CN114268537B - Deterministic network-oriented network slice generation and dynamic configuration system and method - Google Patents
Deterministic network-oriented network slice generation and dynamic configuration system and method Download PDFInfo
- Publication number
- CN114268537B CN114268537B CN202111424669.XA CN202111424669A CN114268537B CN 114268537 B CN114268537 B CN 114268537B CN 202111424669 A CN202111424669 A CN 202111424669A CN 114268537 B CN114268537 B CN 114268537B
- Authority
- CN
- China
- Prior art keywords
- network
- deterministic
- module
- flow
- slice
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 22
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 35
- 230000005540 biological transmission Effects 0.000 claims abstract description 28
- 239000011159 matrix material Substances 0.000 claims description 42
- 238000005259 measurement Methods 0.000 claims description 27
- 230000003068 static effect Effects 0.000 claims description 15
- 230000002787 reinforcement Effects 0.000 claims description 14
- 238000010276 construction Methods 0.000 claims description 10
- 241001522296 Erithacus rubecula Species 0.000 claims description 5
- 238000002955 isolation Methods 0.000 claims description 5
- 230000001934 delay Effects 0.000 claims description 4
- 238000013461 design Methods 0.000 claims description 4
- 238000002347 injection Methods 0.000 claims description 4
- 239000007924 injection Substances 0.000 claims description 4
- 238000012544 monitoring process Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 4
- 238000012549 training Methods 0.000 claims description 4
- 230000001172 regenerating effect Effects 0.000 claims description 3
- 230000007246 mechanism Effects 0.000 claims description 2
- 238000012821 model calculation Methods 0.000 claims 1
- 230000009471 action Effects 0.000 description 9
- 230000006870 function Effects 0.000 description 6
- 238000011160 research Methods 0.000 description 4
- 239000000243 solution Substances 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 239000007788 liquid Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000013468 resource allocation Methods 0.000 description 2
- 230000007704 transition Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000008929 regeneration Effects 0.000 description 1
- 238000011069 regeneration method Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention relates to the field of Internet, and discloses a deterministic network-oriented network slice generation and dynamic configuration system and method. The deterministic network-oriented network slice generation and dynamic configuration system comprises an information acquisition module, a software control module and a network transmission module. The network slice generation and dynamic configuration method guarantees bandwidth resources of deterministic traffic by dividing a link space of a network layer, achieves division of the network slice through a network slice algorithm, achieves reservation and division of the network resources based on deterministic traffic flows, avoids network congestion queuing and resource robbing generated when a plurality of deterministic data flows arrive at an end node at the same time by scheduling time slots of traffic in the deterministic network virtual slice, divides the bandwidth of the deterministic traffic from a time dimension, increases flexibility of the slice through the time slots, and can be adjusted through the time slots without resetting the slice if the traffic is different.
Description
Technical Field
The invention relates to the field of Internet, in particular to a deterministic network-oriented network slice generation and dynamic configuration system and method.
Background
With the increasing number of applications of industry 4.0, remote driving, remote operation and the like, the requirements on ultra-low time delay and micro-jitter of a network are higher and higher. The industrial Internet of things requires end-to-end delay to be microsecond to millisecond, and jitter to be microsecond; the haptic Internet (teleoperation) requires 3-10 ms of end-to-end delay, and jitter is not more than 2ms; the driving assistance requires an end-to-end delay of 100-250 mus and a jitter of a few microseconds. Remote driving requires not only low delay jitter but also higher transmission rates. To meet the above application requirements for the network, the time sensitive network (time sensitive network, TSN) and deterministic network (deterministic network, detNet) optimize the link layer and network layer of the ethernet network, respectively, improving its support capability for time sensitive streaming.
Definition of certainty in network 5.0 refers to providing all-round deterministic capability guarantees including deterministic latency, deterministic jitter, deterministic paths beyond the forwarding capability of the IP network "best effort" to meet the stringent requirements of future traffic on network quality. In order to guarantee deterministic traffic demands, the DetNet optimizes the ethernet L3 layer from three aspects of time certainty, resource certainty, and path certainty. Wherein the resource reservation in the key technology in the resource certainty relates to resource allocation, resource isolation and the like. The network slice at present can reasonably allocate equipment resources through NFV (network function virtualization ), convert the equipment resources into a plurality of end-to-end network slices with different granularity and high isolation, and virtually adapt to different services and mutually insulated sub-network slices respectively in an on-demand networking mode.
At present, no concept related to slicing is proposed in the deterministic network field, but many researches on resource reservation for guaranteeing Qos service quality and 5G for slicing are proposed in the network, most of the researches only achieve simple isolation between virtualized slicing and services, many researches on slicing do not consider service variability of slicing and fairness among different services, and most of the researches on slicing are realized from historical statistics of service requests, but the network resource allocation is not realized by combining the perception angle of current service flow, namely, resources are allocated only by adopting network future load states predicted by a simple statistical method, which does not accord with the concept of deterministic flow priority in the deterministic network.
In summary, the current network slicing method based on deterministic network has the following problems:
1. the characteristics of deterministic traffic are not considered, the variability of traffic services and fairness between different services are not considered
2. The deterministic traffic priority concept is not considered and the dynamic adjustment of the slice is not considered.
