CN117596158A - Train communication network flow scheduling optimization method based on PMOA algorithm - Google Patents
Train communication network flow scheduling optimization method based on PMOA algorithm Download PDFInfo
- Publication number
- CN117596158A CN117596158A CN202311343901.6A CN202311343901A CN117596158A CN 117596158 A CN117596158 A CN 117596158A CN 202311343901 A CN202311343901 A CN 202311343901A CN 117596158 A CN117596158 A CN 117596158A
- Authority
- CN
- China
- Prior art keywords
- algorithm
- scheduling
- flow
- communication network
- traffic
- 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.)
- Pending
Links
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 91
- 238000004891 communication Methods 0.000 title claims abstract description 73
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000005457 optimization Methods 0.000 title claims abstract description 56
- 230000005540 biological transmission Effects 0.000 claims abstract description 60
- 230000008569 process Effects 0.000 claims abstract description 25
- 238000012163 sequencing technique Methods 0.000 claims abstract description 8
- 230000004044 response Effects 0.000 claims description 32
- 210000000349 chromosome Anatomy 0.000 claims description 30
- 239000011159 matrix material Substances 0.000 claims description 18
- 238000012545 processing Methods 0.000 claims description 10
- 235000008694 Humulus lupulus Nutrition 0.000 claims description 5
- 238000002955 isolation Methods 0.000 claims description 5
- 108090000623 proteins and genes Proteins 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 2
- 230000002068 genetic effect Effects 0.000 abstract description 3
- 206010063385 Intellectualisation Diseases 0.000 abstract description 2
- 230000000875 corresponding effect Effects 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 5
- 238000010606 normalization Methods 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 238000004088 simulation Methods 0.000 description 3
- 230000003864 performance function Effects 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000001174 ascending effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 239000002243 precursor Substances 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/14—Network analysis or design
- H04L41/142—Network analysis or design using statistical or mathematical methods
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/14—Network analysis or design
- H04L41/145—Network analysis or design involving simulating, designing, planning or modelling of a network
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/12—Shortest path evaluation
- H04L45/124—Shortest path evaluation using a combination of metrics
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/10—Flow control; Congestion control
- H04L47/12—Avoiding congestion; Recovering from congestion
- H04L47/125—Avoiding congestion; Recovering from congestion by balancing the load, e.g. traffic engineering
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/10—Flow control; Congestion control
- H04L47/24—Traffic characterised by specific attributes, e.g. priority or QoS
- H04L47/2425—Traffic characterised by specific attributes, e.g. priority or QoS for supporting services specification, e.g. SLA
- H04L47/2433—Allocation of priorities to traffic types
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/12—Protocols specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks
Landscapes
- Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Mathematical Optimization (AREA)
- Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Probability & Statistics with Applications (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Analysis (AREA)
- General Physics & Mathematics (AREA)
- Algebra (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention relates to the technical field of train communication networks, in particular to a traffic scheduling optimization method of a train communication network based on a PMOA algorithm. Meanwhile, a PMOA algorithm is designed according to a physical constraint condition and a traffic transmission constraint condition of an actual train communication network switching node, a second generation non-dominant sorting genetic algorithm which is jointly improved and a joint scheduling algorithm based on multi-level routing. The algorithm can optimize flow sequencing, and can select the flow route while balancing the loads of different paths in the train communication network, so that the flow scheduling process can be planned more efficiently. The method and the system effectively solve the problems that part of service flows cannot be scheduled and the instantaneity is poor under the background of rapid increase of data volume after train intellectualization.
Description
Technical Field
The invention relates to the technical field of train communication networks, in particular to a train communication network flow scheduling optimization method based on a PMOA algorithm.
Background
With the rapid development of the intellectualization of rail traffic equipment, the variety and the quantity of traffic in a train communication network show a trend of remarkable increase, and the real-time performance and the reliability of the train communication network are more severely required. In recent years, the data transmission rate of a train communication network is successfully improved by introducing an industrial Ethernet technology, but the industrial Ethernet technology still faces the problems of insufficient protocol interoperability, uncertainty of transmission delay and the like, and the high requirements of an intelligent train on the real-time performance and the reliability of data transmission are difficult to meet.
In this context, time-sensitive networks (TimeSensitiveNetworking, TSN) have attracted considerable attention in the field of train communications due to their advantages of excellent transmission rates, deterministic latency, and protocol interoperability.
However, in the research of the real-time flow scheduling optimization method of the train communication network based on the TSN, most of the research is limited to considering only a single real-time index when constructing an objective function model, and even if the research which considers a plurality of optimization targets occurs, the relation among the optimization targets is not concerned; in the model optimization process, although most of scheduling algorithms optimize the routing problem of the flow, the method is not combined with the actual condition of path congestion in a train communication network, and the complex relationship between the load and the routing length is not considered; in addition, although the ordering of the flows is a key direction that can improve the optimization performance of the algorithm, there is little research interest.
Disclosure of Invention
The invention aims to provide a traffic scheduling optimization method of a train communication network based on a PMOA algorithm, which balances the path load and the routing length of traffic in the train communication network by optimizing traffic sequencing, thereby improving the real-time index of the traffic on the premise of ensuring the traffic scheduling success rate.
