Neural network-based node mobility and network connectivty predictions for mobile ad hoc radio networks

• US 7,555,468 B2
• Filed: 06/26/2006
• Issued: 06/30/2009
• Est. Priority Date: 06/26/2006
• Status: Active Grant

• Current Assignee: The Boeing Co.
• Original Assignee: The Boeing Co.
• Inventors: El-Damhougy, Hesham
• Primary Examiner(s): Vincent; David R
• Assistant Examiner(s): Fernandez Rivas; Omar F
• Application Number:
  US11/426,419
• Publication Number:
  US 20070299794A1
• Time in Patent Office: 1,100 Days
• Field of Search: 706 14- 19, 706/22, 370254-258, 370/310, 370/316, 370/328, 370/338, 370/351, 370/352, 370/395.21, 370/395.3, 370/395.31, 370/395.32, 455/3.01, 455/431, 455/437, 455440-441, 455/448, 455/449, 455/456.1, 455/456.5, 455/456.6, 455/7, 455/95
• US Class Current: 706/15
• CPC Class Codes: H04W 40/18 based on predicted events
• CPC Class Codes: H04W 40/20 based on geographic position or location
• CPC Class Codes: H04W 40/246 Connectivity information discovery
• CPC Class Codes: H04W 40/32 for defining a routing cluster membership
• CPC Class Codes: H04W 84/18 Self-organising networks, e.g. ad-hoc networks or sensor networks

Abstract

A self managed ad hoc communications network nodes and node mobility management. Nodes include an Artificial Neural Network (ANN) that determines connection to other network nodes. The ANN may use free space propagation link life estimation, inverse modeling for partition prediction, Stochastic Approximation, and/or coarse estimation. The node includes storage storing network tables and matrices indicating network connectivity and connection to other nodes. Also, a wireless communications unit provides for wireless communicating with other nodes.

Claims

1. A node for use in an ad hoc communications network including a plurality of connected said nodes, said node comprising:

an Artificial Neural Network (ANN) element;

a storage unit communicatively coupled to said ANN element; and

a wireless communications unit communicatively coupled to said ANN element, said ANN element being capable of predicting a link life for direct connection to at least one other said node.

2. A node as in claim 1, wherein for each directly connected other said node, said ANN element determines a relative velocity and predicts said link life for each based on the respective distance to each and a given propagation range.

3. A node as in claim 1, wherein said ANN element determines said link life as the time for said node to separate from another said node by said given propagation range.

4. A node as in claim 1, wherein said node includes a locational unit determining a location of said node.

5. A node as in claim 4, wherein said ANN element determines link life to another directly connected node based on the determined said location of said node and a determined location of said other directly connected node and each of said node and said other directly connected node are moving at known velocities.

6. A node as in claim 5, wherein said ANN element determines a propagation loss relationship responsive to locations of said node and another said node and a link life relationship responsive to said known velocities.

7. A node as in claim 6, wherein said ANN element determines an expected loss of a direct connection, said expected loss occurring at a zero of the difference of said link life relationship and a maximum propagation range indicated by said propagation loss relationship.

8. A node as in claim 1, wherein said ANN element further determines a reconnect time for a disconnected one of other said nodes.

9. A node as in claim 1, wherein said node is a cluster leader.

10. A node as in claim 1, wherein said node is a Network Management System (NMS) node managing said ad hoc communications network.

11. A node as in claim 1, further comprising an input/output (I/O) unit, a node user communicating with at least one other said node through said I/O unit.

12. A node as in claim 1, wherein said ANN element is an ANN neuron determining connection to other said nodes in said ad hoc communications network, said ANN neuron storing network tables and matrices indicating network connectivity and connectivity to said other nodes and communicating wirelessly with said other nodes through said wireless communications unit.

13. An ad hoc communications network comprising a plurality of nodes as in claim 12, wherein at least one of said plurality of connected nodes is a wireless mobile node.

14. An ad hoc communications network comprising:

a plurality of nodes comprising an Artificial Neural Network (ANN), each node in communication with other said nodes of said ad hoc communications network;

a plurality of network clusters, each cluster including one or more nodes of said plurality of nodes, one node in each cluster determining link life for connections for other nodes in said cluster and reconnect times for previously disconnected nodes;

a storage unit in each said node, said node storing network tables and matrices indicating network connectivity, cluster and connectivity connection between said one or more nodes; and

a wireless communications unit in each said node, said node communicating with at least one other said node wirelessly through said wireless communications unit.

15. An ad hoc communications network as in claim 14, wherein one of the plurality of said nodes is a Network Management System (NMS) node managing said ad hoc communications network and at least one other said nodes of the plurality of said nodes are cluster leaders managing communications between a respective cluster and other clusters.

16. An ad hoc communications network as in claim 14, wherein at least one said node determines a relative velocity connection to at least one said node and predicts a link life for each one based on the respective distance from each said node relative to a given propagation range.

17. An ad hoc communications network as in claim 16, wherein said one node determines said link life as the time to separate two nodes from each other by said given propagation range.

18. An ad hoc communications network as in claim 14, wherein each said node of the plurality of said nodes include a locational unit determining a location of each said node.

19. An ad hoc communications network as in claim 18, wherein at least one said node predicts said link life based on the determined said location of a first node and a determined location of second directly connected node and each of said first node and said second directly connected node are moving at known velocities.

20. An ad hoc communications network as in claim 19, wherein at least one said node determines a propagation loss relationship responsive to locations of said first node and said second directly connected node and a link life relationship responsive to said known velocities.