3. The emergency situation of deterministic flow and the type situation of flow are not considered, and slicing has no flexibility.
Disclosure of Invention
The invention aims to provide a deterministic network-oriented network slice generation and dynamic configuration system and method, which overcome the defects and the shortcomings of the prior art.
In order to solve the above technical problems, the present invention provides a deterministic network-oriented network slice generation and dynamic configuration system, which mainly includes: the system comprises an information acquisition module, a software control module and a network transmission module;
the information acquisition module is used for acquiring the information of the network transmission module and outputting flow information in a deterministic network to the software control module. The information acquisition module comprises a deterministic network dynamic measurement module and a deterministic network static measurement module; wherein,
the deterministic network dynamic measurement module is used for measuring the link load state and the network flow state in the network, the obtained flow in the deterministic network is represented in the form of a flow matrix, and the bandwidth, time delay and jitter information required by the deterministic flow are stored and represented by text documents;
and the deterministic network static measurement module is used for measuring static information in a network, including router connection conditions, port conditions, link bandwidth conditions and the like.
The software control module is used for predicting, constructing, scheduling and adjusting deterministic network slices and outputting information to the network transmission module; the software control module comprises a deterministic network slice prediction module, a deterministic network slice construction module, a deterministic network slice scheduling module and a deterministic network slice adjustment template; the deterministic network slice prediction module is used for predicting deterministic network traffic for a period of time in the future through an RNN (RNN round robin network) method.
And the deterministic network slice construction module is used for reserving bandwidth resources for the links through a network slice method of deep reinforcement learning to output the duty ratio of the links occupied by the deterministic network slice.
And the deterministic network slice scheduling module is used for scheduling resources in the deterministic network slice on the basis of time slots through the required bandwidth, time delay and jitter of the real-time deterministic traffic in the network output by the measurement module.
And the deterministic network slice adjustment template is used for outputting the real-time state of the link through the network dynamic measurement module, and if the load of the deterministic network exceeds a set threshold value, performing re-deterministic network slice construction.
The output end of the network transmission module is connected with the information acquisition module and the underlying network topology and is used for issuing a deterministic network flow table and configuring deterministic network slices; the network transmission module comprises a deterministic network flow table issuing module and a deterministic network slice configuration module, wherein the deterministic network flow table issuing module is used for carrying out port configuration and flow table issuing on a network layer through flow table information. And the deterministic network slice configuration module is used for carrying out link bandwidth allocation and configuration of the deterministic switch through slice information output by the network slice module, wherein deterministic and nondeterministic ports are divided for configuration.
The invention also provides a deterministic network-oriented network slice generation and dynamic configuration method, which comprises the following steps:
step 1: and acquiring a deterministic flow data set, generating a deterministic flow matrix and required jitter and time delay information, inputting the acquired deterministic flow matrix and related information into a depth model according to a time sequence to perform flow prediction, and performing deterministic network flow prediction.
Step 2: inputting the obtained deterministic network flow prediction data set with bandwidth, time delay and jitter requirements and topology basic information in the network into a network slicing algorithm based on reinforcement learning to form bandwidth and port information required by deterministic slicing, and configuring the deterministic network slicing according to the bandwidth and port information required by deterministic slicing.
Step 3: and (3) carrying out resource scheduling in the deterministic slice of the time slot layer by using a bandwidth scheduling algorithm strategy of the time slot layer in the deterministic network slice, and reducing network congestion queuing and resource preemption generated when a plurality of deterministic data streams arrive at the end node at the same time in the time dimension.
Step 4: calculating network benefits according to link bandwidth, topology and deterministic flow information in a network, and returning to the step 1 to regenerate and dynamically configure a slicing strategy if the network benefits do not meet a certain design threshold; otherwise, monitoring of link bandwidth, topology and deterministic traffic information in the network is maintained.
Preferably, step 1 comprises the sub-steps of:
step 1-1: collecting historical flow data in a network by utilizing output data of a data acquisition module, and classifying the historical flow data according to different deterministic service requirements to form a deterministic network flow matrix and a non-deterministic network flow matrix;
step 1-2: modeling the formed flow matrix to obtain a flow matrix containing jitter and time delay information related information required by deterministic flow;
step 1-3: then, predicting deterministic traffic through an RNN (RNN round robin) network algorithm to form a predicted traffic matrix of the deterministic traffic related to time sequence;
preferably, step 2 comprises the sub-steps of:
step 2-1: the link bandwidth size in the deterministic network is collected by a network resource static collection module.
Step 2-2: inputting the data set of the deterministic network flow matrix predicted by the RNN algorithm in the step 1 into a reinforcement learning network slicing algorithm, and generating network slicing module configuration parameters of port numbers and reserved bandwidths of deterministic network resources which can be reserved.
Step 2-3: and reserving resources on the space bandwidth in a network slice configuration module of the transmission module to form an initial network slice.
