CN110381470B - A Joint Optimization Method for Access Control Oriented to Quality of Service Guarantee in Railway Internet of Things - Google Patents

Abstract

The invention discloses an access control joint optimization method facing to service quality guarantee in a railway internet of things in the technical field of the internet of things, which comprises the following steps: introducing white space to improve network throughput; introducing a utility function to ensure fairness and stability of users of an access control scheme and improve QoS (quality of service) in access control; the method comprises the steps of providing an Internet of things access control joint optimization model with both rate control and spectrum allocation; the given access control procedure mainly comprises: performing greedy coloring-based spectrum initial allocation; then, according to a distributed rate control algorithm, the stability constraint of the combined optimization model is satisfied; and finally, performing new allocation on the frequency spectrum by using a dynamic frequency spectrum allocation algorithm. After joint optimization is implemented, the access control of the railway Internet of things can be guaranteed with QoS.

Description

A Joint Optimization of Access Control Oriented to Quality of Service Guarantee in Railway Internet of Things method

technical field

The invention relates to a technical method in the technical field of the Internet of Things, in particular to a joint optimization method for access control that combines initial spectrum allocation and dynamic spectrum allocation technology after introducing a blank spectrum and a utility function to ensure the service quality of the railway Internet of Things .

Background technique

The Internet of Things is the Internet between objects, and it is an extension and extended application of the Internet and communication networks. It makes full use of existing network technologies and combines sensing and radio frequency identification technologies to enable information interaction and communication between objects and people, and between objects and objects. Seamless link. It is an inevitable development trend to use Internet of Things technology to cross-integrate technology with other fields. After the introduction of IoT technology, the railway industry has undergone major changes. First, the deployment of IoT devices can make rail transport more secure. For example, by deploying a sufficient number of sensors around the railway track, data such as track status information and train axle temperature information can be collected well. At the same time, important personnel such as drivers carry wireless wearable devices to collect physical health data, so as to respond to emergencies. situation to take measures. Furthermore, train operators can use IoT devices to geo-locate trains in real time and analyze their running status, capture and analyze the number of passengers in the train and the number of passengers waiting at the station, so as to ensure driving safety and operate trains more efficiently. Finally, IoT devices can provide passengers with ticketing information, location, consumption, etc., so that train operators can provide passengers with more personalized and high-quality services. However, with the huge number of IoT devices deployed in the railway system, the railway system is more dependent on wireless connection, which means it is more vulnerable to external interference, intrusion and cyber attack. With dense rail networks, even minor disruptions can have serious consequences as passenger numbers increase and travel speeds increase. Therefore, the resolution of security issues is increasingly important. The network connection in the Internet of Things is controlled by the access layer, which can realize the communication between the sensor and the Internet. This kind of Unicom requires the assistance of network infrastructure equipment, such as the current base station, WiFi, satellite network, etc. Therefore, in order for the railway Internet of Things to have sufficient security, it is necessary to ensure the security of access. In the network access technology of the railway Internet of Things, the main concern is the "last mile" local access of smart devices in the Internet of Things or sensor networks. Among them, related wireless access technologies include Zigbee, Bluetooth, LoRa, SigFox, NB-IoT and LTE-eMTC. In information exchange and communication during access, the pair of indicators of delay and throughput occupy an important position. However, the two are contradictory to a certain extent, and the transmission quality is usually ignored for the improvement of the throughput, and the delay is increased for the improvement of the transmission quality.

After searching the existing technical literature, it was found that Fang Zhengyue et al.’s patent "A Power Consumption Optimization Method and Terminal in a Narrowband Internet of Things System" proposes to adopt different low-power configurations according to different system states to achieve Energy efficiency optimization purpose. The patent "A Method for Optimizing the Selection of Access Channels of IoT Nodes" by Ma Li et al. uses the Markov model for network access on the basis of classifying services, which not only meets the business needs of each terminal, but also improves the network resource utilization. The article "White Space Networking with Wi-Fi likeConnectivity" by Paramvir Bahl et al. utilizes white space spectrum to achieve an increase in system throughput in wireless local area networks. In the article "Delay-Based Network Utility Maximization" by M.J.Neely et al., a network utility optimization model is proposed to optimize the delay. However, these works only consider one indicator, such as energy consumption, delay, and throughput, and perform a single optimization on it.

