Wang, "Reducing Internal and External Fragmentations of OVSF Codes in WCDMA Systems with Multiple Codes", IEEE Wireless Communications and Networking

Conf. (WCNC), 2003.

18

Tseng’s multi-code assignment

Order of Assignment: increasing decreasing

Co-location of Codes: united strategy separated strategy

Assignment of Individual Codes: Random Leftmost Crowded-first-space Crowded-first-code

19

otherwiseif

iNnandni

n

ifini

i

n

)2(2

)(2

12

)(lg1lg

lg

1 2 3 …

n: number of multicode N(i): ideal (optimal)

N(i)

…

4

n n n n single code

multi-code

N(i)=Number of 1s in (i)2

For any given i, we can find a N(i)

Number of code

3. Code Placement and Replacement Strategies

Tseng’s internal fragmentation solution

20

Internal Fragmentations

3. Code Placement and Replacement Strategies

21

Tseng’s multi-code assignment

Possible candidates for 6R (n=2: 4R+2R) (decreasing): Leftmost: {C8,1 , C16,3}Crowded-first-space: {C8,8 , C16,14}Crowded-first-code: {C8,3 , C16,7}

22

Tseng’s multi-code re-assignment

Dynamic code assignment (DCA) scheme was proposed to solve the single-code reassignment problem

Authors utilize the DCA scheme as a basic construction block. When moving codes around. Authors also consider where to place those codes that are migrated so as to reduce the potential future reassignment cost (this issue is ignored in DCA).

23

Tseng’s multi-code re-assignment

New requested call: 6R (n=2: 4R+2R) (decreasing): Free capacity: 9R Leftmost

24

Our Single-Code Placement and Replacement Strategies Yuh-Shyan Chen and Ting-Lung Lin, "Code Placement

and Replacement Schemes for W-CDMA Rotated-OVSF Code Tree Management," is submitted to The International Conference on Information Networking, ICOIN 2004, Feb. 18 - Feb. 20, 2004, Korea.

25

OutlineOutline

I. IntroductionII. Background KnowledgeIII. Code Placement and Replacement

StrategiesIV. Performance AnalysisV. Simulation ResultsVI. Conclusion

26

I. I. IntroductionIntroduction

This paper proposes a code replacement scheme based on ROVSF code tree

This scheme aims to develop Code placement strategyCode placement strategy

Reduce blocking probabilityReduce blocking probability Code replacement strategyCode replacement strategy

Reduce reassignment costReduce reassignment cost

27

MotivationMotivation

Existing OVSF-based scheme has a lower spectral efficiency and a higher system overhead

This study aims to develop a more efficient channelization code scheme

28

ContributionsContributions

An alternative solution for code placement and replacement schemes is proposed

Advantage of the ROVSF-based schemeLower blocking probability

Better spectral efficiencyLower reassignment cost

Keep the system overhead low

29

II. Background KnowledgeII. Background Knowledge

Related Works OVSF Code Tree Rotated-OVSF Code Tree

Linear-Code Chain

30

Related WorksRelated Works

OVSF-based Scheme Dynamic Code Assignment

IEEE Journal on Selected Areas in Comm., Aug. 2000 Single-code Placement & Replacement

Proc. of IEEE Trans. on Mobile Computing, 2002. (Y.C. Tseng) Multi-code Assignment

IEEE Wireless Comm. and Networking Conf., 2003. (Y.C. Tseng)

OVSF-like Scheme FOSSIL

Proc. of IEEE ICC, 2001.

31

Review of OVSF Property

(C)

(C,C)

(C,-C)

C8,8 = (-1,-1,-1,-1,-1,-1,-1,-1)

C8,7 = (1,-1,-1,1,1,-1,-1,1)

C8,6 = (1,-1,1,-1,-1,1,-1,1)

C8,5 = (1,-1,1,-1,1,-1,1,-1)

C8,4 = (1,1,-1,-1,-1,-1,1,1)

C8,3 = (1,1,-1,-1,1,1,-1,-1)

C8,2 = (1,1,1,1,-1,-1,-1,-1)

C8,1 = (1,1,1,1,1,1,1,1)

C4,4 = (1,-1,-1,1)

C4,3 = (1,-1,1,-1)

C4,2 = (1,1,-1,-1)

C4,1 = (1,1,1,1)

C2,2 = (1,-1)

C2,1 = (1,1)

C1,1 = (1) . . .