Preferably, step 3 comprises the sub-steps of:
step 3-1: in the step 2, two virtual sub-networks of deterministic network slices and non-deterministic network slices are generated through deterministic network configuration, and real-time deterministic network traffic and types, time delays and jitter of the deterministic network traffic and the types, time delays and jitter are extracted and counted through a network state measurement module;
step 3-2: and quantifying the bandwidth in the deterministic network slice into time slots according to different types of deterministic traffic, dividing the time slots in the deterministic network for the flexible entering deterministic traffic, and carrying out resource scheduling on the bandwidth in the deterministic slice through a time slot algorithm.
Step 3-3: and according to the division result of the time slot algorithm, adjusting the injection time of the scheduling flow in the network transmission module to realize the resource scheduling algorithm in the deterministic network slice based on the time slot.
Preferably, step 4 comprises the sub-steps of:
step 4-1: and carrying out state monitoring on the deterministic traffic in a deterministic network dynamic measurement module.
Step 4-2: and (3) processing the network state data monitored in the step (4-1), analyzing the network state condition, and regenerating and configuring the deterministic network slice when the bandwidth, time delay and jitter average service guarantee rate of the deterministic traffic are smaller than a design threshold value.
The invention guarantees the bandwidth resources of deterministic service by dividing the link space of the network layer, realizes the division of network slices by a network slicing algorithm, realizes the reservation and division of network resources based on deterministic service flows, and avoids network congestion queuing and resource robbing generated when a plurality of deterministic data flows arrive at an end node at the same time by scheduling the flow in the deterministic network virtual slices, and the division of the bandwidth of deterministic service is carried out from the time dimension, thereby increasing the flexibility of slicing by time slots, and adjusting by time slots without resetting the slices if the services are different.
Drawings
The technical scheme of the invention is further specifically described below with reference to the accompanying drawings and the detailed description.
Fig. 1 is a block diagram showing the construction of the modules of the system of the present invention.
Fig. 2 is a functional logic block diagram of the modules of the system of the present invention.
Fig. 3 is a general flow chart of the method of the present invention.
Detailed Description
With reference to fig. 1 and fig. 2, the deterministic network slice generation and dynamic configuration system of the present invention includes an information acquisition module, a software control module, and a network transmission module;
and the information acquisition module is used for acquiring the information of the network transmission module and outputting the flow information in the deterministic network to the software control module. The information acquisition module comprises a deterministic network dynamic measurement module and a deterministic network static measurement module; the deterministic network dynamic measurement module is used for measuring a link load state and a network flow state in a network, the obtained flow in the deterministic network is represented in a flow matrix form, and bandwidth, time delay and jitter information required by the deterministic flow are stored and represented by text documents; and the deterministic network static measurement module is used for measuring static information in a network, including router connection conditions, port conditions, link bandwidth conditions and the like.
The software control module is used for predicting, constructing, scheduling and adjusting deterministic network slices and outputting information to the network transmission module; the software control module comprises a deterministic network slice prediction module, a deterministic network slice construction module, a deterministic network slice scheduling module and a deterministic network slice adjustment template; the deterministic network slice prediction module is used for predicting deterministic network traffic for a period of time in the future through an RNN (RNN round robin network) method. And the deterministic network slice construction module is used for reserving bandwidth resources for the links through a network slice method of deep reinforcement learning to output the duty ratio of the links occupied by the deterministic network slice. And the deterministic network slice scheduling module is used for scheduling resources in the deterministic network slice on the basis of time slots through the required bandwidth, time delay and jitter of the real-time deterministic traffic in the network output by the measurement module. And the deterministic network slice adjustment template is used for outputting the real-time state of the link through the network dynamic measurement module, and if the load of the deterministic network exceeds a set threshold value, performing re-deterministic network slice construction.
The output end of the network transmission module is connected with the information acquisition module and the underlying network topology and is used for issuing a deterministic network flow table and configuring deterministic network slices; the network transmission module comprises a deterministic network flow table issuing module and a deterministic network slice configuration module, wherein the deterministic network flow table issuing module is used for carrying out port configuration and flow table issuing on a network layer through flow table information. And the deterministic network slice configuration module is used for carrying out link bandwidth allocation and configuration of the deterministic switch through slice information output by the network slice module, wherein deterministic and nondeterministic ports are divided for configuration.
Referring to fig. 2 and 3, the deterministic network slice generation and dynamic configuration process of the present invention is as follows:
step 1: and acquiring a deterministic flow data set from the information acquisition module, generating a deterministic flow matrix and required jitter and time delay information, inputting the acquired deterministic flow matrix and related information into the depth model according to a time sequence to perform flow prediction, and performing deterministic network flow prediction.