In order to achieve the above purpose, the invention provides a train communication network traffic scheduling optimization method based on a PMOA algorithm, which comprises the following steps:
s1: establishing a train communication network topology model based on TSN, and defining various parameters in the train communication network topology model, including the performance parameters of a TSN switch, the performance parameters of a terminal node and link parameters;
s2: establishing a traffic scheduling model based on the train communication network topology model, defining performance parameters of communication traffic, and defining corresponding constraints based on a traffic scheduling process;
s3: establishing an objective function model taking scheduling success rate, average response time delay and total transmission time length as optimization indexes on the basis of the flow scheduling model and constraint, and considering optimization priorities among indexes, wherein the scheduling success rate has the highest optimization priority among the indexes, and the optimization priorities of other real-time indexes are the same;
s4: taking the improved NSGA-II algorithm as a main body framework, combining a JMRSS algorithm for calculating fitness functions of all chromosomes, planning route allocation and offset time slots aiming at the sending sequence of traffic of a train communication network, and designing the whole flow of the PMOA algorithm;
s5: and (3) carrying out scheduling optimization on the train communication network flow defined in the step (S2) in the PMOA algorithm to obtain a group of Pareto optimal solutions with highest scheduling success rate and excellent real-time performance.
The method comprises the steps of establishing a train communication network topology model based on TSN, defining various parameters in the train communication network topology model, including performance parameters of a TSN switch, performance parameters of a terminal node and link parameters, and further comprising:
the performance parameters of the TSN switch comprise the number of nodes, the number of node ports and the processing rate of the node ports of the TSN switch; the terminal node performance parameters comprise the port number of the terminal node and the port transmission rate; the link parameters include link bandwidth and number of links.
The method comprises the steps of establishing a traffic scheduling model based on the train communication network topology model, defining performance parameters of communication traffic, and defining corresponding constraints based on a traffic scheduling process, wherein the steps further comprise:
flow f i The performance parameters of (1) are composed of five-tuple { f i .src,f i .des,f i .siz,f i .prd,f i Ddl }, where f i .src,f i Des represents the source address and destination address of the stream, f i Size represents the data size of the stream, f i Prd the cycle size of the stream, f i Ddl represents the deadline of the stream; any one of the streams f i Features in scheduling are all dyadicRepresentation of->Representing stream f i On the path [ v a ,v b ]Lower offset, +_>Representing stream f i Is on path v a ,v b ]Length of transmission time per stream f i Is routed by { [ v ] i1 ,v i2 ],...,[v i(ni-1) ,v i(ni) ]The specific scheduling process of the flow in the route is related to the constraint of the flow in the transmission process, and the constraint of the flow in the transmission process is divided into no-waiting constraint, link constraint, frame constraint and delay constraintFrame isolation constraints.
The method comprises the steps of establishing an objective function model taking a scheduling success rate, average response time delay and total transmission time length as optimization indexes on the basis of the flow scheduling model and constraint, and considering optimization priorities among indexes, wherein the scheduling success rate has the highest optimization priority among the indexes, and other real-time indexes have the same optimization priority, and the steps further comprise:
the scheduling success rate function satisfies:
where n is the flow F in the flow set F i Num.sub is the number of scheduled successful streams;
the average response time delay is the average time of returning response after processing after a certain device or node in the network system receives the request, and the specific objective function of the average response time delay is as follows:
in the formula, hop represents the flow f i The number of hops of the route,representing the total transmission time length of each flow in the route, which is flow f i An offset at the start of transmission in the first period;
the total transmission time length is the total time which passes from the beginning of data transmission from the transmitting end to the receiving end and the processing is completed, and the specific objective function thereof is as follows:
the improved NSGA-II algorithm is used as a main body framework, a JMRSS algorithm for calculating fitness functions of all chromosomes is combined, route allocation and offset time slots are planned according to the sending sequence of traffic of a train communication network, and the whole flow of the PMOA algorithm is designed, wherein the method further comprises the following steps:
the improved NSGA-II algorithm adopts an integer coding mode, the arrangement sequence of the streams is used as genes of chromosomes to generate one hundred initial chromosomes, part of the initial chromosomes comprise preset sequences, the chromosomes are respectively sequenced according to the size of a period and the size of flow, the chromosomes possibly comprise better solutions, the searching efficiency is improved, and the rest of the chromosomes are randomly generated;
taking three objective functions of scheduling success rate, average response time delay and total transmission time length as fitness functions, and calculating fitness function values of all chromosomes by adopting a JMRSS algorithm;
screening the chromosomes using improved non-dominant ordering;
obtaining filial generation through inheritance, intersection and variation, obtaining a new population by adopting elite strategy, calculating the crowding degree of the new population and improving non-dominant order, repeating the above processes until the iteration frequency requirement is met, and constructing a complete PMOA algorithm.
Wherein, the JMRSS algorithm will plan a group of traffic from two aspects of route and dispatch, the steps include:
the JMRSS algorithm obtains all possible route modes of the current flow, and selects a route with the smallest load in each route grade according to a load matrix based on TCN topology to add into an alternative route set;
sequentially attempting to schedule the traffic by traversing the alternative route set, wherein the traffic is scheduled in a greedy manner, the greedy schedule sets the initial offset time of the traffic to 0, and if the traffic collides with the previous traffic, the traffic is offset until the traffic is scheduled successfully;
after completing the dispatching attempts of all the alternative routes, selecting the successfully dispatched routes, and comparing the real-time indexes of the successfully dispatched routes, wherein the route with the optimal real-time index is the route of the current flow;
the load matrix information based on the TCN topology is updated for the route selected by the current flow, and the process is continuously repeated for the next flow until the process is finished.
The load matrix based on TCN topology defines the load of each port of the TSN switch, can reflect the congestion degree of the path, and aims at different nodes v a And v b Defining an initialized load matrix m:
after initialization, each time a route of a traffic is determined, the corresponding element in the load matrix m increases the corresponding load:
when flow f i Taking ofIn the routing mode of (a) the total load of all occupied ports +.>
In the middle ofFlow f in i Routing means of->For measuring routing->Is a congestion degree of the vehicle.
Wherein the routing levelThe definition is as follows:
in the method, in the process of the invention, for flow f i From arbitrary terminal equipment v a To v b A set of all routing means in +.>For any one of the routing means in the set, < >>Representing all route classes->Set of->Representing route->Is>Representing a route set->Is the minimum number of route hops.