Contents of the invention

The present invention provides a quality-of-service-oriented access control joint optimization method in the railway Internet of Things, which is applied to the access network of the Railway Internet of Things with dynamic spectrum access capabilities, and can ensure stable network capacity by introducing a utility function and selecting On the basis of reasonable and full use of blank spectrum, the two factors of rate control and spectrum allocation of each access device in the railway Internet of Things are optimized.

In order to achieve the above-mentioned joint optimization method, the technical solution of the present invention includes:

The access control system on the railway Internet of Things of the present invention is mainly composed of three parts: a controller, a number of users, and a number of wireless access points, wherein the controller monitors and controls all wireless access points, and periodically Calculate the interference between all wireless access points and users; the access point can have multiple wireless interfaces, but the user has only one interface and can only be connected with one access point; abstractly, represent the wireless of all access points with I Interface set, C represents the user set, R represents the effective transmission link set between the user and the access point, N represents the interference graph between the wireless interfaces of all access points, N(i) represents the set of other wireless interfaces that interfere with the wireless interface i, Note i∈N(i).

在源节点a处会维护流向目的节点b的报文缓冲队列/>

该缓冲队列的报文准入速率为/>

此外，假设所有节点间的传输都是独立同分布的；各速率间存在以下相关约束关系：The present invention adopts a discrete time slot system that divides time into small slots on average, and adopts a first-in-first-out FIFO message buffer management mode; a∈I∪C, that is, both wireless access points and users can be used as source nodes; b∈C , that is, all users can be used as nodes; in the network layer, a rate controller is placed to effectively control the rate of packets entering the network layer from the transport layer; starting from the source node a, the packet sent to the destination node b The text sending rate is denoted as

At the source node a, a message buffer queue to the destination node b will be maintained />

The packet admission rate of this buffer queue is />

In addition, it is assumed that the transmissions between all nodes are independent and identically distributed; the following correlation constraints exist between the rates:

Among them, ω _a,max (t) represents the maximum aggregate admission rate that the network at node a can achieve, and μ _a,max (t) represents the maximum aggregate transmission rate that the network at node a can achieve.

和/>

的长期时间平均值表达：Additionally, given the following

and />

The long-term time mean expression of :

Dynamically update the queue from source node a to destination node b, expressed as:

Among them, [ ] ⁺ means that when . The access point does not need to cache the packets of users connected to the wired network or other access points; the third formula indicates that the destination nodes of all transmissions do not need to cache the relevant packets.

The present invention introduces a virtual subnet queue to evaluate the wireless interface load of the access point; the virtual subnet queue is a variable that can reflect the load of the subnet, and the larger the value, the heavier the load of the subnet , so that more spectrum resources are needed to increase the channel capacity, and finally reduce the buffer queue length of the network and alleviate the congestion of the subnet;

Express the dynamic update of the virtual subnet queue as:

V _i (t+1)＝[V _i (t)-μ _i (t)+ω _i (t)] ⁺

μ _i (t)≤Capa _i (t)

Among them, Capa _i (t) represents the channel capacity of wireless interface i, μ _i (t) represents the aggregate transmission rate of the wireless link between wireless interface i and the connected users, ω _i (t) represents the channel capacity of the corresponding subnet The aggregate admission rate of all nodes, the last formula means that the maximum aggregate transmission rate of all transmissions in the corresponding subnet is equal to the channel capacity of the wireless interface i.

其中ω_i是弹性系数，对数形式会减小x_i的大小，从而在一定程度上避免大波动的异常出现，具体的，本发明中使用的效用函数形式为：The present invention uses the utility function of logarithmic synthesis, and its general form is

Wherein ω _i is the elastic coefficient, and the logarithmic form will reduce the size of _xi , thereby avoiding the abnormal occurrence of large fluctuations to a certain extent, specifically, the form of the utility function used in the present invention is:

是弹性系数，/>

表示时间平均的虚拟子网队列，/>

表示基于子网的时间平均聚合传输速率，τ为平衡/>

和/>

两者对频谱分配的影响因子。in,

is the coefficient of elasticity, />

Represents a time-averaged virtual subnet queue, />

Indicates the time-average aggregate transmission rate based on subnetwork, τ is the balance />

and />

The impact factors of the two on spectrum allocation.

