Presentation is loading. Please wait.

Presentation is loading. Please wait.

Smart Networks Project UC Berkeley / Cisco May 27, 2003.

Similar presentations


Presentation on theme: "Smart Networks Project UC Berkeley / Cisco May 27, 2003."— Presentation transcript:

1 Smart Networks Project UC Berkeley / Cisco May 27, 2003

2 UCB - Smart Networks Project2 May 27, 2003 Team CISCO  David Jaffe UCB  Venkat Anantharam  David Tse  Pravin Varaiya  Jean Walrand COLLABORATION WITH  UIUC (J. Hou)  Georgia Tech (R. Fujimoto)  UCLA (R. Bagrodia)  Eric Chi  Antonios Dimakis  Linhai He  Bill Hodge  Zhanfeng Jia  John Musacchio  Jun Shu  Wilson So  Teresa Tung  Undergraduate Students

3 UCB - Smart Networks Project3 May 27, 2003 Outline Goals Approach Illustrations  Routing  Clustering  Channel Allocation Contributions Future Work

4 UCB - Smart Networks Project4 May 27, 2003 Goals Scalable & Robust QoS Routing in MANET Scalable: Hundreds of Nodes Robust: Fast rerouting as network changes QoS: Urgent Messages (CFF), Urgent Flows (Voice), Expedited Flows (Video Streams), BE Video Voice Video

5 UCB - Smart Networks Project5 May 27, 2003 Approach Hierarchical: for scalability Routing: two levels Clustering: randomized algorithm Channel allocation: simulated annealing

6 UCB - Smart Networks Project6 May 27, 2003 Approach Hierarchical

7 UCB - Smart Networks Project7 May 27, 2003 Approach Dynamic Clustering Gateways

8 UCB - Smart Networks Project8 May 27, 2003 Benefits of Clustering Decomposes problem Simplifies tables Limit change announcements Reduces control messages Acceptable sub-optimality

9 UCB - Smart Networks Project9 May 27, 2003 Approach: Routing Decomposition Decomposed into intra-cluster and inter-cluster Source Dest

10 UCB - Smart Networks Project10 May 27, 2003 Approach: Dynamic Clustering Distributed randomized algorithm Limits cut-sets Targets a cluster size

11 UCB - Smart Networks Project11 May 27, 2003 Approach: Channel Allocation Objective: Maximize capacity while maintaining connectivity Approach: Simulated annealing

12 UCB - Smart Networks Project12 May 27, 2003 Illustration: Routing Routing  Hierarchical link-state routing protocol  Dynamic load balancing  Considers wireless channel sharing effects  More in the demo …

13 UCB - Smart Networks Project13 May 27, 2003

14 UCB - Smart Networks Project14 May 27, 2003

15 UCB - Smart Networks Project15 May 27, 2003

16 UCB - Smart Networks Project16 May 27, 2003 Illustration: Clustering Clustering  An example with 100 nodes

17 UCB - Smart Networks Project17 May 27, 2003 + : node w/ longer range radio. : node w/ shorter range radio : gateway nodes of a cluster 50km Clusters Shorter range radio: 3.5km

18 UCB - Smart Networks Project18 May 27, 2003

19 UCB - Smart Networks Project19 May 27, 2003

20 UCB - Smart Networks Project20 May 27, 2003

21 UCB - Smart Networks Project21 May 27, 2003

22 UCB - Smart Networks Project22 May 27, 2003

23 UCB - Smart Networks Project23 May 27, 2003

24 UCB - Smart Networks Project24 May 27, 2003

25 UCB - Smart Networks Project25 May 27, 2003

26 UCB - Smart Networks Project26 May 27, 2003

27 UCB - Smart Networks Project27 May 27, 2003

28 UCB - Smart Networks Project28 May 27, 2003 Illustration: Channel Allocation Channel Allocation  Example 1: Single cluster; each node has 1 radio.  Example 2: 4 clusters; gateways have 2 radios, others have 1.  More in the demo …

29 UCB - Smart Networks Project29 May 27, 2003

30 UCB - Smart Networks Project30 May 27, 2003

31 UCB - Smart Networks Project31 May 27, 2003

32 UCB - Smart Networks Project32 May 27, 2003

33 UCB - Smart Networks Project33 May 27, 2003

34 UCB - Smart Networks Project34 May 27, 2003

35 UCB - Smart Networks Project35 May 27, 2003

36 UCB - Smart Networks Project36 May 27, 2003

37 UCB - Smart Networks Project37 May 27, 2003

38 UCB - Smart Networks Project38 May 27, 2003

39 UCB - Smart Networks Project39 May 27, 2003

40 UCB - Smart Networks Project40 May 27, 2003

41 UCB - Smart Networks Project41 May 27, 2003

42 UCB - Smart Networks Project42 May 27, 2003 Channel Allocation (4 clusters example)

43 UCB - Smart Networks Project43 May 27, 2003

44 UCB - Smart Networks Project44 May 27, 2003

45 UCB - Smart Networks Project45 May 27, 2003

46 UCB - Smart Networks Project46 May 27, 2003

47 UCB - Smart Networks Project47 May 27, 2003

48 UCB - Smart Networks Project48 May 27, 2003

49 UCB - Smart Networks Project49 May 27, 2003

50 UCB - Smart Networks Project50 May 27, 2003

51 UCB - Smart Networks Project51 May 27, 2003

52 UCB - Smart Networks Project52 May 27, 2003

53 UCB - Smart Networks Project53 May 27, 2003

54 UCB - Smart Networks Project54 May 27, 2003

55 UCB - Smart Networks Project55 May 27, 2003

56 UCB - Smart Networks Project56 May 27, 2003

57 UCB - Smart Networks Project57 May 27, 2003

58 UCB - Smart Networks Project58 May 27, 2003

59 UCB - Smart Networks Project59 May 27, 2003

60 UCB - Smart Networks Project60 May 27, 2003

61 UCB - Smart Networks Project61 May 27, 2003

62 UCB - Smart Networks Project62 May 27, 2003

63 UCB - Smart Networks Project63 May 27, 2003

64 UCB - Smart Networks Project64 May 27, 2003

65 UCB - Smart Networks Project65 May 27, 2003

66 UCB - Smart Networks Project66 May 27, 2003

67 UCB - Smart Networks Project67 May 27, 2003

68 UCB - Smart Networks Project68 May 27, 2003 Contributions New Algorithms  Clustering: Randomized Algorithm  Channel Allocation: Simulated Annealing with a suitable potential function  Routing: New structure of constraints Evaluation and Testing Environment  Simulates Mobility  Models Flows  Implements Algorithms  Evaluates Performance

69 UCB - Smart Networks Project69 May 27, 2003 Future Work Fundamental Problems:  Packet-based Channel Allocation  QoS through bandwidth limits  Fast rerouting of multiple flows  Proof of correctness and convergence  Model of TDM radios  Traffic sensitive channel allocation


Download ppt "Smart Networks Project UC Berkeley / Cisco May 27, 2003."

Similar presentations


Ads by Google