Step 1 comprises the following sub-steps:
step 1-1: collecting historical flow data in a network by utilizing output data of a data acquisition module, and classifying the historical flow data according to different deterministic service requirements to form a deterministic network flow matrix and a non-deterministic network flow matrix; the step 1-1 specifically comprises the following steps: acquiring a data set which is marked with traffic types in a deterministic network, wherein the data set has end-to-end traffic requirements and is marked with all traffic types, and the traffic types in the deterministic network are classified into 10 types, wherein 0, 1 and 2 are non-deterministicThe flow rate of the liquid is controlled,class 3-9 as deterministicThe flow rate of the liquid is controlled,the non-deterministic traffic is internally divided into background traffic, best effort traffic and guaranteed traffic to the greatest extent. Deterministic traffic has different guarantees of latency, bandwidth, jitter,3 rdThe class is the class of calculation force and the class of emergency guarantee flow;4 thClass is time delay<Deterministic traffic for video conferencing class traffic for 100 ms;5 thClass is time delay<10ms voice call deterministic traffic;6 thClasses are internet route control protocols (OSPF, RIP, DNS, etc. traffic.No. 7The class is holographic communication and VR interactive class traffic,8 thClass is remote backhaul class traffic (multispectral video and sensor traffic in industrial control, telemedicine and remote driving),9 thThe class is a remote control class flow (industrial control, telemedicine, and in-remote-drive control instruction class flow). They contain deterministic network traffic related demand information including bandwidth, jitter, latency, etc. Will beThey areThe types are classified into a matrix of deterministic traffic types and a non-deterministic traffic matrix according to different types.
Step 1-2: modeling the formed flow matrix to obtain a flow matrix containing jitter and time delay information related information required by deterministic flow; the step 1-2 specifically comprises the following steps: traversing deterministic network information according to the deterministic traffic data classified in step 1-1; sorting end-to-end traffic in a network, and processing deterministic traffic and non-deterministic traffic into a traffic matrix form, wherein deterministic network topology information is represented by an undirected graph g= (V, E), V represents a set of nodes and |v|=n; e represents a set of links and |e|=l; each time data set generates two flow matrices of deterministic and non-deterministic flows, the value of each element representing the size of the flow demand sent between the corresponding OD pairs. The generated traffic matrix format is expressed as follows:
wherein m is i,j Representing the flow demand from i to j.
Step 1-3: then, predicting deterministic traffic through an RNN (RNN round robin) network algorithm to form a predicted traffic matrix of the deterministic traffic related to time sequence; the steps 1-3 specifically comprise the following steps: building an RNN model according to a deterministic flow matrix generated by a historical data set, inputting the deterministic flow matrix with time correlation and deterministic flow demand data, automatically building the RNN model of network flow, training, calculating the model after training, and outputting a flow matrix of a predicted deterministic network in a later period of time.
Step 2: and inputting the obtained deterministic network flow prediction data set with bandwidth, time delay and jitter requirements and topology basic information in the network into a network slicing algorithm based on reinforcement learning, and outputting bandwidth and port information required by forming deterministic slices into a network configuration module. Step 2 comprises the following sub-steps:
step 2-1: the link bandwidth size in the deterministic network is collected by a network resource static collection module. The step 2-1 specifically comprises the following steps: the static resources such as bandwidth, route and the like in a deterministic network are collected through a network static resource collection module, and end-to-end link bandwidth information forms an end-to-end link bandwidth matrix, wherein the matrix is expressed as follows:
wherein w is i,j Representing the link bandwidth from i to j.
Step 2-2: inputting the data set of the deterministic network flow matrix predicted by the RNN algorithm in the step 1 into a reinforcement learning network slicing algorithm, and generating network slicing module configuration parameters of port numbers and reserved bandwidths of deterministic network resources which can be reserved. The step 2-2 specifically comprises the following steps: inputting the prediction data set of the deterministic network traffic matrix obtained in the step 1 and the static bandwidth resource data of the link obtained in the step 2-1 into a network slicing algorithm based on reinforcement learning, establishing a Markov decision model, searching an optimal strategy, and maximizing the future expected return rewards. The deep reinforcement learning provides a general algorithm framework for the resource reservation of the slice, and comprises 3 basic elements of a State space State, action space actions and a Reward return function Reward. For deterministic network scenarios, the following is defined:
1) State, which represents the State of the deterministic network and comprises three kinds of information, namely, the reservation ratio of resources of the current deterministic slice, the link utilization rate in the deterministic network and the resource guarantee rate of deterministic traffic, and can be specifically represented by the following 3 numerical values. Resource reservation ratio S of slice s The ratio of deterministic slice to the whole system resource is indicated, and the resource utilization rate R in the link s Refers to the ratio of the actually used link bandwidth to the total link bandwidth; wherein the flow resource guarantee rate in the deterministic slice is divided into three indexes, and the three indexes are respectively time delay guarantee D s Jitter protection J s Bandwidth guarantee W s . Resource guarantee rate A s Is a comprehensive representation of three indexes, and for the deterministic network, the State set is defined as [ S ] s ,R s ,A s ]
2) An Action represents a set of actions performed. The DRL will perform the acquisition of the state and then select and execute an action based on the greedy algorithm. In deterministic networks, the action operation is to dynamically adjust the system duty cycle of the bandwidth resources in the link. On the original bandwidth allocation of the link, the increase allocation and the decrease allocation of the deterministic traffic link bandwidth are performed.
