Congestion control method for overloaded service of smart substation under high concurrent users

Abstract

Overload service in the communication network of smart substation will cause congestion, resulting in low overload service throughput, high congestion rate and long congestion control time in the average smart substation. A congestion control method for overloaded services in smart substations with high concurrent users is proposed. According to the characteristics of overload service request of smart substation, the mathematical model of the algorithm is defined by describing the overload service request of smart substation on the basis of network topology model. Combined with the wavelength rotation strategy, the congestion rate of overloaded services in smart substations is reduced, and the throughput rate of overloaded services in smart substations is improved. Considering the factors of high concurrent users, by judging and feeding back the congestion of the overloaded services of smart substations, the congestion control of overloaded services of smart substations under high concurrent users is realized. The experimental results show that the proposed method has better effect and scalability in the congestion control of the overloaded service of the smart substation, and can effectively shorten the congestion control time of the overloaded service of the smart substation.

Keywords

High concurrent users smart substation wavelength rotation strategy overload service congestion control D_WA algorithm overload service congestion model congestion degree value interest packet forwarding rate

1 Introduction

At present, the State Grid proposes to build a smart grid with informatization, digitization, automation, and interaction as the basic technical characteristics. Smart substation is an important part of smart grid, which has the basic characteristics of digitalization of all station information, networked communication platform, and standardization of information sharing [1-3]. The information communication network is the basis for realizing the smart substation, and the smart substation business has extremely high requirements on the real-time, reliability and security of communication [4, 5]. People have put forward the need for unified networking of the whole station with the further development of the smart grid, it is that the structure of the communication network of the smart substation is developing towards the mode in which all the intelligent electronic devices in the station are connected to the same network. This networking method can maximize information sharing and improve network efficiency, but it also increases the uncertainty of network load and network delay. When the smart substation is overloaded with services, the burst traffic in the network may cause service congestion. The switch becomes the bottleneck of the communication network, the quality of network service is degraded, the communication delay of key business groups in the station increases, and the action behavior of sensitive equipment such as relay protection is affected, thereby affecting the safe and stable operation of the power grid. Therefore, it is of great significance to control the overloaded service congestion of smart substations.

At present, scholars in related fields have conducted research on congestion control of overloaded services in smart substations, and have made certain progress. Reference [6] proposes an adaptive congestion control method for EV charging in smart grids. The control method is decentralized and relies only on congestion signals generated by sensors deployed on the network. To dynamically tune the parameters of this congestion control method, the problem is transformed into multi-agent reinforcement learning, where each charge point is an independent agent that learns the parameter. Simulation results on a test distribution network with 33 primary distribution nodes, 1760 low-voltage end nodes and 500 electric vehicles are confirmed. The proposed method tracks the available capacity of the network in real time, prevents the problems of transformer overload and voltage limit violations over a longer period of time, and is able to adapt to changes in the distribution network. However, this method has the problem of high congestion rate in practical application, which leads to the congestion control effect not reaching the best. Reference [7] proposed a MPTCP congestion control method based on deep reinforcement learning for air-ground integrated railway network handover awareness. Firstly, the space-terrestrial integrated network-oriented high-speed rail (SGIN-HSR) and multipath transmission control protocol are introduced. Then, using the cross-layer information, based on deep reinforcement learning, a novel cross-layer auxiliary multi-path transmission control protocol congestion control mechanism for high-speed rail oriented to space-ground integrated networks is designed to alleviate the performance degradation caused by handover. Finally, the experimental results show that in the high-speed railway environment of the space-ground integrated network with frequent handovers, the high-speed railway congestion control significantly improves the effective throughput. However, the above method still has the problems of poor control effect and scalability, and long control time. Reference [8] proposes a bandwidth-adaptive congestion control protocol optimization method in V2 G networks. This algorithm perceives the network state by predicting the congestion level in packets, estimates the available bandwidth to judge the type of packet loss, and realizes the adaptive adjustment of congestion window. However, the congestion control time of this method is long in practical application, which leads to poor effect in practical application. Reference [9] puts forward the path congestion control algorithm for networks with large delays under active queue management. This method predicts the congestion of network paths of adjacent routing nodes and analyzes the queue status of network routing nodes. In order to optimize the subsequent node queue, transmission distance and transmission direction, the network path is optimized from three perspectives: path probability selection, packet discard function and WSN ant colony route selection, so as to realize path congestion control. However, this method has poor network throughput under high concurrent users.

