Communication network scheduling method based on improved particle swarm optimization algorithm

Abstract

The invention relates to a communication network scheduling method based on an improved particle swarm optimization algorithm, and belongs to the technical field of communication networks. The method mainly comprises the steps of communication network model construction, communication energy consumption model construction, feasible path exploration, particle swarm node coding, loss function construction, particle swarm initialization, particle swarm iterative optimization and scheduling scheme sorting. According to the method, the defects of long calculation time, low convergence speed and easiness in local optimization are overcome, and rapid and efficient communication network scheduling is realized.

Description

Technical Field

[0001] This invention belongs to the field of communication network technology and relates to a communication network scheduling method based on an improved particle swarm optimization algorithm. Background Technology

[0002] Communication network scheduling is the core mechanism for ensuring efficient and reliable data transmission in modern information society. Its goal is to dynamically allocate resources in complex network environments and optimize the path and rate of information flow to meet multi-dimensional requirements such as real-time performance, security, and quality of service. The core of communication network scheduling lies in the dynamic management of network resources, including bandwidth, channels, and computing power.

[0003] Traditional scheduling algorithms include First-In-First-Out (FIFO), Shortest Job First (SJF), and Highest Response Ratio First (HRRR). Currently, dynamic scheduling strategies are also applied to traditional algorithms, dynamically adjusting resource allocation and task execution order based on network load and performance changes. Multi-path scheduling is also an optimization method that improves network utilization and fault tolerance by selecting multiple paths to transmit data simultaneously.

[0004] Chinese invention patent CN119011674A discloses a communication network scheduling method and system based on a genetic algorithm, relating to the field of communication networks. The method includes the following steps: collecting task request information from all users at the current moment; then, based on the collected source and destination nodes, calculating the set of all feasible paths for each user; then, encoding each path selected by each user as a gene, and performing selection, crossover, and mutation operations; then iterating repeatedly, and finding the optimal solution in the population from the last operation. However, this method mainly has the following problems:

[0005] (1) This method takes a long time to solve for the set of feasible paths. As the number of network nodes increases, global traversal can easily consume too many computing resources and waste time during the initialization process.

[0006] (2) This method has a slow convergence speed. When there are many network nodes, the genetic algorithm is strong in global search but weak in local search, and its efficiency is low when it is close to the optimal solution; moreover, the algorithm is prone to converge to a local optimum too early. Summary of the Invention

[0007] The technical problem solved by this invention is to overcome the shortcomings of the prior art and provide a communication network scheduling method based on an improved particle swarm optimization algorithm.

[0008] The solution of this invention is: a communication network scheduling method based on an improved particle swarm optimization algorithm, comprising: constructing a communication energy consumption model; exploring feasible paths; encoding communication nodes, encoding the m-th particle as S.m = {s1 m , s2 m , …, s n m}, where s n m represents the feasible path selected by the nth user; construct a loss function, and define the loss f m = 1 / C S m , where C S m is the energy consumption of the mth user for communication, and the probability pro m that particle m is selected is pro m = f j / ∑f j , where f

[0009] Furthermore, the steps for constructing the communication network model are as follows:

[0010] Read the global network nodes and initialize the communication network model G = (V, E); where, V = {v k |0 < k ≤ |V|} represents the node set, and E = {e ij |i, j ∈ V, i ≠ j} represents the link set, v k represents the kth node in the network, and e ij represents the link in the network connecting node v i and node v j , v i represents the ith node in the network, and v j represents the jth node in the network;

[0011] Calculate the bandwidth occupied by each link. Let x ij u represent the status of link e ij required for the uth user to communicate. x ij u ∈ {0, 1} indicates whether link e ij is in the active state. If x ij u = 1, then link e ij is in the active state; if x ij u = 0, then link e ij is in the dormant state; the total bandwidth w ij occupied by link e ij is expressed as:

[0012] wij =∑ u∈N w ij u ×x ij u

[0013] Where L is the number of users, u represents the u-th user, 1≤u≤L, w ij u Indicates link e ij Bandwidth usage;

[0014] Calculate the bandwidth used by each node, let x k u Node v represents the communication time of the u-th user. k The state, x k u ∈{0、1} represents node v k Is it in an active state? If x k u =1 then node v k In the active state, i.e. v k ∈V u V u Let x represent the set of nodes for the u-th user; if x k u =0 then node v k In a dormant state; node v k Total bandwidth used w k Represented as: w k =∑ u∈ N w k u ×x k u w k u Represents node v k The bandwidth occupied.

