Generic distributed processing for multi-agent systems

• US 8,898,218 B2
• Filed: 02/03/2014
• Issued: 11/25/2014
• Est. Priority Date: 02/01/2013
• Status: Active Grant

• Current Assignee: Introspective Power, Inc.
• Original Assignee: Introspective Power, Inc.
• Inventors: Thompson, Anthony Scott
• Primary Examiner(s): JEAN, FRANTZ B
• Application Number:
  US14/171,143
• Publication Number:
  US 20140222886A1
• Time in Patent Office: 295 Days
• Field of Search: 709/201, 709/203, 709/202, 709/248, 706 10- 12, 706 14- 16, 706/61, 719/318
• US Class Current: 709/201
• CPC Class Codes: G06F 9/541 via adapters, e.g. between incompatible applications
• CPC Class Codes: G06F 9/542 Event management; Broadcasting; Multicasting; Notifications
• 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 67/51 Discovery or management thereof, e.g. service location protocol [SLP] or web services

Abstract

Generic Distributed Processing Unit (DPU) for Multi-Agent Systems (MAS) provides a Machine to Machine (M2M) interface that is fast, flexible, redundant and scalable. It allows the handling of unlikely edge cases that Human Machine Interfaces (HMI) cannot. It also allows the processing of excessive amounts of ancillary data that are not processed easily with an HMI arrangement. In the digital ecosystem, any like DPU can back up any other, making the system exceedingly robust.

Claims

1. A distributed control system providing a unifying reactive intelligence for new and legacy agents, the system comprising:

N Distributed Processing Units each comprising a processor providing an associated event handler configured to receive and process events having a common system event format and to generate commands having a common system command format;

distributed databases synchronized with one another using a synchronous data flow, each of the N Distributed Processing Units being configured to access at least one of the distributed databases;

plural agents that generate agent specific events, the format of agent specific events enerated b at least one of the lural a ents differin from the format of a ent s ecific events generated by at least another of the plural agents, the agents being configured to discover at least one of the N Distributed Processering Units; and

event communication protocol abstraction structure comprising;

a transformer connected to receive generated agent specific events, the transformer being configured to transform the received agent specific events into the common system event format;

an event dispatcher connected to receive the common system event formats, the event dispatcher configured to dispatch the events in the common system event formats to the event handler of the discovered Distributed Processing Unit; and

a commander connected to receive commands an event handler generates in the common system command format, the commander transforming commands from the common system command format into agent specific commands for sending to one or more of the agents, for action and response.

2. The system of claim 1 wherein the agents discover Distributed Processing Units by (a) setting up intentional race conditions using a broadcast request for services, and (b) choosing the first Distributed Processing Unit to respond to said broadcast request with only a single round trip response, the responding Distributed Processing Unit with the lowest latency response being selected using zero compute cycles.

3. The system of claim 1 wherein the plural agents discover a Distributed Processing Unit by broadcasting a message setting up an intentional race between the N Distributed Processing Units and selecting a first of the N Distributed Processing Units to respond to the broadcasted message.

4. A control process providing a unifying reactive intelligence for new and legacy agents, the process comprising:

sending a request to N Distributed Processing Units, where integer N>

1, each of the N Distributed Processing Units comprising a processor providing an associated event handler;

with the event handlers of the N Distributed Processing Units, events having a common system event format and generating commands having a common system command format;

synchronizing distributed databases using a synchronous data flow, each of the N Distributed Processing Units being configured to access at least one of the distributed databases;

discovering at least one of the N Distributed Processing Units, including using plural agents to generate agent specific events, wherein the format of agent specific events generated by at least one of the plural agents differs from the format of agent specific events generated by at least another of the plural agents, wherein discovering includes;

determining which of the N Distributed Processing Units respond sooner, and

selecting, sooner-responding of the Distributed Processing Unit(s);

transforming, with a transformer connected to receive generated agent specific events, the received agent specific events into the common system event format; and

dispatching, with an event dispatcher, the events in the common system event formats to the event handler of the discovered at least one Distributed Processing Unit; and

with a commander, receiving commands an event handler generates in the common system command format, and transforming commands from the common system command format into agent specific commands for sending to one or more of the agents for action and response.

5. The process of claim 4 wherein the determining measures latency of an intentional race condition between at least some of the N Distributed Processing Units.

6. The process of claim 4 further including establishing a connection and associated control relationship with the selected Distributed Processing Unit(s) and repeating the method upon detection that the established connection is broken.

7. The process of claim 4 further including:

discovering one or more appropriate agents to receive commands and forwarding the agent commands to the discovered one or more appropriate agents; and

checking response(s) and sending more commands if needed.

8. The method of claim 4 further including delaying response based on resource availability.

9. The method of claim 8 wherein the resource comprises at least one of processing time, memory and network bandwidth.

10. A distributed control system providing a unifying reactive intelligence comprising:

N Distributed Processing Units, N>

1, the N Distributed Processing Units each comprising an associated event handler configured to receive and process events having a common system event format and to generate commands having a common system command format;

a network coupled to the N Distributed Processing Units;

distributed databases synchronized with one another using a synchronous data flow, each of the N Distributed Processing Units being configured to access at least one of the distributed databases;

plural agents that generate agent specific events, the format of agent specific events generated by at least one of the plural agents differing from the format of agent specific events generated by at least another of the plural agents, the agents being configured to discover at least one of the N Distributed Processing Units by sending messages to the N Distributed Processing Units over the network and determining, based on replies thereto, which of the N distributed processing units responds first or is otherwise available; and

event communication protocol abstraction structure comprising;

a transformer connected to receive generated agent specific events, the transformer being configured to transform the received agent specific events into the common system event format,

an event dispatcher connected to receive the common system event formats, the event dispatcher configured to dispatch the events in the common system event formats to the event handler of the discovered at least one Distributed Processing Unit; and

a commander connected to receive commands an event handler generates in the common system command format, the commander transforming commands from the common system command format into agent specific commands for sending to one or more of the agents for action and response.

11. The system of claim 10 wherein the databases comprise N databases each associated with a different one of the N distributed processing units, and the system further comprises a communication link for synchronizing all of the N databases.

12. The system of claim 10 wherein the databases comprise databases numbering in X associated with the N Distributed Processing Units where N>

=X and the system further comprises, for cases where X!=1, a communication link for synchronizing all of the X number of databases.

13. The system of claim 10 wherein the agents determines the winners of intentional race conditions between the N Distributed Processing Units.

14. The system of claim 10 wherein the agent establishes a connection with the at least one discovered Distributed Processing Unit and repeats discovery upon detection that the established connection is broken.

15. The system of claim 10 wherein the system is configured to:

discover one or more appropriate agents to receive the commands and forward the agent commands to one or more agents; and

check response and sending more commands if needed.

16. The system of claim 10 wherein at least some of the N Distributed Processing Units delay responses based on a predetermined condition.

17. The system of claim 16 wherein the predetermined condition is selected from the group consisting of available processing time, available memory, and available bandwidth.