Aiming at the above problems, this paper proposes a congestion control method for overload service of intelligent substation with high concurrent users. By describing the overload service request of smart substation and combining the wavelength rotation strategy, D_WA algorithm is introduced to define the mathematical model of the algorithm. Moreover, considering the factor of high concurrent users, the number of overload service of smart substation is calculated. Based on this number, the overload service congestion level of smart substation is recalculated to judge the congestion situation. The congestion control of overload service in smart substation is realized under high concurrent users. This study is expected to effectively alleviate the congestion of overloaded services in smart substations with high concurrent users.

2 Wavelength rotation strategy for smart substation overload service

Based on the characteristics of the overload service request of the smart substation, the relevant variables of the algorithm are first defined on the basis of the network topology model [8-10], and then the overload service request of the smart substation is described to define the mathematical model of the algorithm. Then, aiming at the resource conflict problem of overloaded services of smart substations, combined with the wavelength rotation strategy on the basis of the mathematical model, it aims to reduce the congestion rate of overloaded services of smart substations and improve the throughput rate of overloaded services of smart substations.

2.1 Description of overload service request of smart substation

The smart substation communication network can be represented by a connected graph G (Q, W, E). Among them, Q represents the node set in the communication network of the smart substation, W represents the link set, and E represents the wavelength set supported on each fiber. |Q|, |W|, and |E| represent the number of nodes, links, and wavelengths in the smart substation communication network, respectively.

The overload service transmission request in the communication network of the smart substation is expressed in the form of the overload service volume matrix R . In addition to the information of the source/destination node of the overloaded service transmission request, the overloaded traffic matrix should also contain the size of the bandwidth and other resources demanded by the overloaded service transmission request. The overloaded traffic matrix refers to the overloaded traffic transmission requests existing in the communication network of the smart substation, which is expressed as:

$R = {[{r^{k}}_{uo}]}_{| Q | | Q |}, k = 1, 2, . . ., n$ (1)

In formula (1), n is the total number of overloaded service requests in the communication network of the smart substation, and r^k_uo (q^k_u, q^k_o, B^k_uo) is an element of the overloaded traffic matrix, representing the overloaded service transmission requests k from the source node u to the destination node o. Wherein, q^k_u, q^k_o ∈ Q represents the source node and the destination node of the k overload service transmission request respectively, and B^k_uo represents the bandwidth required by the k overload service transmission request.

This paper considers two parameters, namely bandwidth and delay, that is, the parameter requirements corresponding to each overloaded service transmission request are expressed as: $A^{k} = (Δ b^{k}, Δ p^{k}), k = 1, 2, . . ., n$ (2)

In formula (2), Δb^k represents the bandwidth requirement interval of the overload service transmission request k, and Δp^k represents the delay requirement interval of the overload service transmission request k.

2.2 Wavelength rotation strategy

It is assumed that each smart substation communication network node maintains a wavelength cost table to record the wavelength cost of each wavelength. When a certain smart substation communication network node completes the routing calculation for a connection request, the wavelength overhead of the wavelength occupied by the route is set to a preset maximum value, and thereafter the overhead value continues to decay with time. The wavelength cost is added to the cost of all possible routes in the process of route calculation, and the route with the least total weight is selected to construct the route with overloaded service. With this strategy, “wavelength rotation” can be realized in route calculation of overloaded services. That is to say, the same wavelength should be avoided as far as possible to establish connection for overloaded services of intelligent substations arriving together, so as to reduce resource conflicts in the process of road construction, thus reducing blocking probability of communication network of intelligent substations, namely, the minimum assembly should be used to build routes for overloaded services, thus reducing blocking probability. This is the basic idea of the wavelength label rotation strategy. Using the wavelength rotation strategy, the overloaded service resource conflict of the smart substation can be avoided. The four-node dual-wavelength network topology is shown in Fig. 1.

Fig. 1

Four-node dual-wavelength network topology.

First, the decay time t of the wavelength overhead is set to 200 ms, the maximum value ρ_max of the wavelength overhead is 10, and the initial value of the wavelength overhead of wavelength 1 and wavelength 2 is both 0.