[0015] Furthermore, the steps for constructing the communication energy consumption model are as follows:

[0016] Use c ij Indicates link e ij The amount of energy consumed, x ij ∈{0、1} represents link e ij The state, x ij A value of 1 indicates activation, while x ij A value of 0 indicates hibernation; use c k Represents node v k The amount of energy consumed, x k ∈{0、1} represents node v k The state, x k A value of 1 indicates activation, xk A value of 0 indicates hibernation;

[0017] For a routing strategy P = {p1, p2, ..., pn} that satisfies the communication needs of n users, ... u ... p n}, p u This represents the routing strategy for the u-th user, whose total energy consumption is the sum of the energy consumption of the links and nodes: C S =C link +C node , where C link For link energy consumption, C node Node energy consumption.

[0018] Furthermore, the link energy consumption C link C represents the total energy consumption of the activated links. link =∑ i∈V ∑ j∈V c ij ×x ij Node energy consumption C node C represents the total energy consumption of the activated nodes. node =∑ vk∈V c k ×x k .

[0019] Furthermore, the steps for exploring feasible paths are as follows:

[0020] (51) For the network model G = (V, E), create an empty queue List to store the communication nodes to be visited, enqueue the starting node i, i ∈ List, and mark it as visited;

[0021] (52) Take the head node of the empty queue List and mark it as the current node. Then traverse the current node along the link e. ij All unvisited adjacent nodes j are enqueued;

[0022] (53) Repeat step (52) until the queue is empty and all nodes have been visited, the search ends, and the feasible paths between different nodes are saved in P, then P = {p1, p2, ..., p...} u ... p n}, where p u ={p u 1 p u 2 ... p u h}, 1≤u≤n, p u Let p be the set of all feasible paths for user u. uh This is the h-th feasible path for user u.

[0023] Furthermore, the initialization of the particle swarm employs the random generation of multiple particle codes S. m ={s1 m s2 m 、…、s n m} and velocity V m ={v1 m v2 m ..., v n m} in this way, v n m Represents s n m The corresponding speed.

[0024] Furthermore, the steps for iteratively optimizing the particle swarm are as follows:

[0025] (71) Based on the calculated fitness value F of each particle in the population m ={f1 m f2 m ... f n m}, F m f is the fitness value of the m-th particle. n m The meaning is the fitness value corresponding to the communication path of the nth user. The highest fitness value is the global optimum gBest and the historical optimum pBest of the mth particle. m ;

[0026] (72) Update the particle velocity;

[0027] (73) Update the particle positions;

[0028] (74) Particle mutation: For particle encoding S m ={s1 m s2 m 、…、s n m First, randomly generate an integer k in the interval [1, n], then s k m Replace with the optimal feasible path s k m’ A new particle S is obtained. m ={s1 m s2 m 、…、s k m’ 、…、s n m};

[0029] (75) Repeat steps (71)-(74) until the number of loops is greater than or equal to the preset number of iterations N.

[0030] Furthermore, step (72) updates the particle velocity as follows:

[0031]

[0032] Where ω is the inertia weight, c1 and c2 are inertia factors, and r1 and r2 are random numbers between 0 and 1; It is the velocity of the nth user of the mth particle in the current iteration period; It is the velocity of the nth user of the mth particle in the next iteration cycle. For the individual's optimal position, Current position This represents the optimal position for all particles globally.

[0033] Furthermore, step (73) updates the particle position as follows:

[0034] For particle encoding S m ={s1 m s2 m 、…、s n m}, fitness value F m ={f1 m f2 m ... f n m The expected fitness value of the particle is updated as follows:

[0035] F m (t+1)=F m (t)+V m

[0036] Among them, F m (t+1) is the fitness value of the m-th particle in the next iteration cycle, F m (t) represents the fitness value of the m-th particle in this iteration cycle, for the particle encoding S m For each scheme, select the fitness value and F. m (t) Similar solutions, and update to S m ’ ={s1 m’ s2 m’ 、…、s n m’}

[0037] The advantages of this invention compared to the prior art are:

[0038] (1) The present invention proposes a feasible path search strategy, which avoids the problem of waste of computing resources caused by the exhaustive method on the premise of traversing global network nodes.

