Network processing

• US 20090141658A1
• Filed: 03/23/2007
• Published: 06/04/2009
• Est. Priority Date: 03/24/2006
• Status: Active Grant

• Current Assignee: BAE Systems Plc
• Original Assignee: BAE Systems Plc
• Inventors: Day, Peter Stephen, Bladon, Peter
• Granted Patent: US 7,933,223 B2
• Time in Patent Office: Days
• US Class Current: 370/256
• CPC Class Codes: H04L 67/10 in which an application is distributed across nodes in the network software deployment G06F8/60; multiprogramming arrangements G06F9/46
• CPC Class Codes: H04L 67/104 Peer-to-peer [P2P] networks
• CPC Class Codes: H04L 67/125 involving control of end-device applications over a network

Abstract

A method and means for processing information across a network is provided, the network comprising a plurality of nodes. In one embodiment, the invention includes algorithm means for addressing the problem of efficient information distribution between nodes in the network. This invention addresses the problem by generating probabilistic models at the nodes in the network and by selectively distributing information between the nodes based on these models and by using a predetermined metric function in such a way as to reduce information differences between the nodes.

Advantageously this invention finds utility in many applications in both distributed networks and centralised networks provided that it is possible to derive an appropriate tree topology representation of the network.

Claims

1. A method of decision making that is applied in a network comprising a plurality of nodes, each node having means for receiving and sending information and means for processing information, and each node being linked to selected other nodes of the network, the method comprising:

(i) providing a tree representation of the nodes and establishing therefrom for each of the nodes (a) a node model comprising a probability distribution of node component probability factors to represent the node state and (b) at least one node link model comprising a probability distribution of link component probability factors to represent the state of each node link associated with each node via which information can be communicated to selected other nodes,

(ii) in response to receiving new information at one or more of the nodes, updating the corresponding node models that are maintained at said one or more of the nodes,

(iii) in response to detecting the availability of a node link which links a node pair in the network, selecting a combination of node component probability factors and link component probability factors associated with said node pair and its node link by calculating the contributions that various possible combinations of node component probability factors and link component probability factors would make to a predetermined function which measures the difference between the node models and the node link models for said node pair and its node link and by selecting that combination of probability factors amongst the various possible combinations of probability factors which provides the largest calculated contribution to said predetermined function,

(iv) transferring said selected combination of node component probability factors and link component probability factors from the originating node to the receiving node via said node link thereby reducing information differences between the nodes and updating the corresponding node models and the node link models that are maintained at the nodes, and

(v) independently reconciling the node models and the node link models associated with the nodes in the network so as to take account of the transferring step (iv).

2. A method as claimed in claim 1 including a step (vi) of repeating the aforesaid steps (i), (ii), (iii), (iv) and (v) in a series of time steps as an iterative procedure.

3. A method as claimed in claim 1, wherein said step (ii) and said steps (iii) to (v) are effected in accordance with different procedures asynchronously.

4. A method as claimed in claim 1, wherein a plurality of node links are provided at a first node and single node links are provided at a second and third node, the first node being linked separately to the second and third nodes via node links, said steps (iii) to (v) being further effected contemporaneously for each of the node links and associated node link models.

5. A method as claimed in claim 1, wherein said predetermined function is the Kullback-Leibler (KL) divergence metric function.

6. A method as claimed in claim 1, comprising a step of basing the node models and the node link models on a Bayesian network probabilistic model.

7. A method as claimed in claim 1, comprising a step of basing the node models and the node link models on a Markov Random Field probabilistic model.

8. A method as claimed in claim 1 implemented over a lossless TCP network protocol.

9. A computer program comprising program code means for performing the method steps of claim 1 when the program is run on a computer.

10. A computer program product comprising program code means stored on a computer readable medium for performing the method steps of claim 1 when the program is run on a computer.

12. An autonomous system for gathering and controlling the distribution of information comprising a plurality of platforms distributed in space adapted and arranged to carry out a method as claimed in claim 1.

13. The method as claimed in claim 2 wherein said step (ii) and said steps (iii) to (v) are effected in accordance with different procedures asynchronously.

14. The method as claimed in claim 3 wherein a plurality of node links are provided at a first node and single node links are provided at a second and third node, the first node being linked separately to the second and third nodes via node links, said steps (iii) to (v) being further effected contemporaneously for each of the node links and associated node link models.

15. The method as claimed in claim 4 wherein said predetermined function is the Kullback-Leibler (KL) divergence metric function.

16. The method as claimed in claim 5, comprising a step of basing the node models and the node link models on a Bayesian network probabilistic model.

17. The method as claimed in claim 5, comprising a step of basing the node models and the node link models on a Markov Random Field probabilistic model.

18. The method as claimed in claim 7 implemented over a lossless TCP network protocol.

19. A computer program comprising program code means for performing the method steps of claim 8 when the program is run on a computer.

20. A computer program product comprising program code means stored on a computer readable medium for performing the method steps of claim 8 when the program is run on a computer.

21. An autonomous system for gathering and controlling the distribution of information comprising a plurality of platforms distributed in space adapted and arranged to carry out a method as claimed in claim 10.

22. A system for decision making in a network comprising a plurality of nodes, each node having means for receiving and sending information and means for processing information, and each node being linked to selected other nodes of the network, the system performing the steps of the method of claim 1.

11. (canceled)