After node 4 selects the (4⟶2⟶1) route of wavelength 1 to build a route for the overloaded service P₃, it changes the wavelength cost of wavelength 1 in the wavelength cost database it maintains to ρ_max = 10. Another overload service request arrives on the communication network node 4 of the smart substation: overload service 2 (4⟶2). Two possibilities of its arrival time are expounded separately: the arrival time interval with the overload service P₃ is 20 ms less than the t overload service; the arrival time interval with the overload service P₃ is 1 s greater than the t non-overload service.

As shown in Fig. 1 (a), since the path construction information of the overload service P₃ has not completed the whole network synchronization at this time, the links (4 ⟶ 2) and (2 ⟶ 1) of the wavelength 1 used by the overload service Q are still idle in the link state database of the node 4. Therefore, the calculated shortest route results are (4⟶2) for wavelength 1 and (4⟶3⟶1⟶2) for wavelength 2, and their routing costs are 1 and 4, respectively. At this time, according to the linear attenuation mode, the calculation formula of the wavelength overhead value of the wavelength α at time t is: $ρ_{α} = {\begin{matrix} 0, 1 \geq σ_{α} + t \\ \frac{ρ_{max} (σ_{α} + t - 1)}{t}, σ_{α} \leq 1 \leq σ_{α} + t \end{matrix}$ (3)

In the formula (3), σ_α is the time when the wavelength α overhead is last set to the preset maximum value ρ_max. The decay time t of the set wavelength overhead can be set as the average data synchronization time of the link state database maintained by each smart substation communication network node after the smart substation communication network state changes. The wavelength cost of wavelength 1 is calculated to be 9, and the wavelength cost of wavelength 2 is still the initial value of 0. Therefore, the total cost of the two routes is 10 and 4 respectively, and node 4 selects the wavelength 2 (4⟶3⟶1⟶2) with the smallest total cost to build a route for the overloaded service P₄.

As shown in Fig. 1 (b), since the road construction information of overloaded service P₃ has been synchronized across the network at this time, the links (4⟶2) and (2⟶1) of wavelength 1 used by it are in the link state database of node 4 in the occupied state. Therefore, the calculated shortest route results are (4⟶3⟶2) for wavelength 1 and (4⟶3⟶1⟶2) for wavelength 2, and their routing costs are 2 and 4, respectively. At this time, according to the calculation formula of the wavelength cost value in the linear attenuation mode, the wavelength cost of wavelength 1 is calculated to be 0. The wavelength overhead of wavelength 2 is still the initial value of 0. Therefore, the total cost of the two routes is still 2 and 4. After that, the node 4 selects the wavelength 1 with the smallest total overhead (4⟶3⟶2) to perform a path construction operation for the overloaded service P₄.

It can be seen that wavelength rotation can successfully avoid resource conflicts of overloaded services, automatically identify overloaded services and non-overloaded services, and select the optimal route for overloaded services on the premise of avoiding resource conflicts as much as possible.

2.3 Basic principles of D_WA algorithm

The core idea of D_WA (Dynamic Wavelength Alternation) algorithm mainly includes two aspects. First, the dynamic weight technology in load balancing is used to make the distribution of communication network resources in smart substations more uniform [11-13]. At present, the unbalanced load of the communication network of the smart substation is mainly due to the unbalanced occurrence of overloaded service requests, mainly from three aspects: overloaded service rate, burstiness, and office-direction distribution.

The D_WA algorithm mainly achieves the purpose of load balancing of communication network resources in smart substations by changing the influencing factors in the dynamic weights. On the basis of the link w_ij weight D_d, the dynamic attenuation value of the wavelength overhead is set, and the influence of the wavelength overhead is considered when calculating the weight. The smart substation communication network resources are used more fully, the resource utilization rate of overloaded services is improved, the road construction conflict of the smart substation communication network is avoided, and the blocking rate is reduced. It is more reasonable to adopt an adaptive routing strategy for overloaded services, then the routing weights in the communication network of the smart substation are as follows: $D_{F} = F_{R} = \sum_{(u, o) \in L} \frac{s (u, o)}{a (z, x) Y (w)}$ (4)

In formula (4), (u, o) is a pair of source/destination nodes in the communication network of the smart substation, L is the set of all source/destination node pairs, and s (u, o) is the minimum capacity between source/destination node pairs (u, o). a (z, x) is the remaining capacity of the link, Y (w) is the remaining capacity of the link F_R, and w is the link cost. The purpose is to guide the overloaded service request to the link with larger remaining capacity.

