Real Time Estimation for Usage-Dependent Reliability on a Dual-Backbone Network Subsystem Meng-Lai Yin Lawrence E. James Rafael R. Arellano Raymond V. Hettwer

1
Real Time Estimation for Usage-Dependent
Reliability on a Dual-Backbone Network
SubsystemMeng-Lai Yin Lawrence E.
JamesRafael R. Arellano Raymond V. Hettwer
2
OUTLINE

• Overview of the system
• Modeling strategy
• Detailed analysis
• Summary

3
OverviewSatellite Navigation System
4
Overview Ground Segment Network Subsystem

5
Real Time Modeling User 1
6
Real Time Modeling User 2
7
Modeling Strategy

• To connect k sites, the problem is a k-terminal
  network reliability problem, which in general is
  an NP-Hard problem.
• Independence assumption can be made between the
  two ring-structured backbone networks and the
  access networks. The hierarchical decomposition
  method is then applied.
• By preprocessing static data, and efficiently
  reference those data during run time, we can
  obtain the dynamic network reliability
  efficiently.

8
Hierarchical Decomposition
System
Space
Ground
Network
Network 2
Network 1
Access Network 1
Backbone Network 1
Access Network 2
Backbone Network 2
9
Real Time Modeling Strategy

• Preprocess network reliability for different
  network requirements
• During run time, obtain the corresponding network
  reliability according to the network requirements
  based on user position and time
• Feed network reliability into the ground segment
  model and obtain the overall system reliability
  in real time

10
Hierarchical Decomposition
Usage dependent
11
Topology of the Network Subsystem
12
Reduced Graph
13
Mapping on the 4-node Ring Network
14
Real Time Network Reliability Estimation

• Preprocessing
• for all Ai
• estimate PAi
• for all mapping types
• estimate Pbackbone network 1(Pbackbone
  network 2)
• Real-Time Processing
• for each user
• mapping required sites onto backbone
  networks
• obtain corresponding Pbackbone network 1and
  Pbackbone network 2
• / from
  preprocessed data
• compute PN1 and PN2
• compute Pnetwork

O(nm)
O(m)
Total time complexity O(nm), compared to
O(2nm) n number of site nodes m number of
backbone nodes
15
SUMMARY

• An efficient method for real time network
  reliability estimation is presented
• Preprocessing is the key to efficiency
• The hierarchical decomposition method and
  several graph simplification techniques are used

16
Future Research

• Apply graph similarity to network reliability
  estimation other than the ring topology
• Efficient methods on usage-dependent reliability
  estimation for satellite navigation systems