Processing event data streams to recognize event patterns, with conditional query instance shifting for load balancing

• US 9,667,549 B2
• Filed: 11/27/2012
• Issued: 05/30/2017
• Est. Priority Date: 11/29/2011
• Status: Active Grant

• Current Assignee: AGT International GMBH
• Original Assignee: AGT International GMBH
• Inventors: Ziekow, Holger
• Primary Examiner(s): Meky, Moustafa M
• Application Number:
  US14/361,339
• Publication Number:
  US 20140317287A1
• Time in Patent Office: 1,645 Days
• CPC Class Codes: G08G 1/0133 for classifying traffic situation
• CPC Class Codes: H04L 41/145 involving simulating, designing, planning or modelling of a network
• CPC Class Codes: H04L 47/125 by balancing the load, e.g. traffic engineering

Abstract

A computer (100) recognizes an event pattern (ABC) for objects (14, 24, 34, 44) that belong to an event domain (150). The computer activates a pattern query (110) that corresponds to the event pattern (ABC) and that has a least one state (S4) with an distinctive transition probability to a final state. The probability is derived from object observations in the event domain (150). The computer continuously receives event representations (*A14, *A24, *A44, *D44, . . . ) that are related to the objects (14, 24, 34, 44) and allocates the event representations to a first processing resource (101) to initiate instances (1.14, 1.24, 1.34, 1.44) of the query (110). It monitors the instances and, upon receiving event representations that cause the instances to reach the state (S4) with the distinct transition probability, it shifts the instances to a second processing resource (102).

Claims

1. A computer-implemented method for recognizing an event pattern for objects that belong to an event domain, the method comprising:

activating a pattern query that corresponds to the event pattern and that has at least one state with a distinctive transition probability to reach a final query state, said distinctive transition probability representing a lower transition probability of said at least one state to reach said final state, said lower probability is lower compared to at least one other transition probability to reach said final state of at least one second state, other than said at least one state, the probability being derived from object observations in the event domain;

wherein an instance reaching said distinctive transition probability state uses a longer time interval than an instance not reaching said distinctive transition probability state;

continuously receiving event representations, that are related to the objects;

allocating the event representations to a first processing resource to initiate instances of the query;

monitoring the instances and, upon receiving event representations that cause an instance of said instances to reach the state with the distinctive transition probability, shifting the instance to a second processing resource that is more optimized for handling longer time interval than said first processing resource.

2. The method of claim 1, wherein the query is activated such that the state with the distinctive transition probability is not associated with the event pattern.

3. The method of claim 1, wherein

in activating, the query has states and event-driven state transitions, wherein transition probabilities to the state classify the states into a first group of states and a second group of states, the transition probabilities being derived from object observations in the event domain, and the second group having the at least one state with the distinctive state transition probability;

in receiving event representations, the event representations are being received with object identifiers;

in allocating, the event representations to the first processing resource, the event representations are being allocated separately for different object identifiers;

in monitoring, the instances are being monitored according to the classification of the states, so that instances that reach states of the first group, continue to be executed by the first processing resource.

4. The method of claim 3, wherein monitoring comprises to monitor the shifted instances according to the classification of the states, for shifted instances that reach states of the first group, re-shifting the previously shifted instances to the first processing resource.

5. The method of claim 3, wherein the query is activated in combination with an event relation model in that most of the transition probabilities for a state that belongs to the first group to a state that belongs to the second group are different than the transition probabilities for the state that belongs to the second group to the states that belongs to the first group.

6. The method of claim 1, wherein such method is applied to objects that interact with sensors and the event representations are received from the sensors.

7. The method of claim 1, wherein the sensors submit representations of the event time.

8. The method of claim 1, wherein executing the instances stops upon reaching a maximum time (Tmax).

9. The method of claim 1, wherein receiving event representations comprises to re-calculate transition probabilities and re-classifying the states into groups.

10. The method according to claim 9, wherein monitoring comprisesto use the re-classified states.

11. The method of claim 1, wherein, a computer program product that, when loaded into a non-transitory computer readable memory of a computing device and being executed by at least one processor of the computing device, performs the steps of the computer-implemented method.

12. The method of claim 1, wherein said pattern query has at least one third state with a second distinctive transition probability to reach a final query state, said second distinctive transition probability representing a higher transition probability of said at least one third state to reach said final state, said higher probability is higher compared to at least one other transition probability to reach said final state of at least one fourth state, other than said at least third state, wherein an instance reaching said second distinctive transition probability state uses a shorter time interval than an instance not reaching said second distinctive transition probability state, and

wherein upon receiving event representations that cause a second instance to reach the third state with the second distinctive transition probability, shifting the second instance to a third processing resource, that is more optimized for handling shorter time interval than said first processing resource.

13. A computer for recognizing an event pattern for objects that belong to an event domain, the computer comprising:

a pattern query activator that corresponds to the event pattern and that activates a pattern query that has at least one state with a distinctive transition probability to reach a final state of the query, said distinctive transition probability representing a lower transition probability of said at least one state to reach said final state, said lower probability is lower compared to at least one other transition probability to reach said final state of at least one second state, other than said at least one state, the probability being derived from object observations in the event domain, wherein an instance reaching said distinctive transition probability state uses a longer time interval than an instances not reaching said distinctive transition probability state;

a receiver that continuously receives event representations that are related to the objects;

an allocator that allocates the event representations to a first processing resource of the computer to initiate instances of the query; and

a monitor that monitors the instances and, upon receiving event representations that causes an instance of said instances to reach the state with the distinctive transition probability, shifts the instance to a second processing resource of the computer that is more optimized for handling longer time interval than said first processing resource.

14. The computer according to claim 13, wherein the pattern query activator activates a query in that the state with the distinctive transition probability is not associated with the event pattern.

15. The computer according to claim 13, wherein

the pattern query activator uses queries with states and event-driven state transitions, wherein transition probabilities classify the states into a first group of states and a second group of states, the transition probabilities being derived from object observations in the event domain, and the second group having the at least one state with the distinctive state transition probability;

the receiver receives the event representations with object identifiers;

the allocator allocates the event representations separately for different object identifiers; and

the monitor monitors the instances according to the classification of the states, so that instances that reach states of the first group continue to be executed by the first processing resource.

16. The computer according to claim 13, wherein the receiver re-calculates transition probabilities and re-classifies the states into groups.

17. The computer according to claim 13, wherein said pattern query has at least one third state with a second distinctive transition probability to reach a final query state, said second distinctive transition probability representing a higher transition probability of said at least one third state to reach said final state, said higher probability is higher compared to at least one other transition probability to reach said final state of at least one fourth state, other than said at least third state, wherein an instance reaching said second distinctive transition probability state uses a shorter time interval than an instance not reaching said second distinctive transition probability state, and

wherein upon receiving event representations that cause a second instance to reach the third state with the second distinctive transition probability, shifting the second instance to a third processing resource, that is more optimized for handling shorter time interval than said first processing resource.