| With the continuous emergence of new vertical industry applications such as indus-trial Internet,autonomous driving,and intelligent factories,network virtualization and white-boxing technologies have become the trend in the future development of 5G net-works.5G networks are gradually moving towards open-source networks.Open-source5G networks not only support the eMBB,URLLC,and m MTC services in traditional5G,but also support personalized customization and expansion according to users’needs,thereby achieving better business adaptation and network optimization to meet the diverse needs of users.To achieve differentiated services,the network needs to have real-time knowledge of the performance metrics of business transmission in order to implement precise and intelligent control.Therefore,this article focuses on studying the measure-ment methods of business performance indicators for open-source 5G network scenarios,providing necessary inputs for intelligent 5G networks.The measurement of one-way delay in the network is crucial for monitoring and opti-mizing the quality of application carrying,such as video conferencing and remote medical treatment.For example,remote medical treatment requires real-time transmission of im-age and sound data,which must arrive at the receiving end within a short period of time,otherwise it will affect diagnosis and treatment.One-way delay measurement can provide important basis for business routing adjustment and optimization,helping network man-agers to timely identify network bottlenecks and faults,and take measures to optimize the network to meet user’s demand for delay.Therefore,Chapter 2 of this article studies a one-way delay measurement scheme based on the Sketch algorithm.This scheme compresses statistical information through Sketch and separates abnormal delay flows through Bloom filters,helping network managers quickly locate relevant information about abnormal de-lay flows.The experiment shows that the one-way delay measurement scheme proposed in this chapter can obtain relatively accurate measurement results with low high-speed storage resource costs.The separation strategy of abnormal delay flows can also quickly locate exceptional flows.The measurement of round-trip delay is very important for optimizing network trans-mission protocols.For example,the TCP protocol controls the sending rate through con-gestion window to avoid further deterioration of network congestion,and the size of the congestion window is directly related to RTT;the QUIC protocol calculates the expira-tion time of data packets through RTT to improve security.The method of monitoring the three-way handshake of TCP is difficult to accurately measure long-duration TCP business flows.Continuous monitoring of TCP business flows requires dealing with prob-lems such as retransmission and reordering.Therefore,Chapter 3 of this article studies a round-trip delay measurement scheme that uses ACK packets to continuously monitor delay.This scheme continuously monitors abnormal states such as retransmission and re-ordering of business flows through programmable network elements,and generates ACK packets when abnormal states occur to assist the sending end in completing measurements to ensure the validity of measurement samples.The experiment shows that when abnor-mal states such as retransmission and reordering occur,programmable network elements can ensure the matching of data packets between the sending and receiving ends by gen-erating packets,avoiding false measurement and missed measurement,and ensuring the validity and accuracy of measurement.In applications such as real-time voice and video communication,even a small amount of packet loss or delay will result in a decline in communication quality,and network managers need to adjust the network in a timely manner to ensure reliable transmission of business.Therefore,measuring the packet loss rate of the business is crucial for net-work management control.Existing network-level packet loss rate measurement methods need to deal with problems such as time synchronization or packet disorder.Therefore,Chapter 4 of this article proposes a network-level packet loss rate measurement scheme based on packet numbering.This scheme adds a numbering field to the data packet us-ing programmable network elements,and can complete the measurement at the receiving end without time synchronization;at the same time,a greedy-based maximum matching algorithm is used to allocate high-speed storage resources to solve the problem of hash con-flicts.The experiment shows that this scheme can reduce the high-speed storage resource cost by 50%with only a 1.5%increase in bandwidth consumption,and packet disorder does not affect the measurement results.The greedy-based maximum matching algorithm can complete all business flow measurement tasks within several batches,avoiding hash conflicts and bandwidth waste. |