[0039] (2) The present invention proposes a solution update strategy based on the global optimal value and the individual optimal value. Under the trend of global optimization, it improves the solution update speed of each individual and realizes the rapid convergence of the solution. At the same time, a particle mutation strategy is proposed. By adding an uncertain term, the solutions of different individuals are updated to avoid falling into the local optimal solution. Brief Description of the Drawings

[0040] Figure 1 It is a flowchart of a communication network scheduling method based on an improved particle swarm optimization algorithm according to the present invention. Detailed Embodiment

[0041] The present invention will be further explained and described below in conjunction with the drawings of the specification and the detailed embodiment.

[0042] As Figure 1 shown, a communication network scheduling method based on an improved particle swarm optimization algorithm mainly includes steps of constructing a communication network model, constructing a communication energy consumption model, exploring a feasible path, encoding communication nodes, constructing a loss function, initializing a population, iteratively optimizing the population, and selecting an optimal solution. The implementation steps of the present invention are as follows:

[0043] 1) The steps of constructing a communication network model are as follows:

[0044] (1) Read global network nodes and initialize the network model G = (V, E). Among them, V = {v k |0 < k ≤ |V|} represents the node set, E = {e ij |i, j ∈ V, i ≠ j} represents the link set, v k represents the k-th node in the network, e ij represents the link connecting node v i and node v j , v i represents the i-th node in the network, v j represents the j-th node in the network.

[0045] (2) Calculate the bandwidth occupied by each link: Let x ij u represent the state of the link e ij required for the communication of the u-th user, x ij u ∈ {0, 1} indicates whether the link e ij is in the active state. If x ij u = 1, then the link eij In an active state; if x ij u =0 then link e ij It is in a dormant state. Therefore, link e ij Total bandwidth used w ij It can be represented as: w ij =∑ u∈N w ij u ×x ij u L represents the number of users; u represents the u-th user, 1≤u≤L, w ij u Indicates link e ij The bandwidth occupied.

[0046] (3) Calculate the bandwidth occupied by each node: Let x k u Node v represents the communication time of the u-th user. k The state, x k u ∈{0、1} represents node v k Is it in an active state? If x k u =1 then node v k In the active state, i.e. v k ∈V u V u Let x represent the set of nodes for the u-th user; if x k u =0 then node v k Currently in a dormant state. Node v k Total bandwidth used w k Represented as: w k =∑ u∈ N w k u ×x k u w k u Represents node v k The bandwidth occupied.

[0047] 2) The steps for constructing the communication energy consumption model are as follows:

[0048] (1) Using c ij Indicates link e ij The amount of energy consumed, x ij ∈{0、1} represents link e ij The state where x ij A value of 1 indicates activation, while x ij A value of 0 indicates sleep mode. Similarly, let c...k Represents node v k The amount of energy consumed, x k ∈{0、1} represents node v k The state where x k A value of 1 indicates activation, while x k A value of 0 indicates hibernation.

[0049] (2) For a routing strategy P = {p1, p2, ..., pn} that satisfies the communication needs of n users, u ... p n}, p u This represents the routing strategy for the u-th user, whose total energy consumption is the sum of the energy consumption of the links and nodes: C S =C link +C node Among them, the energy consumption of the link is C. link C represents the total energy consumption of the activated links. link =∑ i∈V ∑ j∈V c ij ×x ij Node energy consumption C node C represents the total energy consumption of the activated nodes. node =∑ vk∈V c k ×x k .

[0050] 3) The steps for exploring feasible paths are as follows:

[0051] (1) For the network model G = (V, E), create an empty queue List to store the communication nodes to be visited, enqueue the starting node i, i ∈ List, and mark it as visited.

[0052] (2) Take the head node of the empty queue List and mark it as the current node. Then traverse the current node along the link e. ij All unvisited adjacent nodes j are enqueued.

[0053] (3) Repeat step (2) of step 3) until the queue is empty and all nodes have been visited, the search ends, and the feasible paths between different nodes are saved in P, then P = {p1, p2, ..., p...} u ... p n}, where p u ={p u 1 p u 2 ... p u h Let p be the set of all feasible paths for user u (1≤u≤n).u h This is the h-th feasible path for user u.

[0054] 4) Encode the m-th particle as S m ={s1 m s2 m 、…、s n m}, where s n m This represents the feasible path chosen by the nth user.

[0055] 5) The loss f of the m-th particle m Defined as: f m =1 / C S m C S m The energy consumption for communication by the m-th user, and the probability of particle m being selected, pro. m The calculation method is as follows: pro m =f m / ∑f j ;f j This represents the loss of the j-th particle.