Wherein, the improved rapid non-dominant ranking is to rank the objective functions, in the rapid non-dominant ranking, the scheduling success rate between individuals is preferentially compared, and the performance of the other two real-time objective functions is paid attention to, when the individual x 1 Dominating individual x 2 When the method is used, the following steps are needed:
f 1 (x 1 )＞f 1 (x 2 )∨
f 1 (x 1 )＝f 1 (x 2 )∧f i (x 1 )＜f i (x 2 )∧f j (x 1 )≤f j (x 2 )∨
f 1 (x 1 )＝f 1 (x 2 )∧f j (x 1 )＜f j (x 2 )∧f i (x 1 )≤f i (x 2 )
wherein F is 1 (x i ) Represents the scheduling success rate of the ith individual, F 2 (x i ) Represents the average response time delay of the ith individual, F 3 (x i ) Representing the total transmission duration of the ith individual;
while when individual x 1 With individual x 2 Under the condition of no mutual control, the following needs to be satisfied:
f 1 (x 1 )＝f 1 (x 2 )∧f i (x 1 )＜f i (x 2 )∧f j (x 1 )＞f j (x 2 )∨
f 1 (x 1 )＝f 1 (x 2 )∧f j (x 1 )＜f j (x 2 )∧f i (x 1 )＞f i (x 2 )∨
f 1 (x 1 )＝f 1 (x 2 )∧f j (x 1 )＝f j (x 2 )∧f i (x 1 )＝f i (x 2 )
under the premise, when x 1 One of the other two objective functions performs better than x 2 One is weaker than x 2 Or the performance on the two other objective functions is equal to x 2 When equal, satisfy individual x 1 With individual x 2 Are not mutually dominant.
The train communication network traffic defined in the step S2 is scheduled and optimized in the PMOA algorithm to obtain a Pareto optimal solution with the highest scheduling success rate and excellent real-time performance, and the steps further include:
the optimization scheme comprises the sequencing, routing and time slot offset of each flow; the optimization result comprises the scheduling success rate of a group of flows, average response time delay and total transmission duration.
The invention discloses a train communication network flow scheduling optimization method based on a PMOA algorithm, which adopts a TSN technology, and fully considers the priority order among optimization targets on the basis of researching optimization targets comprising flow scheduling success rate, average response time delay and total transmission time length. Meanwhile, a PMOA algorithm is designed according to physical constraint conditions and traffic transmission constraint conditions of an actual train communication network switching node, a second generation Non-dominant sorting genetic algorithm (Non-dominated Sorting Genetic Algorithm II, NSGA-II) which is improved in a combined mode and a Joint Multi-level Routing Selection and Scheduling (JMRSS) algorithm based on Multi-level routing. The algorithm can optimize flow sequencing, and can select the flow route while balancing the loads of different paths in the train communication network, so that the flow scheduling process can be planned more efficiently. Compared with the traditional scheduling method, the flow scheduling method provided by the invention can obviously improve the scheduling success rate and generate a group of Pareto solution sets with better real-time performance through the simulation of an example, thereby effectively solving the problems that part of service flows cannot be scheduled and the real-time performance is poor under the background of the rapid increase of the data quantity after the intelligent train.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below.
Fig. 1 is a flowchart of a method for optimizing traffic scheduling of a train communication network based on a PMOA algorithm of the present invention.
Fig. 2 is a network topology model of the train of the present invention.
Fig. 3 is a graph comparing scheduling success rates of five scheduling schemes according to an embodiment of the present invention.
Fig. 4 is a graph of total transmission duration and average response time delay for five scheduling schemes according to an embodiment of the present invention.
FIG. 5 is a graph of overall performance versus solution for five scheduling schemes in an embodiment of the present invention.
Detailed Description
The following detailed description of embodiments of the invention, examples of which are illustrated in the accompanying drawings and, by way of example, are intended to be illustrative, and not to be construed as limiting, of the invention.
Referring to fig. 1 to 5, fig. 1 is a flowchart of a traffic scheduling optimization method for a train communication network based on a PMOA algorithm according to the present invention. Fig. 2 is a network topology model of the train of the present invention. Fig. 3 is a graph comparing scheduling success rates of five scheduling schemes according to an embodiment of the present invention. Fig. 4 is a graph of total transmission duration and average response time delay for five scheduling schemes according to an embodiment of the present invention. FIG. 5 is a graph of overall performance versus solution for five scheduling schemes in an embodiment of the present invention. The invention provides a train communication network flow dispatching optimization method based on a PMOA algorithm, which comprises the following steps: the method comprises the following steps:
s1: and establishing a train communication network topology model based on TSN, and defining various parameters in the train communication network topology model, including the performance parameters of the TSN switch, the performance parameters of the terminal nodes and the link parameters.
Specifically, the performance parameters of the TSN switch mainly include the number of nodes, the number of node ports and the processing rate of the node ports of the TSN switch, which all have a storage forwarding type switching technology; the terminal node performance parameters mainly comprise the port number of the terminal node and the port transmission rate; the link parameters include link bandwidth and number of links, and the links all employ duplex communication technology.
S2: and establishing a flow scheduling model based on the train communication network topology model, defining performance parameters of communication flow, and defining corresponding constraints based on a flow scheduling process.
Specifically, the performance parameter of the flow fi can be defined by five-tuple { f i .src,f i .des,f i .siz,f i .prd,f i Ddl }, where f i .src,f i Des represents the source address and destination address of the stream, f i Size represents the data size of the stream, f i Prd the cycle size of the stream, f i Ddl represents the deadline of the stream; any one of the streams f i Features in the schedule may be composed of two tuplesThe representation is: wherein->Representing the flow f i On the path [ v a ,v b ]Lower offset, +_>Representing the flow f i Is on path v a ,v b ]The length of transmission time below. Each stream f i Can be routed by { [ v ] i1 ,v i2 ],...,[v i(ni-1) ,v i(ni) ]And } represents. The specific scheduling process of the flow in the route is related to the constraint of the flow in the transmission process.