Assuming that β_c records the wavelength cost value, the allocated wavelength is recorded as a maximum value and the value decays linearly with the formula (3), then ρ_max is the sum of the link weight and the wavelength cost value, namely:

$\begin{matrix} D_{F} = & F_{R} + ρ_{α} = \sum_{(u, o) \in L} \frac{s (u, o)}{a (z, x) Y (w)} \\ + \frac{ρ_{max} (σ_{α} + t - 1)}{t} \end{matrix}$ (5)

When the number of remaining wavelengths β_f (w_ij) on link w_ij is equal to 0, there are no available resources on link w_ij, and the current overload service connection request cannot be carried. Therefore, the link cost F_R is set to infinity;

When the number of remaining wavelengths β_f (w_ij) on the link w_ij is not equal to 0, but when the wavelength overhead has an instantaneous value, the total weight depends on the sum of the two parts. In this way, the resources that have been used are prevented from being reused, resulting in resource conflicts in the communication network of the smart substation.

2.4 The basic process of D_WA algorithm

The above has analyzed the resource conflict problem in the overload service of smart substations and the process of avoiding the road construction conflict by the wavelength rotation strategy. The basic flow of D_WA algorithm is shown in Fig. 2.

According to Fig. 2, the basic process of D_WA algorithm is divided into five steps, which are as follows:

Fig. 2

Basic flow of d_WA algorithm.

Step 1: Generate plane routes for each wavelength. When an overloaded service request arrives at a certain node of the communication network of the smart substation, the control plane server of the communication network of the smart substation of the node uses the dynamic optimization algorithm to generate the route of each wavelength plane for the sub-database of each wavelength in the link state database surface.

Step 2: Find the route with the lowest cost of routing overloaded service requests. In the routing table of each wavelength plane described in step 1, respectively find the route with the minimum routing cost for the overloaded service request, and at most can find multiple routes with the same number of wavelengths as the number of wavelengths configured in the communication network of the smart substation; If one or more routes can be found, go to the next step; otherwise, it is considered that congestion occurs, and the route construction failure information is output, and the route calculation and route construction process ends.

Step 3: Calculate the total weight of wavelength cost and routing cost. Query the wavelength cost value of the wavelength used by the multiple routes described in step 2 in the wavelength cost database, and add the value to the route cost value to obtain their respective total weights.

Step 4: Select the route with the smallest total weight for dynamic road construction. Before road construction, set the wavelength cost value of the wavelength used by the route as a preset maximum value ρ_max in the wavelength cost database. This value is required to be much larger than the maximum possible routing overhead value in the communication network of the smart substation, after which the overhead value decays linearly with formula (3).

Step 5: Dynamic construction of intelligent substation communication network overload service request channel. The control plane server of the communication network of the smart substation adopts the resource reservation protocol to dynamically build a road for the overload service request of the communication network of the smart substation. After the road construction is successful, the link state database maintained by each node is updated and synchronized, and the road construction process ends.

3 Research on congestion control method for overloaded services under high concurrent users

When the overloaded service of the smart substation is congested, considering the factors of high concurrent users will greatly improve the performance of the communication network of the smart substation and reduce the congestion of the overloaded service. Therefore, the congestion control strategy of the overloaded service of smart substations under high concurrent users is studied.

3.1 Judgment of overloaded service congestion under high concurrent users

The reason for considering the situation of high concurrent users is that high concurrent users are common in the communication network of real smart substations. There are often different types of overloaded services of smart substations in the communication network of smart substations, and they have different demands on the communication network resources of smart substations. It is necessary to add the attribute field and priority field of the overload service of the smart substation to the data packet in order to easily identify the types of data packets in the communication network of the smart substation. The smart substation overload service attribute field is used to mark the smart substation overload service type, and the priority field is used to distinguish data packets of different priorities of the same smart substation overload service. When carrying out the congestion control of the overloaded service of the smart substation, first provide different rate adjustment strategies according to the different overloaded services of the smart substation, and then consider the priority of the overloaded service of the same type of smart substation.

There may be requests from multiple high concurrent users in the interest queue for a router interface, or there may be multiple different requests from the same high concurrent user. The simplified overload service congestion model under high concurrent users is shown in Fig. 3.

Fig. 3

Simplifies the overloaded service congestion model under high concurrent users.

