DECENTRALIZED SYNCHRONIZATION OF MULTIPLE AGENTS

• US 20190326916A1
• Filed: 06/28/2019
• Published: 10/24/2019
• Est. Priority Date: 06/28/2019
• Status: Active Grant

• Current Assignee: Intel Corporation
• Original Assignee: Intel Corporation
• Inventors: Parra Vilchis, Jose I., Guzman Leguel, Anthony K., Gutierrez, David Gomez, Campos Macias, Leobardo E., Gonzalez, Rafael de la Guardia, Lopez, Rodrigo Aldana
• Granted Patent: US 10,763,868 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 1/10 Distribution of clock signals , e.g. skew
• CPC Class Codes: H03L 7/00 Automatic control of frequency or phase; Synchronisation
• CPC Class Codes: H03L 7/0991 the oscillator being a digital oscillator, e.g. composed of a fixed oscillator followed by a variable frequency divider H03L7/0995 takes precedence; fixed oscillators with means for selecting among various phases H03L7/0814
• CPC Class Codes: H04J 3/0697 Synchronisation in a packet node
• CPC Class Codes: H04L 67/1059 Inter-group management mechanisms, e.g. splitting, merging or interconnection of groups
• CPC Class Codes: H04L 69/40 for recovering from a failure of a protocol instance or entity, e.g. service redundancy protocols, protocol state redundancy or protocol service redirection management of faults, events, alarms or notifications in data switching networks H04L41/06

Abstract

In one embodiment, a distributed oscillator computes a first error value based on a first state value for the distributed oscillator and first state values for other distributed oscillators, and computes a second error value based on a second state value for the distributed oscillator and second state values for the other distributed oscillators. The distributed oscillator computes a new first state value for the distributed oscillator based on the first error value, the first state value for the distributed oscillator, and the second state value for the distributed oscillator, and computes a new second state value for the distributed oscillator based on the second error value, the first state value for the distributed oscillator, and the second state value for the distributed oscillator. The distributed oscillator transmits the new first state value and the new second state value to the other distributed oscillators.

Claims

1. A distributed oscillator apparatus comprising:

receiver circuitry to;

obtain first and second state values for one or more other distributed oscillators;

compute a first error value based on a first state value for the distributed oscillator apparatus and the first state values for the other distributed oscillators; and

compute a second error value based on a second state value for the distributed oscillator apparatus and the second state values for the other distributed oscillators;

a first oscillator circuit to compute a new first state value for the distributed oscillator apparatus based on the first error value, the first state value for the distributed oscillator apparatus, and the second state value for the distributed oscillator apparatus;

a second oscillator circuit to compute a new second state value for the distributed oscillator based on the second error value, the first state value for the distributed oscillator apparatus, and the second state value for the distributed oscillator apparatus; and

transmitter circuitry coupled to the first and second oscillator circuits to transmit the new first state value and the new second state value to the other distributed oscillators.

2. The distributed oscillator apparatus of claim 1, wherein:

the first error value is computed based on respective differences between the first state value for the distributed oscillator apparatus and each of the one or more first state values for the other distributed oscillators; and

the second error value is computed based on respective differences between the second state value for the distributed oscillator apparatus and each of the one or more second state values for the other distributed oscillators.

3. The distributed oscillator apparatus of claim 2, wherein:

the first error value is based on a sum of the respective differences between the first state value for the distributed oscillator apparatus and each of the one or more first state values for the other distributed oscillators;

the second error value is based on a sum of the respective differences between the second state value for the distributed oscillator apparatus and each of the one or more second state values for the other distributed oscillators.

4. The distributed oscillator apparatus of claim 1, wherein the first and second oscillator circuits are to compute the respective new state values based on the equation x_i[n+1]=x_i[n]+Δ

t(ƒ

(x₁, x₂)+u_ii[n]), where x_i[n+1] represents the respective new state value, x_i[n] represents the respective state value, Δ

t represents a discrete time step between computation of the state values, ƒ

(x₁, x₂) represents an oscillator function, and u_i[n]=k_ie_i, where e_i represents the respective error value and k_i is a constant.

5. The distributed oscillator apparatus of claim 1, wherein the first and second oscillator circuits each comprise a multiplexer to select an oscillator function from a set of oscillator functions, and the respective new state values computed by the first and second oscillator circuits are based on the selected oscillator function.

6. The distributed oscillator apparatus of claim 5, wherein the set of oscillator functions comprises at least one of a Van der Pol-based oscillator function and an Andronov-based oscillator function.

7. The distributed oscillator apparatus of claim 1, further comprising circuitry to update an oscillation waveform based on the new first state value and the new second state value.

