Chapter 11 Code 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
Jan 01, 2016
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
Example of OVSF Code Tree
: used code : new request
Code blocking:2R
11
Example of OVSF Code Tree
: used code : new request
Internal fragmentation: 3R
12
Example of OVSF Code Tree
: used code : new request
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
: used code : new request
(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)
: used code : new request
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
Mean arrival rate (calls/unit time)
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
Mean arrival rate (calls/unit time)
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
Mean arrival rate (calls/unit time)
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
Mean arrival rate (calls/unit time)
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
Mean arrival rate (calls/unit time)
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
Mean arrival rate (calls/unit time)
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
Mean arrival rate (calls/unit time)
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
Mean arrival rate (calls/unit time)
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
Mean arrival rate (calls/unit time)
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