(a) (b)

SF = 1 SF = 2 SF = 4 SF = 8

32

Our of ROVSF Property

(a) (b)

RC2i, 2j-2 = (-A,-A)

RC2i, 2j-3 = (A,-A)

RC2i, 2j-1 = (B,-B)

RC2i, 2j = (-B,-B)

RCi, j-1 = (B)

RCi, j = (A)

SF = i SF = 2i

RC8,8 = (1,1,1,1,-1,-1,-1,-1)

RC8,7 = (1,-1,-1,1,1,-1,-1,1)

RC8,6 = (-1,-1,1,1,1,1,-1,-1)

RC8,5 = (1,1,-1,-1,1,1,-1,-1)

RC8,4 = (1,-1,1,-1,-1,1,-1,1)

RC8,3 = (-1,1,-1,1,-1,1,-1,1)

RC8,2 = (-1,1,1,-1,1,-1,-1,1)

RC8,1 = (1,-1,-1,1,1,-1,-1,1)

RC4,4 = (-1,-1,1,1)

RC4,3 = (1,1,1,1)

RC4,2 = (-1,1,1,-1)

RC4,1 = (1,-1,1,-1)

RC2,2 = (-1,1)

RC2,1 = (-1,-1)

RC1,1 = (1)

SF = 1 SF = 2 SF = 4 SF = 8

. . .

33

Important Properties of ROVSF Code Tree

A ROVSF code is cyclic orthogonal to its two children codes

: used code : orthogonal codes

34

Important Properties of ROVSF Code Tree (cont.)

A ROVSF code is cyclic orthogonal to any descendent codes

: used code : orthogonal codes

35

Important Properties of ROVSF Code Tree (cont.)

A ROVSF code is not cyclic orthogonal to any descendent of its brother code

: used code

X X X X

36

Linear-Code Chain

A collection of mutually orthogonal codes Every node of a OVSF code tree is mapping to the

corresponding node of a ROVSF code tree to form the linear-code chain

Prior to designate where to allocate each supported request Rate restriction of transmission requests Reduce blocking of high-rate request

37

Code Placement in OVSF Code Tree

: used code

38

Example of Linear-Code Chain

: used code

39

Two Types of Linear-Code Chain

(a)[8R, 4R, 2R, 1R] [8R, 4R, 2R, 1R, 1R]

(b)

[8R, 4R]

(e)[8R, 4R, 4R]

(f)

[8R, 4R, 2R]

(c)[8R, 4R, 2R, 2R]

(d)

40

III. Code Placement and Replacement Strategies

Placement Scheme Linear-Code Chain (LCC) Placement Phase Non-linear-Code Chain (NCC) Placement Phase

Replacement Scheme Dynamic Adjustment Operation of Linear-Code

Chain

41

LCC Placement Phase

If exists (bk, bk-1, bk-2, 0,…, 0) and β< j, then the assignment is failed even if bβ= 0

1R


(1, (1, 1), 0) (1, 0, 0)(1, 0, 1)

X X X X

42

Example of LCC Placement Phase

If bβ= 1 and there is bγ= 1 and γ<β, then the assignment is failed

(1, 1, 1) (1, 0, 1)

2R

(1, 1, 1)


X

43

NCC Placement Phase

If YR is failed in LCC placement phase, then enters NCC placement phase

If there exists linear-code chain (bk=1, 0,…,0), where γ =log2Y and γ= k, we may assign YR to neighboring node of node N of linear-code chain on the same level of ROVSF code tree, where transmission rate of node N is 2k

44

Example of NCC Placement Phase

(0, (1, 1), 0)(1, 1, 1)

2R

X X XX X X X

45

Summary of Code Placement

More codes are assigned in linear-code chain will result in a lower blocking probability

Dynamic adjustment operation of linear-code chain is introduced in code replacement scheme

46

Replacement Scheme

The purpose of this procedure Force the code blocking probability to zero

We adopt the same concept of DCA algorithm ROVSF-version DCA algorithm

Our proposed placement strategy is adopted while relocating each code

47

Example of Replacement Scheme

4R

: used code : minimum-cost branch : occupied code

cost = 1 cost = 2 cost = 4 cost = 3

48

Dynamic Adjustment Operation

Aims to overcome drawbacks of fixed length of LCC Maximum transmission rate is limited Not applicable to variable traffic patterns

If exists BW=(bk, bk-1, bk-2,…, b1, b0), where bi = 0 If an incoming transmission rate is 2k+t, where 1 ≤ t ≤ n-k, we

can adjust the length of linear-code chain to be k+t+1

49

Example of Dynamic Adjustment Operation

4R

: used code : minimum-cost branch : occupied code

cost = 1 cost = 2

1R 2R

50

IV. Performance Analysis

We define the set of allowable states to be

The steady-state probability πv can be determined using the following equation:

where π0 is the steady-state probability being in state 0:

51

Call Blocking Probability

Then we have call blocking probability PB(i)for iR as:

where

is the call blocking states for iR Therefore, the overall call blocking probability PB is simply

given by:

52

Call Blocking Probability at Different Traffic Load when max SF = 16

53

V. Experimental ResultsV. Experimental Results

Simulation environment Capacity test : code-limited Maximum spreading factors are 64 and 256 Call arrival process is Poisson distributed with mean

arrival rate λ=1 -16 calls/unit time (SF=64), λ=4 -64 calls/unit time (SF=256)