8. The distributed oscillator apparatus of claim 1, further comprising circuitry to generate one or more of a timestamp and a trigger signal based on the new first state value and the new second state value.

9. The distributed oscillator apparatus of claim 8, wherein the circuitry is to generate a timestamp by unwrapping a phase value based on the new first state value and the new second state value.

10. The distributed oscillator apparatus of claim 8, wherein the circuitry is to generate a trigger signal based on a zero-cross detection of a waveform based on the new first state value and the new second state value.

11. The distributed oscillator apparatus of claim 1, wherein:

the receiver circuitry is further to set a flag value in response to receiving a number of packets greater than a threshold number of packets, wherein the packets comprise the first and second state values for the other distributed oscillators; and

the transmitter circuitry is further to transmit the new first state value and the new second state value based on the flag value being set.

12. The distributed oscillator apparatus of claim 1, wherein the receiver circuitry is to compute the first error value further based on a previous first error value and compute the second error value further based on a previous second error value.

13. The distributed oscillator apparatus of claim 1, wherein:

the first oscillator circuit is further to reset the first error value after computing the new first state value;

the second oscillator circuit is further to reset the second error value after computing the new first state value.

14. The distributed oscillator apparatus of claim 1, further comprising:

a first register coupled to the first oscillator circuit to store state values output by the first oscillator circuit; and

a second registers coupled to the second oscillator circuits to store state values output by the second oscillator circuit.

15. A method, comprising:

receiving, at a first distributed oscillator of a first agent in a multi-agent system, a set of state values for one or more second distributed oscillators of respective second agents in the multi-agent system;

computing a set of error values for the first distributed oscillator based on a current set of state values for the first distributed oscillator and the set of state values for the second distributed oscillators, wherein each error value corresponds to a particular state value in the current set of state values and is based on respective differences between the particular state value and state values for the second distributed oscillators corresponding to the particular state value;

computing a new set of state values for the first distributed oscillator based on the set of error values and the current set of state values for the first distributed oscillator; and

transmitting the new set of state values for the first distributed oscillator to the second distributed oscillators.

16. The method of claim 15, wherein computing the set of error values for the first distributed oscillator comprises:

computing a first error value based on a first state value of the current set of state values and first state values for the second distributed oscillators;

computing a second error value based on a second state value of the current set of state values and second state values for the second distributed oscillators.

17. The method of claim 15, wherein the current set of state values for the first distributed oscillator comprises a current first state value and a current second state value, the set of error values comprises a first error value corresponding to the current first state value and a second error value corresponding to the current second state value, and computing the new set of state values comprises:

computing a new first state value based on the first error value, the current first state value, and the current second state value; and

computing a new second state value based on the second error value, the current first state value, and the current second state value.

18. The method of claim 15, wherein the new set of state values for the first distributed oscillator are computed based on the equation x_i[n+1]=x_i[n]+Δ

t(ƒ

(x₁, x₂)+u_i[n]), where x_i[n+1] represents a new i-th state value for the first distributed oscillator, x_i[n] represents a current i-th state value for the first distributed oscillator, Δ

t represents a discrete time step between computation of the current and new state values, ƒ

(x₁, x₂) represents an oscillator function, and u_i[n]=k_ie_i, where e_i represents the i-th error value and k_i is a constant.

19. A multi-agent system, comprising:

a plurality of agents communicably coupled to at least one other agent in the multi-agent system, wherein each agent comprises a local distributed oscillator to;

obtain first and second state values for each distributed oscillator of the other agents in the multi-agent system, wherein the first state values correspond to a first local state value of the distributed oscillator, and the second state values correspond to a second local state value of the distributed oscillator;

compute a first error value based on the first local state value and the first state values for the distributed oscillators of the other agents in the multi-agent system;

compute a second error value based on the second local state value and the second state values for the distributed oscillators of the other agents in the multi-agent system;

compute a new first local state value for the distributed oscillator based on the first error value, the first local state value, and the second local state value;

compute a new second state value for the distributed oscillator based on the second error value, the first local state value, and the second local state value; and

transmit the new first local state value and the new second local state value to the distributed oscillators of the other agents in the multi-agent system.

20. The multi-agent system of claim 19, wherein the distributed oscillator of each agent is to compute the new local state values based on the equation x_i[n+1]=x_i[n]+Δ

t(ƒ

(x₁, x₂)+u_i[n]), where x_i[n+1] represents a new i-th local state value, x_i[n] represents a current i-th local state value, Δ

t represents a discrete time step between computation of the current and new local state values, ƒ

(x₁, x₂) represents an oscillator function, and u_i[n]=k_ie_i where e_i represents the i-th error value and k_i is a constant.