3) Reward, which represents rewards and rewards fed back by environment interaction, and rewards are carried out according to executed actions, wherein in a deterministic network scene, the rewards and the resource guarantee rate of deterministic traffic are related, and are defined as:
A s ＝αW s +βJ s +γD s
assume that the current policy is expressed asBut->Is the current rewards function, then the current rewards function is:
the optimal equation for the Q value can be expressed as:
where γ is the decay factor of the Markov process, so the decision function is defined as the difference of the actions of this state transition to the next bandwidth reservation state, P is the probability of the state transition, assuming the next action is A t Then the decision function is:
the definition above allows deterministic network slicing to maximize future expected rewards. The ratio of the reserved bandwidth slice to the whole link is obtained, and the port number of the reserved deterministic network resource, the reserved bandwidth network slice data are obtained.
Step 2-3: and reserving resources on the space bandwidth in a network slice configuration module of the transmission module to form an initial network slice. The step 2-3 specifically comprises the following steps: the reserved data of the deterministic network slice generated by reinforcement learning is configured into a deterministic network through a network slice configuration module of a network transmission module, the bandwidth of a link is configured in a limited flow mode mainly through a queue mechanism in a deterministic switch, and the link bandwidth of the deterministic network and the link bandwidth of a non-deterministic network are subjected to virtualized isolation, so that the bandwidths of the deterministic network and the non-deterministic network are not interfered with each other.
Step 3: and (3) carrying out a bandwidth scheduling algorithm strategy of a time slot layer in the deterministic network slice configured by the network transmission module, and carrying out a resource scheduling algorithm in the deterministic network slice of the time slot layer, so as to reduce network congestion queuing and resource robbing generated when a plurality of deterministic data streams arrive at the end node at the same time in a time dimension. Step 3 comprises the following sub-steps:
step 3-1: in step 2, two virtual sub-networks of deterministic network slices and non-deterministic network slices have been generated by configuring the deterministic network, and real-time deterministic network traffic and non-deterministic network traffic are measured by a network state measurement moduleThey areExtracting and counting the type, time delay and jitter of (1); the step 3-1 specifically comprises the following steps: the state measurement in the deterministic network slice is carried out by an in-band measurement method in the network measurement module, the category classification of deterministic traffic within 5 seconds of the period and the service requirements and quality such as time delay, bandwidth, jitter and the like required by the deterministic traffic are extracted, and modeling is carried out, wherein the deterministic traffic of different categories are divided into different deterministic traffic matrixes.
Step 3-2: and quantifying the bandwidth in the deterministic network slice into time slots according to different types of deterministic traffic, dividing the time slots in the deterministic network for the flexible entering deterministic traffic, and carrying out resource scheduling on the bandwidth in the deterministic slice through a time slot algorithm. The step 3-2 specifically comprises the following steps: due to the difference of the arrival types and time of deterministic traffic, when bandwidth allocation of different deterministic traffic is performed in deterministic slices, bandwidth resources can be quantized into time slots through indexes such as the type of deterministic traffic and required bandwidth, time delay, jitter and the like which are achieved within 5s, wherein the total bandwidth resources in the deterministic slices are denoted as WS, the period is denoted as T, and when the period is divided into N time slots, the bandwidth granularity corresponding to each time slot in the period T is
bw＝WS/N
The time slot allocation is carried out preferentially for the service with less time delay requirement through the scheduling algorithm, if the number of the time slots with high priority in the deterministic traffic is n, the bandwidth resource allocated in the period T in the deterministic network is that
BW1＝bw*n
Step 3-3: and according to the division result of the time slot algorithm, adjusting the injection time of the scheduling flow in the network transmission module to realize the resource scheduling algorithm in the deterministic network slice based on the time slot. The step 3-3 specifically comprises the following steps: the strategy of reserved time slots of each type of deterministic flows in the deterministic slice is obtained through the step 3-2, the injection time of various deterministic type flows is adjusted in the network transmission module according to the strategy, the flexibility in the deterministic slice is ensured, and the resource scheduling algorithm in the deterministic network slice based on the time slots in the deterministic network is realized.
Step 4: and (3) regenerating a slicing strategy by using a method of designing a threshold value according to the link load condition and the statistical condition of deterministic traffic in the network measured by the network measurement module, adjusting and generating the network slicing, and repeating the steps of the steps 1-4. Step 4 comprises the following sub-steps:
step 4-1: and carrying out state monitoring on the deterministic traffic in a deterministic network dynamic measurement module. The step 4-1 specifically comprises the following steps: policies and calculations for utilization of links, utilization of deterministic network slices, bandwidth satisfaction of deterministic traffic for deterministic network slices and overall network state are performed by a deterministic network dynamic measurement module. The utilization rate of the link is analyzed to be VL, the utilization rate of the deterministic network slice is analyzed to be DL, the bandwidth satisfaction rate of deterministic traffic is analyzed to be WL, and the bandwidth satisfaction rate of non-deterministic traffic is analyzed to be UL.