The constraints of the stream in the transmission process can be divided into no-wait constraints, link constraints, frame constraints, delay constraints and frame isolation constraints:
the wait-free constraint satisfies:
v in a ,v b ,v c Is any three switches or end nodes in node set V and needs to satisfy [ V ] b ,v c ]Is [ v ] a ,v b ]The next hop. R represents a real set. The constraint ensures that the flow does not need to waste time on queuing and waiting, and maximizes the throughput and efficiency of the system;
the link constraint satisfies:
wherein F is the set of flow rates, lcm (F j .prd,f i Prd) as stream f j ，f i The least common multiple of cycles. v a ，v b For any two adjacent switches or end nodes in node set V. When the flow is forwarded according to the optimized route, the correct transmission time sequence is needed, and the service flow on the corresponding path can be transmitted on the next link after the transmission of the previous link is completed. Link constraint believesIf a link arrives and transmits a plurality of streams at the same time, the problems of frame loss, large delay jitter and the like can be caused in the transmission process. The link constraint requires that only one flow can pass through the same link at the same time, so that the determined time delay can be ensured, frames of each flow cannot be staggered and confused, and the time correctness of the service flow in path transmission is ensured;
the frame constraint satisfies:
all data frame slot offsets involved in scheduling must not be negative and the scheduling time window for each frame must not exceed the transmission period of the stream. The constraint meets the real-time requirement of stream transmission by limiting the time slot offset of the stream;
the delay constraint satisfies:
wherein [ v (n-1), vn]Representing stream f i The last hop of the route that is being traversed,representing stream f i How many times in the overcycle. The constraint meets the real-time requirements of the stream by limiting the size of the stream response delay.
The frame isolation constraint satisfies:
wherein [ v ] x ,v a ]、[v y ,v a ]Is in link v a ,v b ]Is a single precursor link.And->Respectively show that when passing through the switch v a Time flow f i And flow f j Is selected in the queue. When multiple flows from different sources and of the same priority arrive at the same switch node and are forwarded from the same egress port, the flows can be transmitted simultaneously in the switch if and only if the flows select different queues in the switch. If the same queue is selected and frames of multiple streams arrive at the same time, the order in which frames of the respective streams enter the queue is uncertain. The frames of different streams are likely to be identicalThe transmissions are interleaved in a queue. This can lead to uncertainty in the end-to-end latency. The frame isolation constraint specifies that the same queue of a switch can only store frames from the same stream at the same time. The constraint ensures the certainty of the end-to-end delay by orderly arranging the frames in the switch queues.
S3: and establishing an objective function model taking the scheduling success rate, the average response time delay and the total transmission time length as optimization indexes on the basis of the flow scheduling model and the constraint, and considering the optimization priorities among the indexes, wherein the scheduling success rate has the highest optimization priority among the indexes, and the optimization priorities of other real-time indexes are the same.
Specifically, maximizing the scheduling success rate is a precondition of ensuring real-time performance, optimizing priority is highest, and the function satisfies:
where n is the flow F in the flow set F i Num.sub is the number of scheduled successful streams;
minimizing the average response time delay and the total transmission time length is a real-time embodiment, and the priorities of the two indexes are the same. The average response time delay refers to average time of returning response after processing after a certain device or node in the network system receives a request, and the specific objective function of the average response time delay is as follows:
in the formula, hop represents the flow f i The number of hops of the route,representing the total transmission time length of each flow in the route, which is flow f i Offset at the start of transmission in the first period.
The total transmission duration refers to the total time that passes from the start of data transmission from the transmitting end to the receiving end and the completion of processing, and the specific objective function thereof satisfies:
s4: the improved NSGA-II algorithm is used as a main body framework, the JMRSS algorithm for calculating fitness functions of all chromosomes is combined, and planning is conducted aiming at the sending sequence, route allocation and offset time slots of the traffic of the train communication network, so that the overall flow of the PMOA algorithm is designed.
Specifically, the improved NSGA-II adopts an integer coding mode, and the sequencing mode of train communication network traffic is described by the gene sequence of chromosomes. For example, when there are five flows in the network, the chromosome string [53412] can represent that the five flows in the default ordering are rearranged in the order of the fifth, third, fourth, first, and second. After the coding mode is determined, NSGA-II generates one hundred initial chromosomes, part of the initial chromosomes contains preset sequences, the chromosomes are respectively sequenced according to the size of a period and the size of flow, the chromosomes possibly contain better solutions, the searching efficiency is accelerated, and the rest chromosomes are randomly generated;
and taking three objective functions of scheduling success rate, average response time delay and total transmission time length as fitness functions, and calculating fitness function values of all chromosomes by adopting a JMRSS algorithm.
The JMRSS algorithm needs to be defined first for the load matrix and the routing level before it is defined. The load matrix based on the TCN topology defines the load of each port of the TSN switch, and can reflect the congestion degree of the path. For different nodes v a And v b Defining an initialized load matrix m:
after initialization, each time a route of a traffic is determined, the corresponding element in the load matrix m increases the corresponding load:
when flow f i Taking ofIn the routing mode of (a) the total load of all occupied ports +.>
In the middle ofFlow f in i Routing means of->For measuring routing->Is a congestion degree of (3);
wherein the routing levelCan be defined as:
in the method, in the process of the invention, for flow f i From arbitrary terminal equipment v a To v b A set of all routing means in +.>Is any routing mode in the collection. />Representing all route classes->Is a set of (3). />Representing route->Is>Representing a route set->Is the minimum number of route hops.