集合组成，其能在保证网络稳定的基础上使得系统吞吐量最优化；避免使用传统的网络容量区域，是因为在此容量区域内求解出的最终结果会导致网络稳定性被破坏，从而实际使用中的成功率极其低下。The present invention selects the network capacity area Π, which consists of all the networks that can be supported by the network in a proportional sharing manner

It is composed of sets, which can optimize the system throughput on the basis of ensuring network stability; avoid using the traditional network capacity area, because the final result obtained in this capacity area will lead to the destruction of network stability, so that the actual use of The success rate is extremely low.

Step 1: Establish a joint optimization model for system access control;

Assuming that the network utility function used has convex, non-decreasing and continuous correlation characteristics, the joint optimization model can be described as:

Among them, the first constraint condition can ensure proportional fairness, and can make the long-term time average admission rate of all nodes in the network capacity area Π, the second constraint condition indicates that the long-term time average admission rate cannot be greater than the long-term time average transmission rate At the same time, it needs to be clear that the spectrum allocation of the wireless interface of the access point constrains the size of Π; if the traditional network capacity area is used, the final result will lead to the destruction of network stability; therefore, using the network capacity area Π can ensure network stability The system throughput is optimized on the basis of

Step 2: Perform initial spectrum allocation for the system;

The present invention uses a spectrum allocation algorithm based on greedy coloring to obtain the initial solution of the spectrum allocation; firstly, the input data is processed, the data is formed into an undirected graph, and the degree of each node is used to arrange and number them in ascending order; at the same time, the colors to be used are also The numbers are arranged in ascending order; then, each node is colored according to the relevant rules; the channels are classified according to the color of each node after coloring; finally, the blank spectrum is allocated by using the number of colors consumed by coloring, and we get Spectrum allocation solution; the relevant rules are: nodes and colors are used according to the number from small to large, and the colors of adjacent nodes must be different.

Step 3: Perform joint optimization of access control;

Using the distributed rate control algorithm based on Lyapunov optimization, the stability constraints of the joint optimization model are satisfied; the specific implementation is composed of the rate controller of the node network layer and the link scheduler of the access point; the former can control the network layer. The rate of the message flow injected by the upper layer protocol, and the buffer queue length of each transmission flow in the node determines the rate; the latter is based on the buffer length difference between the access point and the user connected to it to perform wireless link scheduling transmission, so that Minimize the overall length of the buffer queue in the subnet.

(1) Rate controller: In each time slot, node a controls the admission rate of all messages transmitted to b, and its optimization problem is described as follows:

Among them, in order to balance the emphasis on optimization of throughput and delay, the γ balance parameter is introduced; the utility function is set to be convex, non-decreasing and continuous, so the optimal untie;

(2), link scheduler: in each time slot, the access point will observe the buffer queue length of all transmissions in the subnet, determined by

It can be seen that the buffer length difference between the access point and the users connected to it can be directly expressed by the transmission buffer queue length; therefore, the access point will preferentially schedule the transmission link with the largest buffer length in the subnet for transmission, and its length is

Step 4: Dynamically access spectrum allocation to users;

和/>

对频谱进行新的分配，最终使得联合优化模型中的网络效用最优化得以实现；将效用函数代入联合优化模型中，得Using the dynamic spectrum allocation algorithm based on Frank-Wolfe, according to the initial spectrum allocation solution S _i obtained in step 1 and the distributed rate control algorithm obtained in step 3

and />

The new allocation of the frequency spectrum finally enables the network utility optimization in the joint optimization model to be realized; substituting the utility function into the joint optimization model, we get

记计算得到的最优解为q^(m),并且记d^(m)＝q^(m)-p^(m)；②判断

是否为真？若为真，输出p^(m)，循环结束；否则，将d^(m)作为可行下降方向，进行有效一维搜索，以此确定一维搜索步长；同时为保证结果在可行域内，规定步长α∈[0,1]；③接着处理问题：/>