Step 4-2: and (3) processing the network state data monitored in the step (4-1), analyzing the network state condition, and when the bandwidth, time delay and jitter average service guarantee rate of deterministic traffic are smaller than a threshold value of 0.8, slicing the repartitioned network and repeating the steps (1-3). The step 4-2 specifically comprises the following steps: the link utilization is VL, the utilization of deterministic network slices is DL, the bandwidth satisfaction rate of deterministic traffic is WL, the bandwidth satisfaction rate of non-deterministic traffic is UL, and the benefits in the deterministic network are considered to be defined as:
since the guarantee is mainly the bandwidth of the deterministic network, it willLet 0.8 be, β be 0.2, and the slice assurance threshold K at the time of generating the just-sliced slice is calculated. If Value in subsequent time period T<The number of K exceeds 30%, then this slice is indicated to be adjusted, regeneration of the slice is started, and steps 1-3 are repeated.
Finally, it should be noted that the above-mentioned embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications and equivalents may be made to the technical solution of the present invention without departing from the spirit and scope of the technical solution of the present invention, and all such modifications and equivalents are intended to be encompassed in the scope of the claims of the present invention.
Claims (6)
1. The deterministic network-oriented network slice generation and dynamic configuration system is characterized by comprising an information acquisition module, a software control module and a network transmission module; wherein,
the information acquisition module is used for acquiring the information of the network transmission module and outputting flow information in a deterministic network to the software control module;
the software control module is used for predicting, constructing, scheduling and adjusting deterministic network slices and outputting information to the network transmission module;
the output end of the network transmission module is connected with the information acquisition module and the underlying network topology and is used for issuing a deterministic network flow table and configuring deterministic network slices;
the software control module comprises a deterministic network slice prediction module, a deterministic network slice construction module, a deterministic network slice scheduling module and a deterministic network slice adjustment template; wherein,
the deterministic network slice prediction module is used for predicting deterministic network traffic in a future period of time through an RNN (RNN round-robin network) method;
the deterministic network slice construction module is used for reserving bandwidth resources of the link through a network slice method of deep reinforcement learning to output the duty ratio of the link occupied by the deterministic network slice;
the deterministic network slice scheduling module is used for scheduling resources in the deterministic network slices based on time slots through the required bandwidth, time delay and jitter of the real-time deterministic traffic in the network output by the measuring module;
the deterministic network slice adjustment template is used for outputting the real-time state of the link through the network dynamic measurement module, and if the load of the deterministic network exceeds a set threshold value, the deterministic network slice construction is performed again;
acquiring a deterministic flow data set from the information acquisition module, generating a deterministic flow matrix and required jitter and time delay information, inputting the acquired deterministic flow matrix and related information into a depth model according to a time sequence to perform flow prediction, and performing deterministic network flow prediction;
inputting the obtained deterministic network flow prediction data set with bandwidth, time delay and jitter requirements and topology basic information in a network into a network slicing algorithm based on reinforcement learning, and outputting bandwidth and port information required by forming deterministic slices into a network configuration module;
and (3) carrying out a bandwidth scheduling algorithm strategy of a time slot layer in the deterministic network slice configured by the network transmission module, and carrying out a resource scheduling algorithm in the deterministic network slice of the time slot layer, so as to reduce network congestion queuing and resource robbing generated when a plurality of deterministic data streams arrive at the end node at the same time in a time dimension.
2. The deterministic network oriented network slice generation and dynamic configuration system of claim 1 wherein the information acquisition module comprises a deterministic network dynamic measurement module and a deterministic network static measurement module; wherein,
the deterministic network dynamic measurement module is used for measuring the link load state and the network flow state in the network, and the obtained flow in the deterministic network is represented in the form of a flow matrix, and the bandwidth, time delay and jitter information required by the deterministic flow are obtained;
and the deterministic network static measurement module is used for measuring static information in a network, including router connection condition, port condition and link bandwidth information.
3. The deterministic network oriented network slice generation and dynamic configuration system of claim 2 wherein the network transport module comprises a deterministic network flow table issuing module and a deterministic network slice configuration module, wherein,
the deterministic network flow table issuing module is used for carrying out port configuration and flow table issuing on a network layer through flow table information;
the deterministic network slice configuration module is used for carrying out link bandwidth allocation and configuration of the deterministic switch through slice information output by the network slice module.