The JMRSS algorithm will plan a set of traffic from both routing and scheduling aspects and calculate fitness values. Firstly, the JMRSS algorithm obtains all possible route modes of the current traffic, selects the route with the smallest load in each route grade according to a load matrix based on TCN topology, adds the route into an alternative route set, and then sequentially tries to schedule the traffic by traversing the alternative route set. The flow dispatching mode adopts greedy dispatching. Greedy scheduling sets the initial offset time of the traffic to 0, and if there is a collision with the previous flow, the offset is performed until the traffic is scheduled successfully. After completing the dispatching attempts of all the alternative routes, the successfully dispatched routes are selected, the real-time indexes are compared, and the route with the optimal real-time index is the route of the current flow. Then updating the load matrix information based on TCN topology for the route selected by the current flow, and continuing repeating the process for the next flow until the process is finished;
the chromosomes are screened using improved non-dominant ordering, wherein the improved rapid non-dominant ordering does not focus on three objective functions simultaneously, but ranks the objective functions. In the rapid non-dominant sorting, firstly, the scheduling success rate between individuals is preferentially compared, and then the performance quality of the other two real-time objective functions is paid attention to. When individual x 1 Dominating individual x 2 When the method is used, the following steps are needed:
f 1 (x 1 )＞f 1 (x 2 )∨
f 1 (x 1 )＝f 1 (x 2 )∧f i (x 1 )＜f i (x 2 )∧f j (x 1 )≤f j (x 2 )∨
f 1 (x 1 )＝f 1 (x 2 )∧f j (x 1 )＜f j (x 2 )∧f i (x 1 )≤f i (x 2 )
wherein F is 1 (x i ) Represents the scheduling success rate of the ith individual, F 2 (x i ) Represents the average response time delay of the ith individual, F 3 (x i ) Representing the total transmission duration of the ith individual. The formula shows that when the individual x 1 Dominating individual x 2 When x is 1 The scheduling success rate of (a) is required to be larger than x 2 The method comprises the steps of carrying out a first treatment on the surface of the Or in the two target functions, x is the same as the scheduling success rate 1 At least one is better than x 2 On the premise that another objective function is better than or equal to x in performance 2 。
While when individual x 1 With individual x 2 Under the condition of no mutual control, the following needs to be satisfied:
f 1 (x 1 )＝f 1 (x 2 )∧f i (x 1 )＜f i (x 2 )∧f j (x 1 )＞f j (x 2 )∨
f 1 (x 1 )＝f 1 (x 2 )∧f j (x 1 )＜f j (x 2 )∧f i (x 1 )＞f i (x 2 )∨
f 1 (x 1 )＝f 1 (x 2 )∧f j (x 1 )＝f j (x 2 )∧f i (x 1 )＝f i (x 2 )
the formula shows that the precondition that the individuals are not mutually controlled is that the scheduling success rates among the individuals must be equal. Under the premise, when x 1 One of the other two objective functions performs better than x 2 One is weaker than x 2 Or the performance on the two other objective functions is equal to x 2 When equal, satisfy individual x 1 With individual x 2 Are not mutually dominant.
And obtaining filial generation through heredity, crossing and mutation. And a new population is obtained by adopting elite strategy. The new population is subjected to congestion degree calculation and improved non-dominant ranking. Repeating the above process until the iteration frequency requirement is met, and constructing a complete PMOA algorithm.
S5: and (3) carrying out scheduling optimization on the train communication network flow defined in the step (S2) in the PMOA algorithm to obtain a group of Pareto optimal solutions with highest scheduling success rate and excellent real-time performance.
Specifically, the optimization scheme comprises sequencing, routing and time slot offset of each flow; the optimization result comprises the scheduling success rate of a group of flows, average response time delay and total transmission duration.
In the invention, the train communication network topology based on the TSN switch is built according to the train standard protocol IEC61375-3-4 by referring to the actual communication scene of the train communication network. As shown in fig. 2, the train communication network topology is composed entirely of 16 TSN switches (CS) and 8 terminal devices (EDs). The topology takes the form of a dual bus, the scalability of which is illustrated in terms of two grouped ethernet networks (ECNs), and 4 train backbone nodes (ETBN) are employed for communication between the grouped ethernet networks. Based on the topology, different numbers of flows are randomly selected and uniformly distributed between 0 and 500, and communication scenes of different scales on a train are simulated. Topology information and alternative data parameter information used in the simulation are shown in table 1.
Table 1 topology information and data parameters
| Data name | Parameter value | Data name | Parameter value |
| Number of links | 38 | Number of streams | 0～500 |
| Number of switches | 16 | Frame size | 64～1500B |
| Synchronization accuracy | 1ms | Period of flow | 10～80ms |
| Bandwidth of a communication device | 100Mbps | Cut-off time | 10ms |
Based on the above scene, the invention adopts four different flow scheduling algorithms, and proves the superiority of the PMOA algorithm by comparing the performance of the scheduling success rate, the real-time index and the like with the PMOA algorithm. The first algorithm is the flow-aware based NW-TAS scheduling algorithm (the flow-awareNW-TASschedulingalgorithm, FANS); the second is a graph DOC aware based stream partitioning algorithm (DASP); the third is a scheduling algorithm (LoadBalancedRouting, LBR) that considers maximum load balancing; the last one is a periodic route aware algorithm (Period-AwareRoutingalgorithm, PAR) that optimizes the flow ordering in a packet ordering manner while taking into account the flow and Period combinability. The simulation tool for the algorithm was MATLABR2020a with hardware of 3.70GHz and 32GBRAM intel bori 9-10900X processors.