In Fig. 3, requests from multiple different high-concurrency users are forwarded to an interface of the router. At this time, the mechanism of congestion control is to calculate a smart substation overload service congestion degree value according to the current packet situation in the transmission queue, and then use the obtained smart substation overload service congestion degree value to judge the congestion state. When calculating the congestion level value of overloaded services in smart substations, different types of overloaded services in smart substations have different bandwidth requirements. Through the formula (6), the value of the congestion degree of the overloaded service of the smart substation can be obtained: $F = z_{q} \times x_{q} + z_{w} \times x_{w} + z_{e} \times x_{e} + z_{r} \times x_{r}$ (6)

In formula (6), x_q, x_w, x_e, x_r and z_q, z_w, z_e, z_r are the resource class, database class, interaction class, and background class, respectively, the number of overloaded services of smart substations and the corresponding bandwidth requirements.

The method of exponentially weighted moving average [14, 15] is used to calculate the average value of the number of overloaded services in various smart substations in this paper, which can eliminate the error caused by the congestion control of overloaded services in smart substations. For a certain smart substation overload service, the average quantity x_avg can be obtained by formula (7): $x_{avg} = (1 - z_{n}) \times x_{avg} + z_{n} \times x$ (7)

In formula (7), x is the detected number of overloaded services of smart substations, z_n is the weight, and this paper takes 0.5. For each smart substation overload service, after calculating the average number by formula (7), substitute formula (6) to calculate the congestion degree value of the smart substation overload service.

In this paper, according to the severity of the congestion degree of the overloaded business of the smart substation, the overloaded business of the smart substation is divided into three categories, and two thresholds F_busy and F_congestion of the overloaded business of the smart substation are set at the same time. When the overloaded service congestion level value of the smart substation is less than F_busy, the link is in an idle state. When the overloaded service congestion level of the smart substation is between the two, the link is in a busy state. When the overloaded service congestion level value of the smart substation is greater than F_congestion, the link is in a congested state.

3.2 Congestion feedback of overloaded services under high concurrent users

The state of the communication network of the smart substation is detected in the forwarding device on the basis of the overload service congestion judgment under the above-mentioned high concurrent users. Then, the communication network status of the smart substation is explicitly fed back to the high-concurrency users, and the high-concurrency users adjust the packet sending rate. The goal is to avoid or deal with the overloaded service congestion of the smart substation without affecting the overall response time, thereby effectively improving the service quality. If it is found that the communication network status of the smart substation is congested according to the calculation of the congestion degree value of the overloaded service of the smart substation, the overloaded service requests of some smart substations in the queue are removed in time.

The method adopted in this paper is to feed back the congestion information of the overloaded service of the smart substation to the receiver after the congestion of the overloaded service of the smart substation. Add two fields of smart substation overload service type and priority to the data packet, so that the forwarding device can easily distinguish different data packets. After detecting that overloaded services are congested in the communication network of the smart substation, the overloaded services with the lowest latency requirements are adjusted first to ensure the latency requirements of most high-concurrency users. If there are multiple data requests in the same smart substation overload service type, the overload service with lower priority is adjusted first to ensure the service quality of the high-priority smart substation. As long as the calculated value of the congestion level of the overloaded service of the smart substation is greater than the threshold F_congestion of the overloaded service of the smart substation, the corresponding requests will continue to be removed from the queue according to the above principles. The overloaded service congestion level of the smart substation is detected at regular intervals to ensure the timeliness of the status feedback of the communication network of the smart substation. The overloaded service congestion feedback algorithm under high concurrent users is described as follows:

Enter the Interest package queue Interest_list. If F > F_congestion, there is a background-type smart substation overload service in the pending_list, and put the Interest package with the lowest priority among the background-type smart substation overload services into the handling_list. The link congestion information is fed back to the corresponding smart substation overload service. This Interest package is deleted from the pending_list. The number of smart substation overload services in the pending_list is calculated as: $x {(p)}_{avg} = (1 - z_{n}) \times x {(p)}_{avg} + z_{n} \times x (p)$ (8)

In formula (8), x (p) is the detected number of overloaded services of smart substations in pending_list.

