1. 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

2. 2 1. Introduction

3. 3

4. 4 2G GSM 2.5G GPRS 3G UMTS 1. Introduction

5. 5 Structural network architecture 3G UMTS system architecture 1. Introduction Uu Cu IubIur Iu USIM ME UE BS RNC MSC/ VLR UTRAN HLR GGSN GMSC SGSN CN Iur-CS Iur-PS UE

6. 6 OVSF code tree 1. Introduction

7. 7

8. 8 2. Problem statement Code placement problem  Code blocking probability  Internal fragmentation Code replacement problem  Code reassignment cost

9. 9 Example of OVSF Code Tree : used code : new request Code blocking :4R

10. 10 Example of OVSF Code Tree : used code : new request Code blocking :2R

11. 11 Example of OVSF Code Tree : used code : new request Internal fragmentation : 3R

12. 12 Example of OVSF Code Tree : used code : new request Internal fragmentation : 3R

13. 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.Code Placement and Replacement Strategies for Wideband CDMA OVSF Code Tree Management Tseng’s Code placement schemes  Random placement scheme  Leftmost placement scheme  Crowded-first placement scheme

14. 14 A new call of rate 2R

15. 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. 16 A Code Replacement Example A new call of rate 8R

17. 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.Reducing Internal and External Fragmentations of OVSF Codes in WCDMA Systems with Multiple Codes

18. 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. 19 123 … n: number of multicode N(i): ideal (optimal) N(i)N(i) … 4 n n single code multi-code N(i)=Number of 1 s 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. 20 Internal Fragmentations 3. Code Placement and Replacement Strategies

21. 21 Tseng’s multi-code assignment Possible candidates for 6R (n=2: 4R+2R) (decreasing): Leftmost: {C 8,1, C 16,3 } Crowded-first-space: {C 8,8, C 16,14 } Crowded-first-code: {C 8,3, C 16,7 }

22. 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. 23 Tseng’s multi-code re-assignment New requested call: 6R (n=2: 4R+2R) (decreasing): Free capacity: 9R Leftmost

24. 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.The International Conference on Information Networking, ICOIN 2004

25. 25 Outline I.Introduction II.Background Knowledge III.Code Placement and Replacement Strategies IV.Performance Analysis V.Simulation Results VI.Conclusion

26. 26 I. Introduction This paper proposes a code replacement scheme based on ROVSF code tree This scheme aims to develop  Code placement strategy Reduce blocking probability Reduce blocking probability  Code replacement strategy Reduce reassignment cost Reduce reassignment cost

27. 27 Motivation 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. 28 Contributions An alternative solution for code placement and replacement schemes is proposed Advantage of the ROVSF-based scheme  Lower blocking probability Better spectral efficiency  Lower reassignment cost Keep the system overhead low

29. 29 II. Background Knowledge Related Works OVSF Code Tree Rotated-OVSF Code Tree Linear-Code Chain

30. 30 Related 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. 31 Review of OVSF Property

32. 32 Our of ROVSF Property

33. 33 Important Properties of ROVSF Code Tree A ROVSF code is cyclic orthogonal to its two children codes : used code: orthogonal codes

34. 34 Important Properties of ROVSF Code Tree (cont.) A ROVSF code is cyclic orthogonal to any descendent codes : used code: orthogonal codes

35. 35 Important Properties of ROVSF Code Tree (cont.) A ROVSF code is not cyclic orthogonal to any descendent of its brother code : used code XXXX

36. 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. 37 Code Placement in OVSF Code Tree : used code

38. 38 Example of Linear-Code Chain : used code

39. 39 Two Types of Linear-Code Chain

40. 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. 41 LCC Placement Phase If exists (b k, b k-1, b k-2, 0, …, 0) and β< j, then the assignment is failed even if b β = 0 1R1R : used code: new request (1, (1, 1), 0)(1, 0, 0)(1, 0, 1) XXXX

42. 42 Example of LCC Placement Phase If b β = 1 and there is b γ = 1 and γ<β, then the assignment is failed (1, 1, 1)(1, 0, 1) 2R2R (1, 1, 1) : used code: new request X

43. 43 NCC Placement Phase If YR is failed in LCC placement phase, then enters NCC placement phase If there exists linear-code chain (b k =1, 0, …,0), where γ =log 2 Y 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 2 k

44. 44 Example of NCC Placement Phase (0, (1, 1), 0)(1, 1, 1) 2R2R XXX XXXX

45. 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. 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. 47 Example of Replacement Scheme 4R4R : used code: minimum-cost branch: occupied code cost = 1cost = 2cost = 4cost = 3

48. 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=(b k, b k-1, b k-2, …, b 1, b 0 ), where b i = 0  If an incoming transmission rate is 2 k+t, where 1 ≤ t ≤ n-k, we can adjust the length of linear-code chain to be k+t+1

49. 49 Example of Dynamic Adjustment Operation 4R4R : used code: minimum-cost branch: occupied code cost = 1cost = 2 1R1R2R2R

50. 50 IV. Performance Analysis We define the set of allowable states to be The steady-state probability π v can be determined using the following equation: where π 0 is the steady-state probability being in state 0:

51. 51 Call Blocking Probability Then we have call blocking probability P B (i)for iR as: where is the call blocking states for iR Therefore, the overall call blocking probability P B is simply given by:

52. 52 Call Blocking Probability at Different Traffic Load when max SF = 16

53. 53 V. 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. 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. 55 Performance 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. 56 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. 57 Impact of Code Placement (SF=256)

58. 58 Impact of Code Placement (SF=64)

59. 59 Impact of Code Replacement

60. 60 Impact of the Length of LCC on Blocking Probability

61. 61 Impact of the Length of LCC on Reassignment Cost

62. 62 Impact of Call Patterns on Blocking Probability

63. 63 Impact of Call Patterns on Reassignment Cost

64. 64 VI. 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