In the invention, the scheduling success rate obtained by scheduling the flows in the same matrix period based on five different algorithms is shown in figure 3. The scheduling success rate is an important factor, which may be affected by various factors. These factors are difficult to parse and analyze individually, so only a few conclusions can be drawn about how the factors affect the scheduling success rate by the algorithm. First, FANS, PAR, and DASP all consider the flow ordering problem. Fass considers best descending ordering performance according to the size of the data, PAR considers best ascending ordering performance according to the traffic cycle without a mutual quality cycle, and DASP does not explicitly account for the flow ordering, but the incremental scheduling process after grouping the traffic is a set of forms that account for the flow ordering. The three algorithms do not conclude the ordering of streams differently, possibly due to the different sets of streams employed. This illustrates that the ordering of streams does not have a fixed choice, where the logical relationship is very complex, likely to change according to changes in the set of streams. At the same time, the necessity of researching the flow ordering is proved; second, both PAR and LBR use quantitative analysis to select routes, but this approach may not be accurate enough when directed to actual scheduling. PMOA overcomes these problems, so scheduling success rate performs best in algorithms.
In the invention, the real-time performance of the flow scheduled based on five different algorithms in the same matrix period is shown in fig. 4, the left graph compares the total transmission time length of the flow after the flow is scheduled, and the right graph compares the average response time delay of the flow after the flow is scheduled. Because the PMOA takes a solution set, the relation between the total transmission time length and the average response time delay in the solution set is not positively correlated, and therefore, the solution set can only guarantee that the value of one objective function is optimal on the premise of ensuring the highest scheduling success rate. When the PMOA curve is drawn, two objective function values of a certain solution are not obtained, but the optimal values of the solutions concentrated in different objective functions are respectively taken to draw the curve. The solution is selected in order to illustrate that the PMOA can provide solutions with variability, and meets various real-time requirements of researchers in the face of practical problems. It is only interesting to consider real-time in case the flows are all scheduled successfully. In this case, the present invention selects the first 250 streams for real-time comparison. As shown in fig. 4, the solution that the PMOA can provide is still optimal for performance in each algorithm, which demonstrates the superiority of the solution set provided by the PMOA.
In the invention, in order to verify the superiority of the PMOA algorithm in all aspects, after the performances of each scheduling algorithm are respectively compared, the multi-objective optimization problem is converted into the single-objective optimization problem, and the performances of each algorithm are comprehensively analyzed. The order of magnitude difference between the total transmission time length and the average response time delay is larger, and the dimension of the multi-objective function can be reduced after normalization processing is carried out on the two objective functions. The normalization formula is as follows:
wherein F is i Representing either one of the two objective functions, F max Refer to the maximum value that can be found in the objective function, F min Refers to the minimum value that can be found in the objective function. The scheduling success rate itself is a fraction from 0 to 1, and the function does not need normalization. After normalization, in order to convert the multi-objective function into a single-objective function and more intuitively look at the performance of each algorithm, the invention designs the following performance functions:
W＝-w 1 ×SUC+w 2 ×AVD+w 3 ×MAK
wherein MAK represents normalized total transmission time length, AVD represents normalized average response time delay, and SUC represents scheduling success rate. In fact, the optimal conditions for these three objective functions are not exactly the same, and it is desirable that the SUC be able to take a maximum value, while the MAK and AVD take a minimum value, so here the invention takes a negative value for the SUC. In this case, the SUC takes the maximum value, and the SUC which takes the negative value takes the minimum value, and the value directions of the two other objective functions are consistent. The coefficients preceding each objective function represent the degree to which the invention is required to place importance. Wherein w is 1 Far greater than w 2 And w is equal to 3 And w is 2 And w is equal to 3 The setting is made according to different requirements for real-time. In this case, the invention solves the solution taken by the PMOA into a single solution, with the aim of optimizing for the performance function W. It may not be the solutions in fig. 3 or fig. 4, but they belong to one solution set. Compared with multiple objective functions, the performance of each algorithm can be seen more intuitively by comparing the single objective functions. The performance of each algorithm is shown in fig. 5, and it is obvious that the PMOA is the best overall performance of the five algorithms.
In the invention, fig. 3, fig. 4 and fig. 5 fully prove that the real-time performance and reliability of the train communication network flow can be remarkably improved by optimizing the sequencing of the flows, the route allocation and the offset time slot, and meanwhile, the PMOA algorithm provided by the invention has more excellent performance compared with other algorithms: the PMOA algorithm not only can provide excellent solution sets with higher scheduling success rate and better real-time performance, but also can provide single solutions with more excellent comprehensive performance, and has better performance in real-time performance and reliability.
The foregoing disclosure is only illustrative of one or more preferred embodiments of the present application and is not intended to limit the scope of the claims hereof, as it is to be understood by those skilled in the art that all or part of the process of implementing the described embodiment may be practiced otherwise than as specifically described and illustrated by the appended claims.
Claims (10)
1. The train communication network flow scheduling optimization method based on the PMOA algorithm is characterized by comprising the following steps of:
s1: establishing a train communication network topology model based on TSN, and defining various parameters in the train communication network topology model, including the performance parameters of a TSN switch, the performance parameters of a terminal node and link parameters;
s2: establishing a traffic scheduling model based on the train communication network topology model, defining performance parameters of communication traffic, and defining corresponding constraints based on a traffic scheduling process;
s3: establishing an objective function model taking scheduling success rate, average response time delay and total transmission time length as optimization indexes on the basis of the flow scheduling model and constraint, and considering optimization priorities among indexes, wherein the scheduling success rate has the highest optimization priority among the indexes, and the optimization priorities of other real-time indexes are the same;
s4: taking the improved NSGA-II algorithm as a main body framework, combining a JMRSS algorithm for calculating fitness functions of all chromosomes, planning route allocation and offset time slots aiming at the sending sequence of traffic of a train communication network, and designing the whole flow of the PMOA algorithm;
s5: and (3) carrying out scheduling optimization on the train communication network flow defined in the step (S2) in the PMOA algorithm to obtain a group of Pareto optimal solutions with highest scheduling success rate and excellent real-time performance.
