avatar mobility in
play

Avatar Mobility in 1 Wei Tsang Ooi Mehul Motani Huiguang Liang - PowerPoint PPT Presentation

Avatar Mobility in 1 Wei Tsang Ooi Mehul Motani Huiguang Liang Ian Tay Ming Feng Neo National University of Singapore 2 My life is so great that I literally wanted a second one! - Dwight Schrute, The


  1. 新加坡國立大學 黃瑋璨 Avatar Mobility in 1

  2. Wei Tsang Ooi Mehul Motani Huiguang Liang Ian Tay Ming Feng Neo National University of Singapore 2

  3. “My life is so great that I literally wanted a second one!” - Dwight Schrute, The Office 3

  4. 256x256 m regions. 4

  5. avatar mobility: who is where, when 5

  6. why do we care? 6

  7. research in systems support for NVE 7

  8. 8

  9. How to partition a world into regions and assign regions to servers considering - communication cost - hand-over rate - balancing server load : 9

  10. 10

  11. How to predict avatar movement (end therefore what a user will see next)? 11

  12. 12

  13. AoI-based scheme 13

  14. How many connections? How stable are the connections? 14

  15. supernode-based scheme 15

  16. How to pick supernodes? How stable are the supernodes? 16

  17. how to simulate avatar mobility? 17

  18. random walk random waypoint clustered movement : 18

  19. or, small-scale implementation 19

  20. no large-scale NVE available until recently 20

  21. 482,594 residents logged in between 2-9 June 2008 21

  22. secondlife.com/whatis/economy-graphs.php 22

  23. • collect mobility traces of avatars in Second Life • what it means w.r.t. systems design for NVEs? 23

  24. collecting traces 24

  25. how do avatars move inside a distributed virtual environment? 25

  26. how are avatars distributed within a region? 26

  27. how long do they stay at a location? 27

  28. do they move in groups? 28

  29. etc. 29

  30. FPS MMORPG NVE 30

  31. Linden, can we get access to the server traces? No. 31

  32. • Wrote our own client • Parses packets using libsecondlife • Insert bots into regions • Log positions of avatars every 10s 32

  33. difficulties 33

  34. running out of memory 34

  35. anti-bots policy 35

  36. over crowded region 36

  37. inter-region tracking 37

  38. • Wrote our own client • Parses packets using libsecondlife • Insert bots into regions • Log positions of avatars every 10s 38

  39. who is where, when (doing what) 39

  40. 40

  41. Freebies 41

  42. The Pharm 42

  43. Isis 43

  44. Ross 44

  45. Mobility Patterns 45

  46. Freebies: number of visits to a cell 46

  47. Freebies: average pause time in a cell 47

  48. Freebies: average speed in a cell 48

  49. Isis: number of visits to a cell 49

  50. caching/prefetching based on popularity of locations? 50

  51. Isis: average pause time in a cell 51

  52. pick supernodes from sticky location? 52

  53. Isis: average speed in a cell 53

  54. mobility model: random walk + pathway ? 54

  55. churn rate 55

  56. 56

  57. 57

  58. Reasonably high churn (up to 6/min) 58

  59. 1 min 10 min 1 hr 2 hr Highly skewed. Some stay for hours. 59

  60. can not pick supernodes uniformly 60

  61. clustering of avatars 61

  62. meeting : encounter between two avatars (within each other AoI) 62

  63. Meet many different avatars. 63

  64. 1 min 1 hr 10 min 2 hr Most meetings are short. 64

  65. Meeting size is large. 65

  66. high overhead in maintaining AoI neighbors 66

  67. meeting stability: avg meeting size over num of avatars met 67

  68. Wide range of stability 68

  69. other tidbits 69

  70. little temporal variations can use historical information to predict future 70

  71. rotate 18% of the time Second Life’s prefetching is wasteful 71

  72. 25-35% revisits the same region in a day region-based caching? 72

  73. proxy-based texture caching 73

  74. why textures? 74

  75. 62 - 81% of traffic are textures 75

  76. 316 MB of textures in Isis 76

  77. 64m 80 o 77

  78. 64m 80 o 78

  79. 403 TB of textures retrieved in Isis in a day 79

  80. clients SL servers texture proxy 80

  81. what caching algorithm to used? 81

  82. 2Q 82

  83. FIFO LRU 83

  84. cache miss FIFO LRU 84

  85. cache hit FIFO LRU 85

  86. cache hit FIFO LRU 86

  87. scan resistant FIFO LRU 87

  88. 88

  89. 3Q 89

  90. FIFO LRU Victim Buffer 90

  91. cache hit FIFO LRU Victim Buffer (sorted by popularity) 91

  92. how to define popularity of texture? 92

  93. Freebies: number of visits to a cell 93

  94. little temporal variations can use historical information to predict future 94

  95. popularity of texture = popularity of cell 95

  96. Per-byte Hit Rate 2Q 3Q OPT Ross 50 MB 0.58 0.62 0.70 Ross 25 MB 0.28 0.36 0.47 Freebies 50 MB 0.48 0.50 0.68 Freebies 25 MB 0.21 0.33 0.50 96

  97. conclusion 97

  98. understanding real avatar mobility is crucial to design good NVEs 98

  99. 謝謝 歡迎發問及指教 99 99

Recommend


More recommend