通过计算得到最优解，记为α_m，令p^(m+1)＝p^(m)+α_m*d^(m)，且使m＝m+1，重复上述过程。Because this formula has been proved to be an NP-hardness problem, an approximate solution is required to calculate the problem; first use the algorithm in step 2 to complete the channel allocation, and then use the Frank-Wolfe idea to solve it; it can be simply described as: select The initial feasible point, that is, the initial channel allocation result p, specifies the allowable error range ε, and sets m=0; ① Carry out linear programming, that is, deal with the problem:

Record the calculated optimal solution as q ^(m) , and record d ^(m) = q ^(m) -p ^(m) ; ② Judgment

Is it true? If it is true, output p ^(m) and the loop ends; otherwise, use d ^(m) as the feasible descending direction to conduct an effective one-dimensional search to determine the one-dimensional search step size; at the same time, to ensure that the result is within the feasible region, the specified step Long α∈[0,1]; ③ Then deal with the problem: />

The optimal solution is obtained by calculation, denoted as α _m , let p ^(m+1) =p ^(m) +α _m *d ^(m) , and let m=m+1, repeat the above process.

The beneficial effects of the present invention are: the present invention innovatively jointly optimizes both the delay and the throughput; The introduction of utility theory reduces the complexity of the system and realizes partial optimization of the access control of the railway Internet of Things; the specific utility function can make the resource allocation scheme ensure the fairness and stability of users, and can improve the quality of service (QoS) of the business .

Description of drawings

Fig. 1 is a schematic composition diagram of the railway internet of things access control system of the present invention;

Fig. 2 is the related utility function image of the present invention;

Fig. 3 is a flow chart of the joint optimization access control method of the present invention;

Fig. 4 is a diagram of inter-user interference in this embodiment.

Detailed ways

The embodiments of the present invention are described in detail below. This embodiment is implemented on the premise of the technical solution of the present invention, and detailed implementation methods and specific operating procedures are provided, but the protection scope of the present invention is not limited to the following implementation example.

Technical scheme of the present invention comprises:

The access control system on the railway Internet of Things of the present invention is mainly composed of three parts: a controller, a number of users, and a number of wireless access points, wherein the controller monitors and controls all wireless access points, and periodically Calculate the interference between all wireless access points and users; the access point can have multiple wireless interfaces, but the user has only one interface and can only be connected with one access point; abstractly, represent the wireless of all access points with I Interface set, C represents the user set, R represents the effective transmission link set between the user and the access point, N represents the interference graph between the wireless interfaces of all access points, N(i) represents the set of other wireless interfaces that interfere with the wireless interface i, Note i∈N(i).

在源节点a处会维护流向目的节点b的报文缓冲队列/>

该缓冲队列的报文准入速率为/>

此外，假设所有节点间的传输都是独立同分布的；各速率间存在以下相关约束关系：The present invention adopts a discrete time slot system that divides time into small slots on average, and adopts a first-in-first-out FIFO message buffer management mode; a∈I∪C, that is, both wireless access points and users can be used as source nodes; b∈C , that is, all users can be used as nodes; in the network layer, a rate controller is placed to effectively control the rate of packets entering the network layer from the transport layer; starting from the source node a, the packet sent to the destination node b The text sending rate is denoted as

At the source node a, a message buffer queue to the destination node b will be maintained />

The packet admission rate of this buffer queue is />

In addition, it is assumed that the transmissions between all nodes are independent and identically distributed; the following correlation constraints exist between the rates:

Among them, ω _a,max (t) represents the maximum aggregate admission rate that the network at node a can achieve, and μ _a,max (t) represents the maximum aggregate transmission rate that the network at node a can achieve.

和/>

的长期时间平均值表达：Additionally, given the following

and />

The long-term time mean expression of :

Dynamically update the queue from source node a to destination node b, expressed as:

Among them, [ ] ⁺ means that when . The access point does not need to cache the packets of users connected to the wired network or other access points; the third formula indicates that the destination nodes of all transmissions do not need to cache the relevant packets.