[0056] 6) By randomly generating multiple particle codes S m ={s1 m s2 m 、…、s n m} and velocity V m ={v1 m v2 m ..., v n m The particle swarm is initialized in the manner described by}, where v n m Represents s n m The corresponding speed.

[0057] 7) The steps of particle swarm optimization are as follows:

[0058] (1) Calculate the fitness value F of each particle in the population according to step 5). m ={f1 m f2 m ... f n m}, F m f is the fitness value of the m-th particle. n m The meaning is the fitness value corresponding to the communication path of the nth user. The highest fitness value is the global optimum gBest and the historical optimum pBest of the mth particle.m ;

[0059] (2) Particle velocity update: For individual velocity V m ={v1 m v2 m ..., v n m},

[0060]

[0061] Where ω is the inertia weight, c1 and c2 are inertia factors, and r1 and r2 are random numbers in the range [0,1]. It is the velocity of the nth user of the mth particle in the current iteration period; It is the velocity of the nth user of the mth particle in the next iteration cycle. For the individual's optimal position, Current position The optimal position for all particles;

[0062] (3) Particle position update: For particle encoding S m ={s1 m s2 m 、…、s n m}, fitness value F m ={f1 m f2 m ... f n m The expected fitness value of the particle is updated as follows:

[0063] F m (t+1)=F m (t)+V m

[0064] Among them, F m (t+1) is the fitness value of the m-th particle in the next iteration cycle, F m (t) represents the fitness value of the m-th particle in this iteration cycle, for the particle encoding S m For each scheme, select the fitness value and F. m (t) Similar solutions, and update to S m '={s1 m’ s2 m’ 、…、s n m’}

[0065] (4) Particle mutation: For individual S m ={s1 m s2 m 、…、sn m First, randomly generate an integer k in the interval [1, n], then s k m Replace with the optimal feasible path s k m’ A new particle S is obtained. m ={s1 m s2 m 、…、s k m’ 、…、s n m}

[0066] (5) Repeat steps (1)-(4) of step 7) until the number of loops is greater than or equal to the preset number of iterations N.

[0067] 8) For the set of routing schemes generated after the loop ends, select the scheme with the highest fitness based on the fitness value. Specific implementation examples:

[0069] Assuming communication is required between node 1 and node 6, between node 2 and node 5, and between node 1 and node 5, the implementation steps of this invention are as follows:

[0070] 1) The steps for constructing a communication network model are as follows:

[0071] (1) Assume the total number of network nodes is 6, and initialize the communication network model G = (V, E). Where V = {v1, v2, v3, v4, v5, v6} represents the set of communication nodes, and E = {e...} 12 e 13 e 15 e 23 e 24 e 34 e 36 e 45 e 56} represents the set of links between nodes, e ij Indicates the connection node v in the network i and node v j The link.

[0072] (2) Calculate the bandwidth occupied by each link: Let x ij u ∈{0、1} represents link e ij Is it in an active state? If x ij u =1 then link e ij In an active state; if x ij u =0 then link e ijCurrently in sleep mode. Assume the required bandwidth w when each link is invoked. ij u If it is 1, then link e ij The total bandwidth occupied can be expressed as: w ij =∑ u∈N x ij u .

[0073] (3) Calculate the bandwidth occupied by each node: Let x k u ∈{0、1} represents node v k Is it in an active state? If x k u =1 then node v k In the active state, i.e. v k ∈V u If x k u =0 then node v k In a dormant state, let w be the bandwidth required when each node is activated. k u 1, node v k The total bandwidth used is expressed as: w k =∑ u∈N x k u .

[0074] 2) The steps for constructing the communication energy consumption model are as follows:

[0075] (1) Using c ij Indicates link e ij The amount of energy consumed, x ij ∈{0、1} represents link e ij The state where x ij A value of 1 indicates activation, while x ij A value of 0 indicates sleep mode. Similarly, let c... k Represents node v k The amount of energy consumed, x k ∈{0、1} represents node v k The state where x k A value of 1 indicates activation, while x k A value of 0 indicates hibernation.

[0076] (2) For a routing strategy P = {p1, p2, ..., pn} that satisfies the communication needs of n nodes, ... n Its total energy consumption is the sum of the energy consumption of the link and the nodes: C S =C link +C node ;

[0077] Among them, the energy consumption of the link C link C represents the total energy consumption of the activated links. link =∑ i∈V ∑ j∈V c ij ×x ij Node energy consumption C node C represents the total energy consumption of the activated nodes. node =∑ vk∈V c k ×x k .