Recalculate the overloaded service congestion level value of the smart substation:

$\begin{matrix} F = & z_{q} \times x_{q} (p) + z_{w} \times x_{w} (p) + z_{e} \times x_{e} (p) \\ + z_{r} \times x_{r} (p) \end{matrix}$ (9)

In formula (9), x_q (p) , x_w (p) , x_e (p) , x_r (p) is the number of overloaded services of smart substations in the pending_list of resource, database, interaction, and background, respectively. In the order of resource class, database class, and interaction class, follow the above steps to determine whether there are other smart substation overload services in pending_list.

The above algorithm feeds back the congestion status of the communication network of the smart substation to the high concurrent users in time when the communication network of the smart substation may be overloaded with service congestion.

3.3 Congestion control of overloaded services under high concurrent users

After the overloaded service congestion feedback of high-concurrency users is used, the overloaded service congestion control of smart substations will be carried out for high-concurrency users. By monitoring the number of overloaded services of smart substations in the queue to be transmitted, the routing node adjusts the sending rate of Interest packets.

It is necessary to calculate the total bandwidth required by the overloaded service of the smart substation in the current interest packet queue according to formula (6) before adjusting the interest packet rate. If the total bandwidth is greater than the capacity of its upstream link, the Interest packet forwarding rate needs to be adjusted. After calculating the total bandwidth required by the overloaded service of the smart substation in the Interest queue, the various streams are divided into three categories, namely, remain_Interest_list, reject_Interest_list, and block_Interest_list. The three respectively store the content name of the stream that ensures the quality of service of the smart substation, the content name that needs to reject the request, and the content name that needs to reduce the transmission rate. remain_Interest_list is initialized to Interest_list. When the total bandwidth demand is greater than the upstream link capacity, the flow with the lowest latency requirement and the lowest priority will be removed from it. Add it to the reject_Interest_list, and this process will continue until the total bandwidth demand of the overloaded service of the smart substation in the remaining_Interest_list is less than the upstream link capacity. The content name in block_Interest_list is selected from reject_Interest_list. How many smart substation overload service flows to reduce the rate can be selected, which can be changed according to the specific situation, and the smart substation overload service flow with the highest priority is selected.

The rate is adjusted according to its bandwidth requirements for the overloaded service flows of smart substations in the remaining_Interest_list. If the overloaded service flow of the smart substation is a resource class and its bandwidth requirement is z _ c, then its Interest packet forwarding rate should be: $v_{c} = z_{c} \times \frac{b_{I}}{b_{D}}$ (10)

In formula (10), b_I, b_D is the average size of the Interest packet and the Data packet of the stream, respectively. For the overloaded service flows of smart substations in block_Interest_list, the remaining bandwidth is required to adjust the rate. If the total bandwidth requirement of the overloaded smart substation service in the remain_Interest_list is M, the maximum bandwidth occupied by the overloaded service of the smart substation in the block_Interest_list is N - M. Among them, N is the upstream link capacity. Taking an overloaded service flow of a smart substation to reduce the rate, the interest packet forwarding rate of this flow is: $v = (N - W) \times \frac{b_{I}}{b_{D}}$ (11)

If there are multiple flows in the block_Interest_list, the remaining bandwidth can be divided equally or in proportion to the bandwidth requirement, and its Interest packet forwarding rate is the same as formulas (10) and (11). For the overloaded service flows of smart substations in reject_Interest_list, send denial-of-service feedback to high-concurrency users of the corresponding flow.

High-concurrency users can take many measures after finding that the requests sent are slowed down, such as abandoning requests on this path or continuing to send requests. Since this paper focuses on the congestion control of overloaded smart substation services under high-concurrency users, it is assumed that high-concurrency users continue to send requests. When a high-concurrency user receives a denial of service feedback, it will suspend sending requests to this path, and then send it again after a period of time. When high concurrent users receive feedback from the reply service, they will continue to send requests to this path. The congestion control of the overloaded service of the smart substation under high concurrent users can be effectively realized through the above process.

4 Experimental simulation and analysis

4.1 Set up the experimental environment

In order to verify the effectiveness of the congestion control method for overloaded services in smart substations with high concurrent users, ndnSIM is used as an experimental platform to conduct simulation experiments. The hardware and system configuration of the host is: Intel Core 2.93 GHz processor, 4.0 G memory, 500 G hard disk, and the operating system is Ubuntu14.04. In the simulation experiment, the number of overloaded services in a 500MB smart substation is selected, and the proposed method, the method of reference [6] and the method of reference [7] are compared to verify the effectiveness of the proposed method.