The present invention introduces a virtual subnet queue, thereby evaluating the wireless interface load of the access point; the virtual subnet queue is a variable that can reflect the load of the subnet, and the larger the value, the heavier the load of the subnet, thus More spectrum resources are needed to increase channel capacity, ultimately reducing the buffer queue length of the network and alleviating subnet congestion;

Express the dynamic update of the virtual subnet queue as:

V _i (t+1)＝[V _i (t)-μ _i (t)+ω _i (t)] ⁺

μ _i (t)≤Capa _i (t)

Among them, Capa _i (t) represents the channel capacity of wireless interface i, μ _i (t) represents the aggregate transmission rate of the wireless link between wireless interface i and the connected users, ω _i (t) represents the channel capacity of the corresponding subnet The aggregate admission rate of all nodes, the last formula means that the maximum aggregate transmission rate of all transmissions in the corresponding subnet is equal to the channel capacity of the wireless interface i.

其中ωi是弹性系数，对数形式会减小x_i的大小，从而在一定程度上避免大波动的异常出现，具体的，本发明中使用的效用函数形式为：The present invention uses the utility function of logarithmic synthesis, and its general form is

Wherein ωi is the elastic coefficient, and the logarithmic form will reduce the size of _xi , thereby avoiding the abnormal occurrence of large fluctuations to a certain extent, specifically, the utility function form used in the present invention is:

是弹性系数，/>

表示时间平均的虚拟子网队列，/>

表示基于子网的时间平均聚合传输速率，τ为平衡/>

和/>

两者对频谱分配的影响因子。in,

is the coefficient of elasticity, />

Represents a time-averaged virtual subnet queue, />

Indicates the time-average aggregate transmission rate based on subnetwork, τ is the balance />

and />

The impact factors of the two on spectrum allocation.

集合组成，它能在保证网络稳定的基础上使得系统吞吐量最优化；避免使用传统的网络容量区域，是因为在此容量区域内求解出的最终结果会导致网络稳定性被破坏，从而实际使用中的成功率极其低下。The present invention selects the network capacity area Π, which consists of all the networks that can be supported by the network in a proportional sharing manner

It is composed of a set, which can optimize the system throughput on the basis of ensuring network stability; avoid using the traditional network capacity area, because the final result obtained in this capacity area will lead to the destruction of network stability, so that the actual use of The success rate is extremely low.

Step 1: Establish a joint optimization model for system access control;

The network utility function used is assumed to have convex, non-decreasing and continuous characteristics, and the joint optimization model can be described as:

Among them, the first constraint condition can ensure proportional fairness, and can make the long-term time average admission rate of all nodes in the network capacity area Π, the second constraint condition indicates that the long-term time average admission rate cannot be greater than the long-term time average transmission rate At the same time, it needs to be clear that the spectrum allocation of the wireless interface of the access point constrains the size of Π; if the traditional network capacity area is used, the final result will lead to the destruction of network stability; therefore, using the network capacity area Π can ensure network stability The system throughput is optimized on the basis of

Step 2: Perform initial spectrum allocation for the system;

The present invention uses a spectrum allocation algorithm based on greedy coloring to obtain the initial solution of the spectrum allocation; firstly, the input data is processed, the data is formed into an undirected graph, and the degree of each node is used to arrange and number them in ascending order; at the same time, the colors to be used are also Numbering, arranged according to the number from small to large; then, color each node according to the rules; classify the channels according to the color of each node after coloring; finally use the number of colors consumed by coloring to allocate the blank spectrum to obtain the spectrum Allocation solution; the rule is: nodes and colors are used according to the number from small to large, and the colors of adjacent nodes must be different.

Step 3: Perform joint optimization of access control;

Using the distributed rate control algorithm based on Lyapunov optimization, the stability constraints of the joint optimization model are satisfied; the specific implementation is composed of the rate controller of the node network layer and the link scheduler of the access point; the former can control the network layer. The rate of the message flow injected by the upper layer protocol, and the buffer queue length of each transmission flow in the node determines the rate; the latter is based on the buffer length difference between the access point and the user connected to it to perform wireless link scheduling transmission, so that Minimize the overall length of the buffer queue in the subnet.