2. A PMOA algorithm-based train communication network traffic scheduling optimization method according to claim 1, wherein a TSN-based train communication network topology model is established, and each parameter in the train communication network topology model is defined, including a performance parameter of a TSN switch, a performance parameter of a terminal node, and a link parameter, and the steps further include:
the performance parameters of the TSN switch comprise the number of nodes, the number of node ports and the processing rate of the node ports of the TSN switch; the terminal node performance parameters comprise the port number of the terminal node and the port transmission rate; the link parameters include link bandwidth and number of links.
3. A PMOA algorithm-based train communication network traffic scheduling optimization method according to claim 1, wherein a traffic scheduling model is established based on the train communication network topology model, performance parameters of communication traffic are defined, and corresponding constraints are defined based on a traffic scheduling process, the steps further comprising:
flow f i The performance parameters of (1) are composed of five-tuple { f i .src,f i .des,f i .siz,f i .prd,f i Ddl }, where f i .src,f i Des represents the source address and destination address of the stream, f i Size represents the data size of the stream, f i Prd the cycle size of the stream, f i Ddl represents the deadline of the stream; any one of the streams f i Features in scheduling are all dyadicRepresentation of->Representing stream f i On the path [ v a ,v b ]Lower offset, +_>Representing stream f i Is on path v a ,v b ]Length of transmission time per stream f i Is routed by { [ v ] i1 ,v i2 ],...,[v i(ni-1) ,v i(ni) ]And the specific scheduling process of the flow in the route is related to the constraint of the flow in the transmission process, and the constraint of the flow in the transmission process is divided into no-waiting constraint, link constraint, frame constraint, delay constraint and frame isolation constraint.
4. A PMOA algorithm-based traffic scheduling optimization method for a train communication network according to claim 1, wherein a target function model is established based on the traffic scheduling model and constraints, wherein the scheduling success rate, the average response delay and the total transmission time length are used as optimization indexes, and optimization priorities among the indexes are considered, and the optimization priorities of the scheduling success rate are the highest, and the optimization priorities of other real-time indexes are the same, and the steps further comprise:
the scheduling success rate function satisfies:
where n is the flow F in the flow set F i Num.sub is the number of scheduled successful streams;
the average response time delay is the average time of returning response after processing after a certain device or node in the network system receives the request, and the specific objective function of the average response time delay is as follows:
in the formula, hop represents the flow f i Hop count of route，Representing the total transmission time length of each flow in the route, which is flow f i An offset at the start of transmission in the first period;
the total transmission time length is the total time which passes from the beginning of data transmission from the transmitting end to the receiving end and the processing is completed, and the specific objective function thereof is as follows:
5. the PMOA algorithm-based train communication network traffic scheduling optimization method according to claim 1, wherein the improved NSGA-II algorithm is used as a main body frame, and the JMRSS algorithm for calculating fitness functions of each chromosome is combined, and the overall flow of the PMOA algorithm is designed by planning for the transmission sequence, the route allocation and the offset time slots of the train communication network traffic, and the steps further include:
the improved NSGA-II algorithm adopts an integer coding mode, the arrangement sequence of the streams is used as genes of chromosomes to generate one hundred initial chromosomes, part of the initial chromosomes comprise preset sequences, the chromosomes are respectively sequenced according to the size of a period and the size of flow, the chromosomes possibly comprise better solutions, the searching efficiency is improved, and the rest of the chromosomes are randomly generated;
taking three objective functions of scheduling success rate, average response time delay and total transmission time length as fitness functions, and calculating fitness function values of all chromosomes by adopting a JMRSS algorithm;
screening the chromosomes using improved non-dominant ordering;
obtaining filial generation through inheritance, intersection and variation, obtaining a new population by adopting elite strategy, calculating the crowding degree of the new population and improving non-dominant order, repeating the above processes until the iteration frequency requirement is met, and constructing a complete PMOA algorithm.
6. A PMOA algorithm-based train communication network traffic scheduling optimization method according to claim 5, wherein said JMRSS algorithm will plan a set of traffic both from a routing and scheduling aspect, said steps comprising:
the JMRSS algorithm obtains all possible route modes of the current flow, and selects a route with the smallest load in each route grade according to a load matrix based on TCN topology to add into an alternative route set;
sequentially attempting to schedule the traffic by traversing the alternative route set, wherein the traffic is scheduled in a greedy manner, the greedy schedule sets the initial offset time of the traffic to 0, and if the traffic collides with the previous traffic, the traffic is offset until the traffic is scheduled successfully;
after completing the dispatching attempts of all the alternative routes, selecting the successfully dispatched routes, and comparing the real-time indexes of the successfully dispatched routes, wherein the route with the optimal real-time index is the route of the current flow;
the load matrix information based on the TCN topology is updated for the route selected by the current flow, and the process is continuously repeated for the next flow until the process is finished.
7. A PMOA algorithm-based train communication network traffic scheduling optimization method according to claim 6,
the load matrix based on TCN topology defines the load of each port of the TSN switch, can reflect the congestion degree of the path and aims at different nodes v a And v b Defining an initialized load matrix m:
after initialization, each time a route of a traffic is determined, the corresponding element in the load matrix m increases the corresponding load:
when flow f i Taking ofIn the routing mode of (a) the total load of all occupied ports +.>
In the middle ofFlow f in i Routing means of->For measuring routing->Is a congestion degree of the vehicle.
8. A PMOA algorithm-based train communication network traffic scheduling optimization method according to claim 6,
the route classThe definition is as follows:
in the method, in the process of the invention, for flow f i From arbitrary terminal equipment v a To v b Is provided with a set of all routing means,for any one of the routing means in the set, < >>Representing all route classes->Set of->Representing route->Is>Representing a route set->Is the minimum number of route hops.