4. The deterministic network-oriented network slice generation and dynamic configuration method is characterized by comprising the following steps:
step 1: collecting a deterministic flow data set, generating a deterministic flow matrix and required jitter and time delay information, inputting the obtained deterministic flow matrix and related information into a depth model according to a time sequence to perform flow prediction, and performing deterministic network flow prediction; step 1-1: collecting historical flow data in a network by utilizing output data of a data acquisition module, and classifying the historical flow data according to different deterministic service requirements to form a deterministic network flow matrix and a non-deterministic network flow matrix; step 1-2: modeling the formed flow matrix to obtain a flow matrix containing jitter and time delay information related information required by deterministic flow; step 1-3: then, predicting deterministic traffic through an RNN (RNN round robin) network algorithm to form a predicted traffic matrix of the deterministic traffic related to time sequence; the steps 1-3 specifically comprise the following steps: building an RNN model according to a deterministic flow matrix generated by a historical data set, inputting the deterministic flow matrix with time correlation and deterministic flow demand data, automatically building an RNN model of network flow, training, and outputting a flow matrix of a predicted deterministic network in a later period after model calculation after training is completed;
step 2: inputting the obtained deterministic network flow prediction data set with bandwidth, time delay and jitter requirements and topology basic information in a network into a network slicing algorithm based on reinforcement learning to form bandwidth and port information required by deterministic slicing, and configuring the deterministic network slicing according to the bandwidth and port information required by deterministic slicing; step 2-1: collecting the link bandwidth size in a deterministic network through a network resource static collection module; step 2-2: inputting the data set of the deterministic network traffic matrix obtained by the RNN algorithm prediction in the step 1 into a reinforcement learning network slicing algorithm, and generating network slicing module configuration parameters of port numbers and reserved bandwidths of deterministic network resources which can be reserved; step 2-3: the obtained configuration parameters are subjected to resource reservation on space bandwidth in a network slice configuration module of the transmission module, so as to form an initial network slice; the step 2-3 specifically comprises the following steps: the obtained configuration parameters are subjected to resource reservation on space bandwidth in a network slice configuration module of the transmission module, so as to form an initial network slice; the step 2-3 specifically comprises the following steps: the reserved data of the deterministic network slice generated by reinforcement learning is configured into a deterministic network through a network slice configuration module of a network transmission module, wherein the configuration of limiting the bandwidth of a link is mainly performed through a queue mechanism in a deterministic exchanger, and the link bandwidth of the deterministic network and the link bandwidth of a non-deterministic network are subjected to virtualization isolation, so that the bandwidths between the deterministic network and the non-deterministic network are not interfered with each other;
step 3: the bandwidth scheduling algorithm strategy of the time slot level is applied to the deterministic network slice, the resource scheduling in the deterministic slice of the time slot level is carried out, and network congestion queuing and resource preemption generated when a plurality of deterministic data streams arrive at the end node simultaneously are reduced from the time dimension;
step 4: calculating network benefits according to link bandwidth, topology and deterministic flow information in a network, and returning to the step 1 to regenerate and dynamically configure a slicing strategy if the network benefits do not meet a certain design threshold; otherwise, keep collection of link bandwidth, topology and deterministic traffic information in the network.
5. The deterministic network oriented network slice generation and dynamic configuration method according to claim 4, wherein said step 3 comprises the sub-steps of:
step 3-1: in the step 2, two virtual sub-networks of deterministic network slices and non-deterministic network slices are generated through deterministic network configuration, and real-time deterministic network traffic and types, time delays and jitter of the deterministic network traffic and the types, time delays and jitter are extracted and counted through a network state measurement module;
step 3-2: quantifying the bandwidth in the deterministic network slice into time slots according to different types of deterministic traffic, dividing the time slots in the deterministic network for the flexibly-entering deterministic traffic, and carrying out resource scheduling on the bandwidth in the deterministic slice through a time slot algorithm;
step 3-3: and according to the division result of the time slot algorithm, adjusting the injection time of the scheduling flow in the network transmission module to realize the resource scheduling algorithm in the deterministic network slice based on the time slot.
6. The deterministic network oriented network slice generation and dynamic configuration method according to claim 5, wherein said step 4 comprises the sub-steps of:
step 4-1: the method comprises the steps of monitoring the state of deterministic traffic in a deterministic network dynamic measurement module;
step 4-2: and (3) processing the network state data monitored in the step (4-1), analyzing the network state condition, and regenerating and configuring the deterministic network slice when the bandwidth, time delay and jitter average service guarantee rate of the deterministic traffic are smaller than a design threshold value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111424669.XA CN114268537B (en) | 2021-11-26 | 2021-11-26 | Deterministic network-oriented network slice generation and dynamic configuration system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111424669.XA CN114268537B (en) | 2021-11-26 | 2021-11-26 | Deterministic network-oriented network slice generation and dynamic configuration system and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114268537A CN114268537A (en) | 2022-04-01 |