(1) Rate controller: In each time slot, node a controls the admission rate of all messages transmitted to b, and its optimization problem is described as follows:

Among them, in order to balance the emphasis on optimization of throughput and delay, the γ balance parameter is introduced; the utility function is set to be convex, non-decreasing and continuous, so the optimal untie;

(2), link scheduler: in each time slot, the access point will observe the buffer queue length of all transmissions in the subnet, determined by

It can be seen that the buffer length difference between the access point and the users connected to it can be directly expressed by the transmission buffer queue length; therefore, the access point will preferentially schedule the transmission link with the largest buffer length in the subnet for transmission, and its length is

Step 4: Dynamically access spectrum allocation to users;

和/>

对频谱进行新的分配，最终使得联合优化模型中的网络效用最优化得以实现；将效用函数代入联合优化模型中，得Using the dynamic spectrum allocation algorithm based on Frank-Wolfe, according to the initial spectrum allocation solution S _i obtained in step 1 and the distributed rate control algorithm obtained in step 3

and />

The new allocation of the frequency spectrum finally enables the network utility optimization in the joint optimization model to be realized; substituting the utility function into the joint optimization model, we get

记计算得到的最优解为q^(m),并且记d^(m)＝q^(m)-p^(m)；②判断

是否为真？若为真，输出p^(m)，循环结束；否则，将d^(m)作为可行下降方向，进行有效一维搜索，以此确定一维搜索步长；同时为保证结果在可行域内，规定步长α∈[0,1]；③接着处理问题：/>

通过计算得到最优解，记为α_m，令p^(m+1)＝p^(m)+α_m*d^(m)，且使m＝m+1，重复上述过程。Because this formula has been proved to be an NP-hardness problem, an approximate solution is required to calculate the problem; first use the algorithm in step 2 to complete the channel allocation, and then use the Frank-Wolfe idea to solve it; it can be simply described as: select The initial feasible point, that is, the initial channel allocation result p, specifies the allowable error range ε, and sets m=0; ① Carry out linear programming, that is, deal with the problem:

Record the calculated optimal solution as q ^(m) , and record d ^(m) = q ^(m) -p ^(m) ; ② Judgment

Is it true? If it is true, output p ^(m) and the loop ends; otherwise, use d ^(m) as the feasible descending direction to conduct an effective one-dimensional search to determine the one-dimensional search step size; at the same time, to ensure that the result is within the feasible region, the specified step Long α∈[0,1]; ③ Then deal with the problem: />

The optimal solution is obtained by calculation, denoted as α _m , let p ^(m+1) =p ^(m) +α _m *d ^(m) , and let m=m+1, repeat the above process.

The composition of the railway Internet of Things access control system given in this embodiment is assumed to be: 1 controller, 3 wireless access points in total, and 16 users; the interference relationship between the 16 users is shown in Figure 4; note one The spectrum allocation solution is S={<lower _i (t), upper _i (t)>} _i∈I , where the upper and lower bounds of the channel used by wireless interface i at time t are upper _i (t) and lower _i (t) ; After introducing the blank spectrum, record WH(f _lower ,f _upper ,WS) as the entire blank spectrum, WS as the total width of the spectrum, and the upper and lower bounds of the blank spectrum are f _upper and f _lower ; in the simulation, 1×n-dimensional Vector plus Gaussian white noise, as blank spectrum WH.

This embodiment is realized through the following steps:

Step 1: Make initial spectrum allocation:

This embodiment uses the spectrum allocation algorithm based on greedy coloring to obtain the initial solution of the spectrum allocation; firstly, the input data is processed, the data is formed into an undirected graph, and the number of nodes is n; a 1×n-dimensional zero matrix is used as each node The initial value of the color is CM; the total number of colors used is CON, and the initial value is 0; then, it is arranged in ascending order based on the degree, and the result is recorded as D; traverse all nodes, i,j=1,2,...,n; Find the color value of the connected adjacent points in the data adjacency matrix W(i,j), choose a minimum and different color value, and assign it to CM(j); if the color is not enough, then CON=CON+1; according to CM for all nodes Color accordingly. Finally, the blank spectrum WH is evenly put into the channels CON: p _i =WS/CON, i=1,2,...,CON, and the spectrum allocation solution at channel i in WH is obtained as