[0078] 3) The steps for exploring feasible paths are as follows:

[0079] (1) For the communication network model G = (V, E), create an empty queue List to store the communication nodes to be visited, enqueue the starting node i, i ∈ List, and mark it as visited.

[0080] (2) Take the head node of the empty queue List and mark it as the current node. Then traverse the current node along the link e. ij All unvisited adjacent nodes j are enqueued.

[0081] (3) Repeat step 3(2) until the queue is empty and all nodes have been visited, the search ends, and the feasible paths between different nodes are saved in P, then P = {p1, p2, ..., p...} u ... p n}, where P u ={p u 1 p u 2 ... p u h Let} (1≤u≤n) be the set of all feasible paths to node u. Taking the information transfer from node 1 to node 6 as an example, the access process of the feasible paths is shown in Table 1. There are a total of nine feasible paths: 1>2>3>4>5>6, 1>2>3>6, 1>2>4>3>6, 1>2>4>5>6, 1>3>2>4>5>6, 1>3>6, 1>3>4>5>6, 1>5>6, and 1>5>4>3>6, numbered 1-9 respectively.

[0082] Table 1. Access process of feasible paths

[0083]

[0084] According to the above method, the feasible paths for nodes 2 and 5 include 8 schemes: 2>3>4>5, 2>3>6>5, 2>1>3>4>5, 2>1>5, 2>1>3>6>5, 2>4>5, 2>4>3>6>5, and 2>4>3>1>5; the feasible paths for nodes 1 and 5 include 7 schemes: 1>5, 1>2>3>4>5, 1>2>4>5, 1>2>3>6>5, 1>3>4>5, 1>3>6>5, and 1>3>2>4>5.

[0085] 4) Encode the chromosome of the communication node as S m ={s1 m s2 m 、…、s n m}, where s n m This represents the feasible path selected by the nth communication node from the set of feasible paths. Taking communication between nodes 1 and 6 as an example, there are 9 feasible paths. If s1... m If the value is 5, then the 5th path "1>3>2>4>5>6" is selected.

[0086] 5) The loss function is defined as the reciprocal of an individual's energy consumption: f m =1 / C S m The probability of individual m being selected is calculated as follows: pro m =f m / ∑f j Calculate the energy consumption of the path from node 1 to node 6. The energy consumption of the nine feasible paths 1>2>3>4>5>6, 1>2>3>6, 1>2>4>3>6, 1>2>4>5>6, 1>3>2>4>5>6, 1>3>6, 1>3>4>5>6, 1>5>6, and 1>5>4>3>6 are 11, 7, 9, 9, 11, 5, 9, 5, 9 respectively. Therefore, the loss function is 1 / 11, 1 / 7, 1 / 9, 1 / 9, 1 / 11, 1 / 5, 1 / 9, 1 / 5, 1 / 9. According to pro... m =f m / ∑f j The probabilities of each option being selected are 0.0776, 0.12213, 0.0950, 0.0950, 0.0776, 0.1711, 0.0950, 0.1711, and 0.0950.

[0087] The probability of each individual being selected in communication between Node 2 and Node 5 is 0.1264, 0.1264, 0.0983, 0.1770, 0.0983, 0.1770, 0.0983, and 0.0983 for each scheme. The probability of each scheme being selected in communication between Node 1 and Node 5 is 0.3043, 0.1014, 0.1304, 0.1014, 0.1304, 0.1304, and 0.1014 for each scheme.

[0088] 6) Population initialization was performed by randomly generating 10 individuals. The set of communication schemes between node 1 and node 6, between node 2 and node 5, and between node 1 and node 5 were randomly generated, as shown in Table 2.

[0089] Table 2 Randomly generated communication schemes

[0090]

[0091]

[0092] 7) The steps of population iterative optimization are as follows:

[0093] (1) Based on the loss function calculation method in step 5, the loss function of 10 individuals is first calculated. Taking individual 1 as an example, the communication between node 1 and node 6 is scheme two 1>2>3>6, the communication between node 2 and node 6 is scheme three 2>1>3>4>5, and the communication between node 1 and node 5 is scheme six 1>3>6>5. A total of 13 nodes are activated, the link consumption is 10, and the total energy consumption is 23. Similarly, the energy consumption values ​​of other individuals are 29, 21, 23, 21, 23, 27, 19, 27, 19, with probabilities of 0.09893, 0.07846, 0.10835, 0.09893, 0.10835, 0.09893, 0.08427, 0.11976, 0.08427, and 0.11976, respectively.