| CN114268537B true CN114268537B (en) | 2023-12-22 |
Family
ID=80825645
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111424669.XA Active CN114268537B (en) | 2021-11-26 | 2021-11-26 | Deterministic network-oriented network slice generation and dynamic configuration system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114268537B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115834430A (en) * | 2022-11-16 | 2023-03-21 | 国网江苏省电力有限公司信息通信分公司 | Time sensitive network testing method, testing bed and storage medium |
| CN115955403B (en) * | 2022-12-06 | 2023-07-18 | 煤炭科学技术研究院有限公司 | Mining network slicing method and system |
| CN115987814B (en) * | 2022-12-23 | 2023-10-31 | 众芯汉创（北京）科技有限公司 | Deterministic network computing system for 5GURLLC scene delay |
| CN115632925B (en) * | 2022-12-26 | 2023-03-14 | 北京交通大学 | Time-certainty fusion network architecture and data communication method |
| CN118524413A (en) * | 2024-07-23 | 2024-08-20 | 北京华迅通信技术有限公司 | Big data-based 5G communication network operation and maintenance platform |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107659419A (en) * | 2016-07-25 | 2018-02-02 | 华为技术有限公司 | Network dicing method and system |
| CN107743100A (en) * | 2017-09-30 | 2018-02-27 | 重庆邮电大学 | A kind of online adaptive network section virtual resource allocation method based on traffic forecast |
| CN109600259A (en) * | 2018-12-11 | 2019-04-09 | 浙江工商大学 | A kind of real-time Transmission mechanism based on software definable |
| CN110048869A (en) * | 2018-01-16 | 2019-07-23 | 中国科学院沈阳自动化研究所 | Resource allocation methods and system towards industrial time-sensitive software defined network |
| CN110519783A (en) * | 2019-09-26 | 2019-11-29 | 东华大学 | 5G network based on enhancing study is sliced resource allocation methods |
| EP3684137A1 (en) * | 2019-01-21 | 2020-07-22 | Nokia Solutions and Networks GmbH & Co. KG | Providing connectivity service to multiple tsn domains by a single 3gpp network |
| CN111866953A (en) * | 2019-04-26 | 2020-10-30 | 中国移动通信有限公司研究院 | Network resource allocation method, device and storage medium |
| CN112511329A (en) * | 2020-06-12 | 2021-03-16 | 中兴通讯股份有限公司 | Information notification method and device and readable medium |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115580905A (en) * | 2018-08-14 | 2023-01-06 | 华为技术有限公司 | Communication method and device based on time perception service quality |
-
2021
- 2021-11-26 CN CN202111424669.XA patent/CN114268537B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107659419A (en) * | 2016-07-25 | 2018-02-02 | 华为技术有限公司 | Network dicing method and system |
| CN107743100A (en) * | 2017-09-30 | 2018-02-27 | 重庆邮电大学 | A kind of online adaptive network section virtual resource allocation method based on traffic forecast |
| CN110048869A (en) * | 2018-01-16 | 2019-07-23 | 中国科学院沈阳自动化研究所 | Resource allocation methods and system towards industrial time-sensitive software defined network |
| CN109600259A (en) * | 2018-12-11 | 2019-04-09 | 浙江工商大学 | A kind of real-time Transmission mechanism based on software definable |
| EP3684137A1 (en) * | 2019-01-21 | 2020-07-22 | Nokia Solutions and Networks GmbH & Co. KG | Providing connectivity service to multiple tsn domains by a single 3gpp network |
| CN111866953A (en) * | 2019-04-26 | 2020-10-30 | 中国移动通信有限公司研究院 | Network resource allocation method, device and storage medium |
| CN110519783A (en) * | 2019-09-26 | 2019-11-29 | 东华大学 | 5G network based on enhancing study is sliced resource allocation methods |
| CN112511329A (en) * | 2020-06-12 | 2021-03-16 | 中兴通讯股份有限公司 | Information notification method and device and readable medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114268537A (en) | 2022-04-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114268537B (en) | Deterministic network-oriented network slice generation and dynamic configuration system and method | |
| Zhang et al. | Elastic and reliable bandwidth reservation based on distributed traffic monitoring and control | |
| CN109614215B (en) | Deep reinforcement learning-based stream scheduling method, device, equipment and medium | |
| CN107454015B (en) | OF-DiffServ model-based QoS control method and system | |
| CN112733303B (en) | Multi-strategy industrial TSN shaper modeling method based on deterministic network algorithm | |
| CN115190082B (en) | Traffic flow scheduling method and device of TSN (traffic stream network) | |
| Li et al. | OPTAS: Decentralized flow monitoring and scheduling for tiny tasks | |
| Kogan et al. | Multi-queued network processors for packets with heterogeneous processing requirements | |
| CN105700940B (en) | A kind of statistical multiplex method of scheduler and scheduler | |
| CN115766884A (en) | Computing task processing method, device, equipment and medium | |
| CN114257554B (en) | Scheduling method for improving bandwidth utilization rate of BE (flow element) stream of TSN (traffic channel network) | |
| Manfredi | Decentralized queue balancing and differentiated service scheme based on cooperative control concept | |
| CN117834552A (en) | Communication queue scheduling optimization method, device, equipment and medium | |
| Kozlovskiy et al. | Development of a modified method of network traffic forming | |
| KR20120055947A (en) | Method and apparatus for providing Susbscriber-aware per flow | |
| Cerroni et al. | Analytical model of quality of service scheduling for optical aggregation in data centers | |
| Mifdaoui et al. | Real-time characteristics of switched Ethernet for" 1553b"-embedded applications: Simulation and analysis | |
| CN116095175B (en) | Data flow scheduling method and device for grid edge computing system | |
| Fang et al. | Token regulated credit based shaper for time sensitive networks | |
| Boussada et al. | A Flow-Level Performance Evaluation of Elastic Traffic Under Low Latency Queuing System | |
| Sarwar et al. | Optical burst switched (OBS) systems with segmentation-based dropping. | |
| Ridwana et al. | Quality-of-Service Performance Comparison: Machine Learning Regression and Classification-Based Predictive Routing Algorithm | |
| Van Bemten | Design, Implementation, and Evaluation of Mechanisms for Predictable Latency in Programmable Networks | |
| Lozhkovskyi et al. | UDC 621. 391 method for approximating the distribution function of the states in the single-channel system with a self-similar traffic | |
| CN114679419A (en) | Demand response service scheduling method and system based on weighted fair queue |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |