CN-121604012-B - Method for detecting packet loss along with flow of mobile network by alternative dyeing

CN121604012BCN 121604012 BCN121604012 BCN 121604012BCN-121604012-B

Abstract

The invention discloses an alternate dyeing mobile network flow-following packet loss detection method, which aims to solve the problem of inaccurate packet loss detection caused by link switching in a mobile network. The method comprises the steps of adding detection heads into service flow messages by encapsulation nodes, alternately changing packet loss dyeing identification values in the service flow messages in a fixed period, identifying the detection heads by transmission and decapsulation nodes, dynamically calibrating a local counting period starting point according to first jump of a dyeing value, then recording period counting and time stamps, reporting data comprising period starting and ending time, the dyeing value and message counting to a collector by each node according to a specific rhythm, reconstructing a counting sequence of each node by the collector according to the reported data, identifying and correcting counting sequence interruption and time dislocation caused by link switching through an intelligent algorithm, and finally accurately calculating hop-by-hop and end-to-end packet loss rate by comparing the counting sequences corrected by upstream and downstream nodes. The invention realizes real-time and accurate detection of the packet loss of the service flow in the mobile network environment.

Inventors

YE YANFEI
LI ZEYOU
LI WENFENG
WANG FEIFEI
NIE XIN
ZHAO KANGLIAN

Assignees

南京大学

Dates

Publication Date: 20260508
Application Date: 20260127

Claims (4)

1. The method for detecting the packet loss of the mobile network along with the flow by alternately dyeing is characterized by comprising the following steps of: s1, detecting strategy configuration and stream dyeing, namely adding a detection head into a message of a service stream to be detected by a packaging node, and counting the message in a fixed period Alternately changing the packet loss dyeing identification value in the detection head; S2, period synchronization and counting, namely, after a detection head is identified, a transmission node and a decapsulation node in a path automatically start a detection function for the service flow to be detected, wherein the transmission node and the decapsulation node start a first counting period by taking the time of first identifying the dyeing identification of the service flow to be detected as a starting point, dynamically calibrate the starting time of a local counting period according to the first jump of a packet loss dyeing identification value in the first counting period, and take the time of first identifying the jump of the packet loss dyeing identification value as the starting time of the counting period if the jump of the packet loss dyeing identification value is not detected in the first counting period, and then take the time of first identifying the jump of the packet loss dyeing identification value as the starting time of the counting period Counting and recording time stamps for messages with different packet loss dyeing identification values in a counting period; S3, reporting the detection data, namely periodically reporting statistical information to a central collector by an encapsulation node, a transmission node and a decapsulation node, wherein the reporting time is that the encapsulation node is in each counting period Reporting detection data to a collector at the end, wherein the transmission node and the decapsulation node are not reported in the first counting period, and reach in each counting period after calibration is completed At the moment, the detection data are respectively reported to the collectors, wherein The transmission node and the decapsulation node clear the counter corresponding to the last counting period and the timestamp thereof after reporting; S4, removing the detection head, namely removing the detection head in the message after the unpacking node completes local statistics and information reporting, recovering the original message format, and ensuring transparency for upper layer application; S5, data aggregation and sequence intelligent reconstruction, wherein the collector receives detection data reported by the packaging node, the transmission node and the decapsulation node, reconstructs a continuous counting sequence of the service to be detected flowing through each node, and is used for calculating the packet loss rate by comparing the counting sequences of adjacent nodes, and the data aggregation and sequence intelligent reconstruction specifically comprises the steps of receiving the data reported by the nodes by the collector and executing the following processing: S5.1, generating a sequence, namely aiming at the same service flow to be tested, according to a starting time stamp in the reported data, respectively and independently arranging all nodes including an encapsulation node, a transmission node and a decapsulation node into a time-ordered counting sequence according to the reported detection data; S5.2 mobility compensation and sequence correction: The interruption identification, namely comparing the counting sequence obtained in S5.1 according to the network topology sequence by the collector, and starting switching judgment if the counting sequence of the downstream node M is suddenly interrupted in time, namely the ending time of the counting sequence is earlier than the ending time of the counting sequence of the upstream node M-1; switching judgment, namely searching a new counting sequence Q with a starting time stamp adjacent to a node M counting sequence ending time stamp in data reported by a subsequent transmission node and a decapsulation node, wherein the judgment criteria comprise time difference Whether or not it is less than a threshold value Its value Judging whether the cycle number is identical or not, if yes, judging that the service flow to be detected is switched from the node M to a new node to which the counting sequence Q belongs, and if not, judging that the switching does not occur and the hop loses the packet to enter the next hop for forwarding, wherein the searched counting sequence belongs to a transmission node of the next hop; the dislocation correction comprises the steps of splicing the counting sequence of the node M with the counting sequence Q, merging counting values and time span merging if the dyeing values of two adjacent counting periods are the same, and reserving the two independent counting periods if the dyeing values are different; Iterative splicing, namely repeating the process until all the segment counting sequences generated by switching are spliced into a complete hop-by-hop counting sequence with continuous time; And S6, calculating the packet loss rate, and based on the corrected complete hop-by-hop counting sequence, accurately calculating by the collector.
2. The method for detecting packet loss along with flow in an alternately dyed mobility network according to claim 1, wherein the period synchronization and counting in step S2 specifically includes: S2.1, initializing, namely creating two counters C0 and C1 and a corresponding time stamp recording unit for a service flow to be detected; s2.2, the period synchronization is that the transmission node and the decapsulation node start a first counting period by taking the time of first identifying the dyeing identifier of the service flow to be detected as a starting point, and the first counting period is that: if the packet loss dyeing mark is detected to generate 0/1 jump, immediately taking the jump time as a new counting period starting point to finish accurate synchronization with the dyeing rhythm of the packaging node; If the jump of the packet loss dyeing identification value is not detected, taking the time of first identifying the jump of the packet loss dyeing identification value as a period starting point; s2.3, continuously counting periods, namely, after synchronization is completed, the transmission node and the decapsulation node enter a stable counting state, wherein the time length of each counting period is T, and the transmission node and the decapsulation node accumulate the message count into the corresponding counters C0 and C1 according to the dyeing value of the message dyeing identification bit, namely 0 or 1, and continuously update the ending time stamp of the current period, namely the arrival time of the last message; and S2.4, releasing resources, namely if the fact that the values of the two counters in the current counting period are always 0 and are not changed is detected when the counting of the continuous periods is finished, judging that the forwarding of the current node to the service flow to be tested is finished, and stopping maintaining the detection information of the service flow to be tested.
3. The method for detecting packet loss along with flow of a mobile network by alternative dyeing according to claim 1, wherein the detecting data reporting in step S3 specifically includes reporting content including a period T, a packet loss dyeing identification value in the period, a total number of messages sent in the period, and a first message time stamp in the period by a transmitting node and a decapsulating node, where the reporting content includes: the packet loss dyeing identification value of the last complete counting period; the start and end time stamps of the last complete count period; Starting from the previous counting period and reporting the time Until the total number of the received messages with the dyeing value is reached.
4. The method for detecting packet loss along with flow in an alternately dyed mobility network according to claim 1, wherein the packet loss rate calculation in step S5 specifically includes: S6.1, for two adjacent A, B nodes, in any same counting period, the packet loss rate= (A node sending count-B node receiving count)/A node sending count; S6.2, comparing the total count of the encapsulation node and the final decapsulation node in the same time interval.