[0094] Particle 8 is selected as the globally optimal particle, gBest.

[0095] (2) Particle velocity update: For particle codes S1 = {Scheme 2, Scheme 3, Scheme 6}, V1 = {0.1, 0.1, 0.1}, F1 = {1 / 7, 1 / 9, 1 / 9}, F Best ={1 / 5, 1 / 7, 1 / 7},

[0096]

[0097] Where ω is the inertia weight, c1 and c2 are inertia factors, and r1 and r2 are random numbers in the range [0,1].

[0098] (3) Particle position update: For particle code S1 = {Scheme 2, Scheme 3, Scheme 6}, fitness value F i ={1 / 11, 1 / 7, ..., 1 / 3}, the expected fitness value of the particle is updated as follows:

[0099] F1(t+1)={1 / 7, 1 / 9, 1 / 9}+{0.0243, 0.0179, 0.0179}={0.167, 0.130, 0.130}

[0100] For each scheme of particle S1, select the scheme with a fitness value close to F1(t+1) and update it to S1' = {Scheme 8, Scheme 3, Scheme 6'} and pBest. m .

[0101] (4) Particle mutation: For particle code S6 = {Scheme 9, Scheme 4, Scheme 4}, first generate an integer 1 in the interval [1, 3], and then replace "Scheme 9" with the optimal feasible path "Scheme 8" to obtain the new particle code S1 = {Scheme 8, Scheme 4, Scheme 4}.

[0102] (5) Repeat steps 7(1)-(4) until the number of loops is greater than or equal to the preset number of iterations N.

[0103] 8) For the set of scheduling schemes generated after the loop ends, the optimal scheme is selected as S7 = {Scheme 8, Scheme 1, Scheme 1}.

[0104] Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make possible changes and modifications to the technical solutions of the present invention by utilizing the methods and techniques disclosed above without departing from the spirit and scope of the present invention. Therefore, any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solutions of the present invention shall fall within the protection scope of the technical solutions of the present invention.

Claims

1. A communication network scheduling method based on an improved particle swarm optimization algorithm, characterized in that, include: Construct a communication network model; Construct a communication energy consumption model; Explore feasible paths; The encoding communication node encodes the m-th particle as S. m ={s1 m s2 m 、…、s n m }, s n m This represents the feasible path chosen by the nth user. Construct a loss function and define the loss for the m-th particle. C S m The energy consumption for communication by the m-th user, and the probability of particle m being selected, pro. m =f m / ∑f j f j The loss of the j-th particle; Initialize the particle swarm; Iterative optimization of particle swarm optimization; For the set of routing schemes generated after the loop ends, select the scheme with the highest fitness based on the fitness value.

2. The communication network scheduling method based on the improved particle swarm optimization algorithm according to claim 1, characterized in that, The steps for constructing the communication network model are as follows: Read the global network nodes and initialize the communication network model G = (V, E); Among them, V = {v k | 0 < k ≤ |V|} represents the node set, E = {e ij | i, j ∈ V, i ≠ j} represents the link set, v k represents the k-th node in the network, e ij represents the link connecting node v i and node v j in the network, v i represents the i-th node in the network, v j represents the j-th node in the network; Calculate the bandwidth occupied by each link, let x be... ij u e represents the link e required for communication by the u-th user. ij The state, x ij u ∈{0、1} represents link e ij Is it in an active state? If x ij u =1 then link e ij In an active state; if x ij u =0 then link e ij In hibernation state; link e ij Total bandwidth used w ij Represented as: In ij =∑ u∈N In ij u ×x ij u Where L is the number of users, u represents the u-th user, 1≤u≤L, w ij u Indicates link e ij Bandwidth usage; Calculate the bandwidth used by each node, let x k u Node v represents the communication time of the u-th user. k The state, x k u ∈{0、1} represents node v k Is it in an active state? If x k u =1 then node v k In the active state, i.e. v k ∈V u， V u Let x represent the set of nodes for the u-th user; if x k u =0 then node v k In a dormant state; node v k Total bandwidth used w k Represented as: w k =∑ u∈N w k u ×x k u w k u Represents node v k The bandwidth occupied.