After network simulation, the original 1 to 16 users are reordered as: D=[4 12 13 14 1 3 5 11 16 26 7 8 9 15 10]; then four user categories can be obtained, successively using blue, green, yellow and black; nodes (users) 1, 3, 4, 6, 12, 13, 14, and 16 are blue; nodes (users) 2, 5, 7, and 11 are green; nodes (users) 8, 9 and 15 are yellow; node (user) 10 is black; that is, the channel can be initially allocated into 4 parts; accordingly, the initial allocation solution of the blank spectrum is: S=[4 4 44]; wherein, S _i is the spectrum at channel i Assignment solution, i represents the i-th column of the above S.

Step 2: Perform joint optimization of access control:

The distributed rate control algorithm is used to satisfy the stability constraints of the joint optimization model; the following rate can be obtained after using the network simulation:

The aggregate transmission rate of the wireless link between the wireless interface i and the connected users is shown in Table 1:

Table 1 Aggregated transmission rate of wireless link between wireless interface i and connected users:

The aggregate admission rate of all nodes in the subnet is shown in Table 2:

Table 2 Aggregate admission rate of all nodes in the subnet:

Step 3: Dynamically allocate spectrum;

The allocation algorithm used is based on the Frank-Wolfe idea;

和/>

对频谱进行新的分配。将效用函数带入

中，可得/>

因为该式已被证明是NP-hardness问题，故为计算该问题，需进行近似求解；According to the second step

and />

Make new allocations to the spectrum. Bring the utility function into

in, available />

Because this formula has been proved to be an NP-hardness problem, an approximate solution is required to calculate this problem;

记计算得到的最优解为q^(m),并且记d^(m)＝q^(m)-p^(m)；随后，判断条件/>

是否为真？若为真，输出p^(m)，循环结束；否则，将d^(m)作为可行下降方向，进行有效一维搜索，以此确定一维搜索步长。为保证结果在可行域内，规定步长α∈[0,1]；接着处理问题：/>

通过计算得到最优解，记为α_m，令p^(m+1)＝p^(m)+α_m*d^(m)，且m＝m+1,重复上述过程；First use the algorithm in step 1 to complete the channel allocation, and then use Frank-Wolfe's idea to solve it. First take the output color number CON in step 1, set m=1, fk=1, specify the allowable error range ε= ^10-5 ; loop the following process to carry out linear programming to solve the problem:

Record the calculated optimal solution as q ^(m) , and record d ^(m) = q ^(m) -p ^(m) ; subsequently, judge the condition

Is it true? If it is true, output p ^(m) and the loop ends; otherwise, use d ^(m) as the feasible descending direction to perform an effective one-dimensional search, so as to determine the one-dimensional search step size. In order to ensure that the result is within the feasible region, the step size α∈[0,1] is stipulated; then deal with the problem: />

Obtain the optimal solution through calculation, denoted as α _m , let p ^(m+1) = p ^(m) + α _m *d ^(m) , and m=m+1, repeat the above process;

设定因子τ＝1，选用p_i＝WS/CON,i＝1,2,...,CON作为初始点。使用网络仿真后可得如下动态分配解：S＝(10^-3×)[0.1545 0.1545 0.1545 0.1545]。Among them, f:R ⁿ →R ¹ is a differentiable function, p ^(m) represents the mth iteration point on p, ε is arbitrarily small (in theory), and can be set according to the actual situation, α _m is the optimal step size of p ^(m) ; in the present embodiment, the differentiable function is

The factor τ=1 is set, and p _i =WS/CON, i=1, 2, . . . , CON are selected as initial points. After using the network simulation, the following dynamic allocation solution can be obtained: S = (10 ^-3 ×) [0.1545 0.1545 0.1545 0.1545].

The optimization model adopted in this embodiment comprehensively considers the relationship between delay and throughput, dynamically allocates frequencies according to the rate, and realizes joint optimization of throughput and delay.

