Abstract: Retransmission of packets in a TCP network is measured by monitoring packets traversing bearer links between nodes and examining the sequence number within its connection of each packet. For each TCP connection being monitored a next expected sequence number (NESN) value is maintained and compared with the actual sequence number of a packet in that connection. If the sequence number is less than the NESN, a retransmission count is incremented by the size of the retransmitted TCP payload; if it is greater than the NESN, a loss counter is incremented by the size of the lost TCP payload. The resulting counts of the lost and retransmitted TCP payload, of the lost and retransmitted packets, enable accurate measurements to be made even at a point remote from the sending node. Comparison of counts from spaced points in a network can enable the location of a fault to be determined at least approximately.

Abstract: Service detail records (SDRs) are generated for at least two layers in a stack of protocols used in a communications network, where protocol data units (PDUs) for a lower-level protocol encapsulate PDUs for a higher-level protocol. SDRs for the higher-level protocol include pointers to SDRs for connections at the lower level which encapsulate PDUs from which the higher-level SDRs were generated.

Abstract: A system and method for automatically monitoring communications networks and for determining network configuration use data stream characterizations. The system comprises a plurality of measurement probes that passively probe the network and collect data packets carried by data streams in the network. The system further comprises characterization computational units that process the collected data packets and produce data stream characterizations from the collected data packets. The data stream characterizations represent individual data streams in an arbitrarily unique manner. The system still further comprises a configuration processing unit that compares data stream characterizations taken at different points in the network and determines data stream paths through the network based on data stream characterization matching.

Abstract: Data cells traversing an Asynchronous Transfer Mode (ATM) network (10) are passively monitored by probes at two spaced points (12,14). A filtering characteristic is applied to select cells comparatively infrequently, and timestamps or sequence numbers (or both) for the cells selected at each point are forwarded to a correlator (28). The correlator identifies similarities in the temporal patterns of intervals between timestamps received from each probe, and uses these similarities to identify and correlate timestamps relating to occurrence of the same cell at the two monitored points. The correlated timestamps are used to determine parameters indicative of the performance of the network, such as one-way delay, variation in that delay, and rate of cell loss.

Abstract: A method of synchronizing the operation of a plurality of measurement probes enables multi-point measurements of a communications network. The method comprises the step of collecting data packets from a data stream of the communications network using the plurality of probes. The data packets are collected after a predetermined start time. The method further comprises the steps of determining a hash value array from an invariant portion of the sampled data packets, and comparing the hash value array to a start hash mask. Collection by a probe of a data packet that produces a hash value array that matches the start hash mask is used as a synchronization event. The method of the present invention allows multiple measurement probes observing the same data stream at different physical locations to be synchronized such that all of the probes observe the same set of packets being carried by the data stream.

Abstract: Retransmission of packets in a TCP network is measured by monitoring packets traversing bearer links between nodes and examining the sequence number within its connection of each packet. For each TCP connection being monitored a next expected sequence number (NESN) value is maintained and compared with the actual sequence number of a packet in that connection. If the sequence number is less than the NESN, a retransmission count is incremented by the size of the retransmitted TCP payload; if it is greater than the NESN, a loss counter is incremented by the size of the lost TCP payload. The resulting counts of the lost and retransmitted TCP payload, of the lost and retransmitted packets, enable accurate measurements to be made even at a point remote from the sending node. Comparison of counts from spaced points in a network can enable the location of a fault to be determined at least approximately.

Abstract: An interactive graphical tool is provided for designing the user interface of a program-controlled instrument. The tool runs on a computer workstation and is used to model the application code of the instrument as a first network in which sessions of user interaction with the application code are represented by respective elements of the network. The actual user interaction sessions themselves are modelled by respective second networks; each second network includes information for defining the interface states of the modelled user interaction session. The full user interface can thereafter be simulated by progressing through the first model until a user interaction element is met and then entering the corresponding second network; the interface state information contained in the second network is used to drive a simulation of the instrument interfaace on the display of the computer workstation.