Flow-based adaptive private network with multiple WAN-paths

A method for time calibration in nodes of a network utilizing characterizations of multiple disparate communication paths across the network which vary in transmission behavior frequently over time, the method comprising:

• sending from a peer node a plurality of request messages to a network control point (NCP) over a first plurality of communication paths, wherein the request messages are each tagged with a send time according to a peer node clock in the peer node;
• selecting a first request message received in the NCP and tagged with an arrival time by a master clock in the NCP;
• sending from the NCP in response to the first request message a plurality of reply messages to the peer node over a second plurality of communication paths, wherein the reply messages are each tagged with a reply send time according to the master clock in the NCP, wherein the reply messages each include the send time;
• selecting a first reply message received in the peer node and tagged with a reply arrival time according to the peer node clock;
• calibrating a network time at the peer node to the master clock in the NCP based on the send time according to the peer node clock, the reply send time according to the master clock in the NCP, and the reply arrival time according to the peer node clock;
• measuring a plurality of one way times (OWTs) over a period of time, wherein an OWT is measured as a difference between one reply arrival time of a plurality of reply arrival times and a corresponding one reply send time of a plurality of reply send times associated with a tagged path message transmitted on a selected communication path;
• and
• selecting the lowest OWT from the plurality of OWTs as a best OWT (BOWT) for calibrating the network time at the peer node, whereby accurate continuous measurements are made of source-to-receiver communication OWTs between the NCP and the peer node for both the first and the second communication paths.

A method for time calibration in nodes of a network utilizing characterizations of multiple disparate communication paths across the network which vary in transmission behavior frequently over time, the method comprising:

• sending from a first peer node and separately from a second peer node a plurality of request messages to a network control point (NCP) over a first plurality of communication paths associated with the first peer node and over a second plurality of communication paths associated with the second peer node, wherein the request messages are each tagged with a send time according to an associated first peer node clock in the first peer node and an associated second peer node clock in the second peer node;
• selecting a first request message received from the first peer node in the NCP, a first request message received from the second peer node, and each tagged with an arrival time by a master clock in the NCP;
• sending from the NCP in response to the first request message received from the first peer node and the first request message received from the second peer node a plurality of reply messages to the first peer node over a third plurality of communication paths associated with the first peer node and over a fourth plurality of communication paths associated with the second peer node, wherein the reply messages are each tagged with a reply send time according to the master clock in the NCP and wherein the reply messages each include the send time;
• selecting a first reply message received in the first peer node, a first reply message received in the second peer node, and each tagged with a reply arrival time according to the corresponding peer node clock;
• and
• calibrating a network time at the first peer node and at the second peer node to the master clock in the NCP based on the send time in the first peer node according to the first peer node clock and on the send time in the second peer node according to the second peer node clock, the reply send time to the first peer node and to the second peer node according to the master clock in the NCP, and the reply arrival time in each peer node according to the first peer node clock and according to the second peer node clock, wherein congestion is determined by comparing how many data packets were sent by the first peer node plus a number of unutilized bytes with an indication how many data packets were received by the second peer node.

A method for time calibration in nodes of a network utilizing characterizations of multiple disparate communication paths across the network which vary in transmission behavior frequently over time, the method comprising:

• sending from a peer node a plurality of request messages to a network control point (NCP) over a first plurality of communication paths, wherein the request messages are each tagged with a send time according to a peer node clock in the peer node;
• selecting a first request message received in the NCP and tagged with an arrival time by a master clock in the NCP;
• sending from the NCP in response to the first request message a plurality of reply messages to the peer node over a second plurality of communication paths, wherein the reply messages are each tagged with a reply send time according to the master clock in the NCP, wherein the reply messages each include the send time;
• selecting a first reply message received in the peer node and tagged with a reply arrival time according to the peer node clock;
• calibrating a network time at the peer node to the master clock in the NCP based on the send time according to the peer node clock, the reply send time according to the master clock in the NCP, and the reply arrival time according to the peer node clock;
• generating heartbeat packets every predetermined period unless data packets are received having an inter-packet interval less than the predetermined period;
• and
• measuring based on the calibrated network time a one way communication time between the NCP and the peer node in response to the heartbeat packets.