21. An ad hoc communications network as in claim 19, wherein at least one said node determines an expected loss of a direct connection, said expected loss occurring at a zero of the difference of said link life relationship and a maximum propagation range indicated by said propagation loss relationship.

22. An ad hoc communications network as in claim 14, wherein at least one said node includes a stochastic approximation engine (SA Engine) receiving information indicating propagation losses between a pair of directly connected nodes.

23. An ad hoc communications network as in claim 22, wherein received propagation loss information includes measured and calculated propagation loss information having both a deterministic component and a random component (ε

) of the form d(x_m,y_m)+ε

_m.

24. An ad hoc communications network as in claim 23, wherein said SA Engine recursively predicts said link life from a relationship having the form D₊₁=D_n−

α

_n{d(x_n,y_n)−

ρ

}, where;

ρ

is a propagation loss threshold

n=0, 1, 2, . . . ;

α

_nis a decreasing sequence of real numbers satisfying;

α

_n→

0, and

Σ

α

_n→

∞

;

d(x_n,y_n)=d(x_n(D_n),y_n(D_n)); and

D_n→

D=ρ

, the spacing between a directly connected pair of nodes at said propagation loss threshold.

25. An ad hoc communications network as in claim 23, wherein said ANN element uses SA Engine in recursively predicting said link life (T) for a first node (x_n) and second node (y_n) located at x_n(T_n), y_n(T_n) from a relationship position based time sequence having the form T_n+1=T_n−

α

_n{d(x_n,y_n)-ρ

}, where;

ρ

is a propagation loss threshold

n=0, 1, 2, . . . ;

α

_n is a decreasing sequence of real numbers satisfying;

α

_n→

0, and

Σ

α

_n→

∞

;

d(x_n,y_n)=d(x_n(T_n),y_n(T_n)); and

T_n→

T as n→

∞

, the link life between a first said node and said second node at said propagation loss threshold.

26. An ad hoc communications network as in claim 14, wherein at least one of the plurality of said nodes further comprise an input/output (I/O) unit, a node user communicating with said other nodes through said I/O unit.

27. A mobile communications station, said mobile communications station being mobile and comprising:

a wireless communications unit;

an Artificial Neural Network (ANN) element connectively coupled to said wireless communications unit, said ANN element being capable of predicting a link life for direct connection to at least one other said mobile communications station; and

a storage unit connectively coupled to said ANN element.

28. A mobile communications station as in claim 27, further comprising a locational unit connectively coupled to said ANN element.

29. A mobile communications station as in claim 28, wherein said locational unit is a Global Positioning Systems (GPS) unit.

30. A mobile communications station as in claim 27, further comprising an input/output (I/O) unit connectively coupled to said ANN element.

31. A mobile communications station as in claim 27, further comprising a stochastic approximation engine (SA Engine) measuring propagation loss with another directly connected mobile communications station and training said ANN element.

32. A mobile communications station as in claim 27, wherein said storage unit having network tables and matrices indicating network connectivity and connection to other mobile communication stations.

33. A mobile communications station as in claim 27, further comprising an unmanned ground vehicle, said wireless communications unit, said ANN element and said storage unit being transported by said unmanned ground vehicle.

34. A mobile communications station as in claim 27, further comprising an unmanned air vehicle, said wireless communications unit, said ANN element and said storage unit being transported by said unmanned air vehicle.

35. A mobile communications station as in claim 27, wherein said ANN element is an ANN neuron determining connection to other said nodes in said ad hoc communications network, said ANN neuron storing network tables and matrices indicating network connectivity and connectivity to said other nodes and communicating wirelessly with said other nodes through said wireless communications unit.

36. A node for use in an ad hoc communications network including a plurality of the nodes, said node comprising:

an Artificial Neural Network (ANN) element;

a stochastic approximation engine (SA Engine) measuring propagation loss with another directly connected mobile communications station and training said ANN element;

a storage unit communicatively coupled to said ANN element; and

a wireless communications unit communicatively coupled to said ANN element.

37. A node as in claim 36, wherein said SA Engine provides measurement results having both a deterministic component and a random component (ε

) and the form d(x_m, y_m)+ε

_m.

38. A node as in claim 37, wherein said ANN element recursively determines said link life from a relationship having the form D_n+1=D_n−

α

_n{d(x_n, y_n)−

ρ

}, where;

ρ

is a propagation loss threshold

n=0, 1, 2, . . . ;

α

_n is a decreasing sequence of real numbers satisfying;

α

_n→

0, and

Σ

α

_n→

∞

;

d(x_n,y_n)=d(x_n(D_n),y_n(D_n)); and

D_n→

D=ρ

, the spacing between said node and said other directly connected node at said propagation loss threshold.

39. A node as in claim 37, wherein said ANN element recursively determines said link life (T) for x_N and y_n located at x_n(T_n), y_n(T_n) from a relationship position based time sequence having the form T_n+1=T_n−

α

_n{d(x_n, y_n)-ρ

}, where;

ρ

is a propagation loss threshold

n=0, 1, 2,

α

_n is a decreasing sequence of real numbers satisfying;

α

_n→

0, and

Σ

α

_n→

∞

;

d(x_n,y_n)=d(x_n(T_n),y_n(T_n)); and

T_n→

T as n→

∞

, the link life between said node (x_n) and said other directly connected node (y_n) at said propagation loss threshold.