4.2 Comparison of congestion control effects of overloaded services in smart substations

The congestion rate is taken as the evaluation index in order to verify the effect of the proposed method on the congestion control of the overloaded service of the smart substation. It is calculated as follows: $μ = \frac{θ_{s}}{θ_{z}}$ (12)

In formula (12), θ_s represents the number of overloaded service requests of smart substations that fail to establish, and θ_z represents the total number of overloaded service requests of smart substations. The method of reference [6], the method of reference [7] and the proposed method are used to compare, and the comparison results of the overload service congestion rate of smart substations of different methods are obtained as shown in Fig. 4.

Fig. 4

Comparison results of overloaded service congestion rates in smart substations with different methods.

According to Fig. 4, when the number of overloaded services in smart substations is 500MB, the average congestion rate of overloaded services in smart substations by the method of reference [6] is 2.87%, and the average congestion rate of overloaded services in smart substations by the method of reference [7] is 4.98%. The average smart substation overload service congestion rate of the proposed method is only 0.82%. It can be seen that the proposed method has a lower congestion rate of the overloaded service of the smart substation, and has a better congestion control effect of the overloaded service of the smart substation.

4.3 Scalability comparison of overloaded business congestion control in smart substations

In order to further verify the scalability of the proposed method for congestion control of overloaded services in smart substations, the throughput rate is used as an evaluation index. The method of reference [6], the method of reference [7] and the proposed method are used to compare, and the comparison results of the overload service throughput rate of smart substations of different methods are obtained as shown in Fig. 5.

Fig. 5

Comparison results of overload service throughput of smart substations with different methods.

Analysis of Fig. 5 shows that when the number of overloaded services in the smart substation is 500MB, the average smart substation overload service throughput rate of the method of reference [6] is 90.8%, and the average smart substation overload service throughput rate of the method of reference [7] is 84.2%. The average smart substation overload service throughput rate of the proposed method is as high as 97.5%. It can be seen from this that the proposed method has a higher throughput rate of the overloaded service of the smart substation and better scalability of the congestion control of the overloaded service of the smart substation.

4.4 Congestion control time comparison of overloaded services in smart substations

On this basis, the proposed method is verified for the overload traffic congestion control time of the smart substation. The proposed method, the method of reference [6] and the method of reference [7] are used for comparison respectively, and the comparison results of congestion control time of overload service in smart substations with different methods are shown in Table 1.

Analysis of Table 1 shows that with the increase in the number of overloaded services in smart substations, the congestion control time of overloaded services in smart substations with different methods increases. When the number of overloaded services of the smart substation is 500MB, the congestion control time of the overloaded services of the smart substation by the method of reference [6] is 20.8 s, and the congestion control time of the overloaded services of the smart substation by the method of the method of reference [7] is 24.8 s. However, the congestion control time of the smart substation overloaded service by the proposed method is only 15.1 s. It can be seen that the congestion control time of the overloaded service of the smart substation of the proposed method is short.

Table 1
Comparison results of congestion control time for overloaded services in smart substations with different methods

Number of The proposed The method of The method of

overloaded services method/s reference [6]/s reference [7]/s

in smart

substations/MB

100 5.9 8.9 10.8

200 7.5 12.1 13.9

300 9.8 15.8 17.1

400 12.3 18.9 20.5

500 15.1 20.8 24.8

Number of	The proposed	The method of	The method of
100	5.9	8.9	10.8
200	7.5	12.1	13.9
300	9.8	15.8	17.1
400	12.3	18.9	20.5
500	15.1	20.8	24.8

4.5 Comparison of anti-interference ability

There is some interference in the actual operation of the network, so it is necessary to analyze the anti-interference capability of the proposed method. On the basis of the experiment in 4.3, interference signals are added to interfere with congestion control, and the changes of throughput under interference are analyzed. The proposed method, the method of reference [6] and the method of reference [7] are respectively used for comparison. The experimental results are shown in Fig. 6.

Fig. 6

Comparison results of anti-interference ability under no-pass method.

According to the data in Fig. 6, the throughput rate of the three methods is reduced in the presence of interference. However, the throughput rate reduction of the proposed method is small, while that of the literature method is large, indicating that the proposed method has strong anti-interference ability. Detailed analysis shows that the throughput rate of the proposed method is still higher than 96.0%, while that of the method of reference [6] is reduced to less than 90.5%, and that of the method of reference [7] is less than 87.3%. Compared with the three methods, the throughput rate of the proposed method is more than 5.5% higher than that of the method in literature. The proposed method has higher anti-interference ability and can effectively resist interference.