9. A PMOA algorithm-based train communication network traffic scheduling optimization method according to claim 5,
the improved rapid non-dominant ranking ranks the objective functions, in the rapid non-dominant ranking, the scheduling success rate between individuals is preferentially compared, the performance quality of the other two real-time objective functions is paid attention to, and when an individual x 1 Dominating individual x 2 When the method is used, the following steps are needed:
f 1 (x 1 )＞f 1 (x 2 )∨
f 1 (x 1 )＝f 1 (x 2 )∧f i (x 1 )＜f i (x 2 )∧f j (x 1 )≤f j (x 2 )∨
f 1 (x 1 )＝f 1 (x 2 )∧f j (x 1 )＜f j (x 2 )∧f i (x 1 )≤f i (x 2 )
wherein F is 1 (x i ) Represents the scheduling success rate of the ith individual, F 2 (x i ) Represents the average response time delay of the ith individual, F 3 (x i ) Representing the total transmission duration of the ith individual;
while when individual x 1 With individual x 2 Under the condition of no mutual control, the following needs to be satisfied:
f 1 (x 1 )＝f 1 (x 2 )∧f i (x 1 )＜f i (x 2 )∧f j (x 1 )＞f j (x 2 )∨
f 1 (x 1 )＝f 1 (x 2 )∧f j (x 1 )＜f j (x 2 )∧f i (x 1 )＞f i (x 2 )∨
f 1 (x 1 )＝f 1 (x 2 )∧f j (x 1 )＝f j (x 2 )∧f i (x 1 )＝f i (x 2 )
under the premise, when x 1 One of the other two objective functions performs better than x 2 One is weaker than x 2 Or the performance on the two other objective functions is equal to x 2 When equal, satisfy individual x 1 With individual x 2 Are not mutually dominant.
10. A method for optimizing traffic scheduling of a train communication network based on a PMOA algorithm according to claim 1, wherein the traffic scheduling of the train communication network defined in step S2 is optimized in the PMOA algorithm to obtain a Pareto optimal solution with the highest scheduling success rate and excellent real-time performance, and the steps further include:
the optimization scheme comprises the sequencing, routing and time slot offset of each flow; the optimization result comprises the scheduling success rate of a group of flows, average response time delay and total transmission duration.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311343901.6A CN117596158A (en) | 2023-10-17 | 2023-10-17 | Train communication network flow scheduling optimization method based on PMOA algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311343901.6A CN117596158A (en) | 2023-10-17 | 2023-10-17 | Train communication network flow scheduling optimization method based on PMOA algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117596158A true CN117596158A (en) | 2024-02-23 |
Family
ID=89915602
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311343901.6A Pending CN117596158A (en) | 2023-10-17 | 2023-10-17 | Train communication network flow scheduling optimization method based on PMOA algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117596158A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117896323A (en) * | 2024-03-15 | 2024-04-16 | 苏州大学 | Priority-based data stream base on-line measurement method and system |
-
2023
- 2023-10-17 CN CN202311343901.6A patent/CN117596158A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117896323A (en) * | 2024-03-15 | 2024-04-16 | 苏州大学 | Priority-based data stream base on-line measurement method and system |
| CN117896323B (en) * | 2024-03-15 | 2024-05-31 | 苏州大学 | Priority-based data stream base on-line measurement method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113300960B (en) | Delay deterministic transmission method based on routing scheduling and joint optimization | |
| CN109818865B (en) | SDN enhanced path boxing device and method | |
| CN110493131B (en) | Design method of spatial information network routing strategy under SDN architecture | |
| CN115460130B (en) | Multipath joint scheduling method in time sensitive network | |
| Sayuti et al. | Real-time low-power task mapping in networks-on-chip | |
| CN114143264B (en) | Flow scheduling method based on reinforcement learning under SRv network | |
| CN103036792B (en) | Transmitting and scheduling method for maximizing minimal equity multiple data streams | |
| CN117596158A (en) | Train communication network flow scheduling optimization method based on PMOA algorithm | |
| CN104954278A (en) | Bee colony optimization based network traffic scheduling method under multiple QoS (quality of service) constraints | |
| CN114827021B (en) | Multimedia service flow acceleration system based on SDN and machine learning | |
| Sarasvathi et al. | QoS guaranteed intelligent routing using hybrid PSO-GA in wireless mesh networks | |
| CN116208567A (en) | Method and system for flow scheduling of SDN network resources of cross-domain data center | |
| CN116455824A (en) | Network traffic load balancing method based on reinforcement learning | |
| CN115378876B (en) | Data center network load balancing method in SDN environment | |
| Nair et al. | Particle swarm optimization for realizing intelligent routing in networks with quality grading | |
| Liu et al. | QoS multicast routing based on particle swarm optimization | |
| CN102447625A (en) | Node state control method based on feedback control mechanism of link between nodes | |
| CN116132353A (en) | TSN route selection method based on DDQN | |
| CN102640457B (en) | Differential frame based scheduling for input queued switches | |
| CN116112440B (en) | Time-sensitive network-oriented vehicle-mounted communication architecture multi-objective optimization method and system | |
| Liu et al. | A novel fast multi-objective evolutionary algorithm for QoS multicast routing in MANET | |
| CN116886630B (en) | Time-sensitive network scheduling method and device for improving system bearing traffic | |
| CN117768372B (en) | Method and device for joint optimization of routing and scheduling in time-sensitive network | |
| Liu et al. | Research on flow scheduling of train communication based on time-sensitive network | |
| Sayuti | Simultaneous Multi-Objective Optimisation for Low-Power Real-Time Networks-On-Chip |
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 |