Description

Method for detecting packet loss along with flow of mobile network by alternative dyeing Technical Field The invention belongs to the technical field of communication network performance detection, and particularly relates to an alternate-dyeing mobile network flow-following packet loss detection method which is used for detecting a specific service flow in real time and accurately in a network (such as a 5G network, a satellite network, a car networking and the like) with mobility characteristics. Background The flow-following detection technology generally refers to a technical means for monitoring and analyzing a real service flow in a network, and one of the flow-following detection technology and the flow-following detection technology is directly embedded or attached to the service flow, namely a data packet, to detect the state and the performance of the network, and periodically report data to a control surface, wherein packet loss measurement is one of core capabilities in flow-following detection and performance monitoring. The alternative labeling method (ALTERNATE MARKING, AM) commonly used in flow-following detection is a core technology defined by the international standards organization IETF, the principle of which is clearly specified in RFC 9341. The mobility network is a network in which nodes in the network have mobility, a topological structure and a link state dynamically change along with time, and is widely applied to scenes such as satellite communication, internet of vehicles, unmanned aerial vehicle clusters, self-organizing networks and the like. When the conventional fixed network forwarding plane transmits node counting, the counting period of the dyeing mark is finished and the counting period of another mark is jumped due to time delay jitter and message disorder, but the message still does not arrive. In order to solve the problem, in the process of calculating the packet loss rate by the IFIT technology, on a counter of a forwarding plane node, the counter is emptied and reported to a collector after the corresponding dyeing identification bit of the previous counting period is counted in the current period by reading the previous counting period at the time point of 1/3-2/3 of the counting period, so that the influence of short-period message disorder and delay jitter on the packet loss rate is solved. Satellite networks are typical representatives of mobility networks, and naturally have topology dynamics, and as inter-satellite and inter-satellite links are switched, the existing methods have the following technical bottlenecks: (1) And the cross-period statistics dislocation problem caused by switching. When node switching occurs in the mobility network, the service flow is cut off and forwarded to another transmission node. When the switched transmission node forwards and counts the service flow messages, the counting period of the first dyed message is quite possibly not a complete counting period, which can lead to the fact that the counting period of the transmission node is not aligned with the packet loss dyeing value of the dyeing period of the encapsulation node, and the counting dislocation of the message crossing periods is caused. Even when the transmission node counts the last complete counting period and the counting number of the packet loss dyeing values of the current last counting period in the second counting period at 1/3-2/3 time point, the counting of the packet loss dyeing values of the first counting period and the third counting period is possible, so that the problem of more serious cross-period counting dislocation is caused. (2) The problem of confusing counting cycle of the dyed marks. In a mobile network, due to the fact that short-time disorder and delay jitter of small-segment messages exist in advance or delay of arrival of the messages, a node cannot judge counting period attribution of different dyeing mark messages received near a period updating time point, and the problem of disordered counting period of the dyeing mark is caused in a collector. Because the alternate dyeing method does not adopt a sequence number mechanism in order to reduce message load, the problem of packet loss counting error caused by continuous disorder of large-section messages cannot be solved. Disclosure of Invention The invention provides an alternate dyeing mobile network packet loss detection method, which is characterized in that under the premise of not changing the content of a message header inserted by the existing protocol, packet loss detection information is compared and spliced in a collector by a double counter cooperation, counting period starting calibration and detection data hop-by-hop period reporting mechanism, and then packet loss information is calculated, so that the problem of packet loss calculation accuracy caused by cross-period statistics dislocation and dye mark counting period confusion caused by switching is solved. The specific