The present invention innovatively jointly optimizes delay and throughput. On the basis of a reasonable trade-off between throughput and delay, this paper introduces the utility theory in the wireless resource allocation of the railway Internet of Things, which reduces the complexity of the system and realizes partial optimization of the access control of the railway Internet of Things. The specific utility function can ensure the fairness and stability of users in the resource allocation scheme, and can improve the service quality (QoS) of the business.

Claims (1)

1. A joint optimization method for access control for quality of service guarantee in the railway Internet of Things, characterized in that: by introducing a utility function and selecting a stable network capacity area to ensure network stability, on the basis of fully and reasonably utilizing blank spectrum, Unified analysis of rate control and spectrum allocation, dynamic spectrum allocation with the help of initial spectrum allocation, virtual subnetwork queues and transmission rates, and a joint optimization model is given to achieve delay and throughput in railway Internet of Things access control The joint optimization of quantity, so as to meet the demand of its service quality assurance; Wherein, the objective function of the joint optimization model is:

where I represents the set of wireless interfaces of all access points, C represents the set of users, WS represents the total width of white space spectrum, CON represents the number of channel classes after the initial allocation,

Represents a time-averaged virtual subnet queue, />

Indicates the time-averaged aggregate transmission rate based on subnetwork, and τ indicates the balance />

and />

The influence factors of the two on spectrum allocation, S _i represents the initial spectrum allocation solution at channel i, μ _i (t) represents the aggregate transmission rate of the wireless link between wireless interface i and the user connected to it; a represents the source node, b represents the destination node, and Cl(i) represents the user set connected to the wireless interface i; Its processing includes the following steps: Step 1: Initially allocate the spectrum: first process the input data, form the data into an undirected graph, and record the number of nodes as n; take a 1×n-dimensional zero matrix as the initial value CM of the color used by each node; the used color The total number is CON, and the initial value is 0; then, it is arranged in ascending order based on the degree, and the result is recorded as D, and all nodes are traversed, i,j=1,2,...,n, and the data adjacency matrix W(i,j ), select a minimum and different color value, and assign it to CM(j); if the color is not enough, then CON=CON+1, and color all nodes according to CM; finally, the blank The spectrum WH is evenly put into the channels of CON: p _i =WS/CON, i=1,2,...,CON, and the spectrum allocation solution at channel i in WH is obtained as

Step 2: Then use the distributed rate control algorithm to satisfy the stability constraints of the joint optimization model; the algorithm is composed of the rate controller at the node network layer and the link scheduler at the access point, The rate controller is used to control the rate of the message stream injected by the upper layer protocol in the network layer, and the buffer queue length of each transmission stream in the node determines the rate; The buffer length difference of the wireless link is scheduled for transmission, so that the overall length of the buffer queue in the subnet is minimized; then, according to the above:

and />

Make new allocations to the spectrum; />

This is an optimization model after substituting a specific utility function. Since this formula has been proven to be an NP-hardness problem, an approximate solution is required to calculate this problem; firstly, take the previous output color number CON, and set m=1, fk= 1. Specify the allowable error range ε=10 ^-5 ; when fk=1, perform linear programming in a loop, that is, deal with the problem: />

The calculated optimal solution is q ^(m) , and record d ^(m) = q ^(m) -p ^(m) Then, the judgment condition />

Is it true; if it is true, let fk=0, output p ^(m) , and the loop ends; otherwise, take d ^(m) as the feasible descending direction, and carry out an effective one-dimensional search, so as to determine the one-dimensional search step size; Ensure that the result is within the feasible region, specify the step size α∈[0,1]; then process: />

Obtain the optimal solution, denoted as α _m , let p ^(m+1) =p ^(m) +α _m *d ^(m) , and m=m+1, repeat the above process, where, f:R ⁿ →R ¹ is a differentiable function, p ^(m) represents the mth iteration point on p, ε is arbitrarily small and can be set according to the actual situation, α _m is the optimal step size of p ^(m) , and differentiable The function is

Set the factor τ=1, select p _i =WS/CON, i=1,2,...,CON as the initial point; The specific form of the utility function described in the joint optimization method for access control oriented to service quality assurance in the railway Internet of Things is:

Images