3. A communication network scheduling method based on an improved particle swarm optimization algorithm according to claim 2, characterized in that, The steps for constructing the communication energy consumption model are as follows: Use c ij Indicates link e ij The amount of energy consumed, x ij ∈{0、1} represents link e ij The state, x ij A value of 1 indicates activation, while x ij A value of 0 indicates hibernation; Use c k Represents node v k The amount of energy consumed, x k ∈{0、1} represents node v k The state, x k A value of 1 indicates activation, x k A value of 0 indicates hibernation; For a routing strategy P = {p1, p2, ..., pn} that satisfies the communication needs of n users, ... u ... p n }, p u This represents the routing strategy for the u-th user, whose total energy consumption is the sum of the energy consumption of the links and nodes: C S =C link +C node , where C link For link energy consumption, C node Node energy consumption.

4. A communication network scheduling method based on an improved particle swarm optimization algorithm according to claim 3, characterized in that, The link energy consumption C link C represents the total energy consumption of the activated links. link =∑ i∈V ∑ j∈V c ij ×x ij Node energy consumption C node C represents the total energy consumption of the activated nodes. node =∑ vk∈V c k ×x k .

5. A communication network scheduling method based on an improved particle swarm optimization algorithm according to claim 2, characterized in that, The steps for exploring feasible paths are as follows: (51) For the network model G = (V, E), create an empty queue List to store the communication nodes to be visited, enqueue the starting node i, i ∈ List, and mark it as visited; (52) Take the head node of the empty queue List and mark it as the current node. Then traverse the current node along the link e. ij All unvisited adjacent nodes j are enqueued; (53) Repeat step (52) until the queue is empty and all nodes have been visited, the search ends, and the feasible paths between different nodes are saved in P, then P = {p1, p2, ..., p...} u ... p n }, where p u ={p u 1 p u 2 ... p u h }, 1≤u≤n, p u Let p be the set of all feasible paths for user u. u h This is the h-th feasible path for user u.

6. A communication network scheduling method based on an improved particle swarm optimization algorithm as described in claim 1, characterized in that, The initialization of the particle swarm uses multiple randomly generated particle codes S m ={s1 m s2 m 、…、s n m } and velocity V m ={v1 m v2 m ..., v n m } in this way, v n m Represents s n m The corresponding speed.

7. A communication network scheduling method based on an improved particle swarm optimization algorithm as described in claim 1, characterized in that, The steps for iteratively optimizing the particle swarm are as follows: (71) Based on the calculated fitness value F of each particle in the population m ={f1 m f2 m ... f n m }, F m f is the fitness value of the m-th particle. n m The meaning is the fitness value corresponding to the communication path of the nth user. The highest fitness value is the global optimum gBest and the historical optimum pBest of the mth particle. m ; (72) Update the particle velocity; (73) Update the particle positions; (74) Particle mutation: For particle encoding S m ={s1 m s2 m 、…、s n m First, randomly generate an integer k in the interval [1, n], then s k m Replace with the optimal feasible path s k m’ A new particle S is obtained. m ={s1 m s2 m 、…、s k m’ 、…、s n m }; (75) Repeat steps (71)-(74) until the number of loops is greater than or equal to the preset number of iterations N.

8. A communication network scheduling method based on an improved particle swarm optimization algorithm as described in claim 7, characterized in that, The step (72) described above updates the particle velocity as follows: Where ω is the inertia weight, c1 and c2 are inertia factors, and r1 and r2 are random numbers between 0 and 1; It is the velocity of the nth user of the mth particle in the current iteration period; It is the velocity of the nth user of the mth particle in the next iteration cycle. For the individual's optimal position, Current position This represents the optimal position for all particles globally.

9. A communication network scheduling method based on an improved particle swarm optimization algorithm as described in claim 7, characterized in that, Step (73) updates the particle position as follows: For particle encoding S m ={s1 m s2 m 、…、s n m }, fitness value F m ={f1 m f2 m ... f n m The expected fitness value of the particle is updated as follows: F m (t+1)=F m (t)+V m Among them, F m (t+1) is the fitness value of the m-th particle in the next iteration cycle, F m (t) represents the fitness value of the m-th particle in this iteration cycle, for the particle encoding S m For each scheme, select the fitness value and F. m (t) Similar solutions, and update to S m ’ ={s1 m’ s2 m’ 、…、s n m’ }

Images