Call duration is exponentially distributed with a mean value of 4 unit of time

Possible transmission rates are 1R, 2R, 4R, and 8R

54

The Compared Targets

OVSF-based scheme Random Leftmost Crowded-first Mostuser-first

ROVSF-based scheme Leftmost Crowded-first

ROVSF code tree + Crowded-first strategy Mostuser-first

ROVSF code tree + Mostuser-first strategy

55

Performance MetricsPerformance Metrics

Blocking Probability The probability of a new request cannot be

accepted because the orthogonality cannot be maintained for this rate, although the system still has enough excess capacity

Utilization of LCC The number of incoming requests assigned on

LCC divided by the total number of accepted requests

56

Performance Metrics (cont.) Performance Metrics (cont.)

Number of Reassigned Codes The total number of necessary reassignments

of all occupied codes to support the new request when occurring code blocking

57

Impact of Code Placement (Impact of Code Placement (SFSF=256)=256)

0

0.05

0.1

0.15

0.2

0.25

0.3

4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64

Mean arrival rate (calls/unit time)

Blo

ckin

gpr

obab

ilit

y

OROLOCOMRLRCRM

(a)

0.7

0.75

0.8

0.85

0.9

0.95

1

8 12 16 20 24 28 32 36 40 44 48 52 56 60 64


Uti

liza

tion

ofL

CC

RLRCRM

(b)

58

Impact of Code Placement (Impact of Code Placement (SFSF=64)=64)

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16


Blo

ckin

gpr

obab

ilit

y

OROLOCOMRLRCRM

(a)

0.75

0.8

0.85

0.9

0.95

1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16


Uti

liza

tion

ofL

CC

RLRCRM

(b)

59

Impact of Code ReplacementImpact of Code Replacement

0

200

400

600

800

1000

1200

1400

1600

4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64


Num

bero

frea

ssig

ned

code

s

OROLOCOMRLRCRM

(a)

0

200

400

600

800

1000

1200

1400

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16


Num

bero

frea

ssig

ned

code

s OROLOCOMRLRCRM

(b)

60

Impact of the Length of LCC on Impact of the Length of LCC on Blocking ProbabilityBlocking Probability

0

0.1

0.2

0.3

0.4

0.5

0.6

4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64


Blo

ckin

gpr

obab

ilit

y RL2 RL3RL4 RL5RL6

(a)

0

0.1

0.2

0.3

0.4

0.5

0.6

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16


Blo

ckin

gpr

obab

ilit

y RL2 RL3RL4 RL5RL6

(b)

61

Impact of the Length of LCC on Impact of the Length of LCC on Reassignment CostReassignment Cost

0

100

200

300

400

500

600

700

800

900

4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64


Num

bero

frea

ssig

ned

code

s RL2RL3RL4RL5RL6

(a)

0

100

200

300

400

500

600

700

800

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16


Num

bero

frea

ssig

ned

code

s RL2RL3RL4RL5RL6

(b)

62

Impact of Call Patterns on Impact of Call Patterns on Blocking ProbabilityBlocking Probability

4

24

4464

0

0.05

0.1

0.15

0.2

0.25

0.3

OLOCRL

OLOCRL

OLOCRL

OLOCRL

1:1:1:14:4:1:1

1:1:4:44:1:1:4Traffic

lo adCode

pat tern

Blocking

probability

(a)

1

6

1116

0

0.05

0.1

0.15

0.2

0.25

OLOCRL

OLOCRL

OLOCRL

OLOCRL

1:1:1:14:4:1:1

1:1:4:44:1:1:4Traffic

lo adCode

pat tern

Blocking

probability

(b)

63

Impact of Call Patterns on Impact of Call Patterns on Reassignment CostReassignment Cost

1

6

1116

0

100

200

300

400

500

600

700

800

OLOCRL

OLOCRL

OLOCRL

OLOCRL

Num

berof

reassignedcodes

1:1:1:14:4:1:1

1:1:4:44:1:1:4Traffic

lo adCode

pat tern

(b)

4

24

4464

0

200

400

600

800

1000

1200

OLOCRL

OLOCRL

OLOCRL

OLOCRL

Num

berof

reassignedcodes

1:1:1:14:4:1:1

1:1:4:44:1:1:4Traffic

lo adCode

pat tern

(a)

64

VI. ConclusionsVI. Conclusions

This paper proposes a novel approach for channelization code in WCDMA Based on the Rotated-OVSF code tree

The simulation results illustrate that our scheme offers a lower blocking probability and lower reassignment cost, compared to OVSF-based scheme

Associate Prof. Yuh-Shyan Chen Dept. of Computer Science and Information Engineering

Documents