5 Conclusion

This paper proposes a congestion control method for overloaded services of smart substations under high concurrent users, and considers the factors of high concurrent users to achieve congestion control of overloaded services in smart substations. The method has better effect and scalability of the congestion control of the overloaded service of the smart substation, and can effectively shorten the time of congestion control of the overloaded service of the smart substation. However, this method only conducts preliminary research on the characteristics of the overload service of smart substations in the process of road construction. However, this method does not take into account the different priorities of different types of traffic and the data loss caused by congestion, which may affect the congestion control. However, when the network topology is complex or the load distribution is unbalanced, the congestion can still be effectively controlled without being affected by the network topology and load distribution. In view of the shortcomings of this method, QoS technology can be adopted, because QoS technology can improve the quality of network service by assigning different priorities to different types of traffic, and can make congestion control more intelligent, and can dynamically adjust the congestion control strategy according to the network load and service quality requirements. Thus, the congestion control effect of overload service in smart substation can be further improved under high concurrent users.

Conflicts of interest

The authors declare that there are no conflicts of interest regarding the publication of this paper.

References

Liu

, Gao

, Du

, Yu

, Modular design of intelligent substation SCD file management and control system, Dianli Xitong Baohu yu Kongzhi/Power System Protection and Control 47(3) (2019), 154–159.

Tao

, Zhong

, Wang

, Dong

, Jiang

Engineering practice of on-line monitoring and intelligent diagnosis system for secondary equipment in smart substation. 2020 IEEE 3rd International Conference on Electronics Technology (ICET). IEEE (2020), 397–403.

Zheng

, Ge

, Zhang

, Xu

, Sun

, High-reliability optical process level network in smart substation, Optical Switching and Networking 36(Feb.) (2020), 100552.1–100552.8.

Qiu

, Qiu

, Lu

, Secure V2X Communication Network based on Intelligent PKI and Edge Computing, IEEE Network 34(2) (2020), 172–178.

Pan

, Yao

, Mai

, Zhang

, Liu

, Scalable traffic control using programmable data planes in a space information network, IEEE Network 35(4) (2021), 35–41.

Zishan

A.A.

, Haji

M.M.

, Ardakanian

, Adaptive congestion control for electric vehicle charging in the smart grid, IEEE Transactions on Smart Grid 12(3) (2021), 2439–2449.

, Ai

, Deep reinforcement learning for handover-aware MPTCP congestion control in space-ground integrated network of railways, IEEE Wireless Communications 28(6) (2021), 200–207.

Jiang

Y.F.

, Liang

X.Y.

, Tang

J.Y.

, Bandwidth adaptive congestion control protocol optimization in V2G networks[J], Application Research of Computerss 38(12) (2021), 3719–3724.

Liu

G.F.

, Zhang

, Path congestion control algorithm for large delay networks based on active queue management[J], Computer Simulation 38(3) (2021), 268–271, 491.

10.

Hautamaki

, Hamalainen

A model of Cyber Threat Information Sharing with the Novel Network Topology. The 12th International Conference on Advances in Information Technology (2021), 1–10.

11.

Zhang

, Qi

, Wei

, Zhang

, Feng

, Hou

, High-priority first dynamic wavelength and bandwidth allocation algorithm in TWDM-PON, (MAR.), Optical Fiber Technology 48 (2019), 165–172.

12.

Bibi

, Khan

F.Z.

, Ahmad

, Naseem

, Mahar

J.A.

, Dynamic wavelength grouping for quality of service in optical packet switching, , IEEE Access 9 (2021), 60946–60959.

13.

, Yu

, Zhao

, Zhang

Dynamic Wavelength and Key Resource Adjustment in WDM based QKD Optical Networks. 2020 Asia Communications and Photonics Conference (ACP) and International Conference on Information Photonics and Optical Communications (IPOC). IEEE (2020), 1–3.

14.

Yan

, Digital multimedia network communication delay fast elimination method simulation, Computer Simulation 36(01) (2019), 218–221.

15.

Habeeb

A.S.

, Compare prediction by autoregressive integrated moving average model from first order with exponential weighted moving average, Journal of Economics and Administrative Sciences 26(120) (2020), 426–439.