Research on the Path of Improving Network Attack and Defense Ability of Students in Applied Colleges and Universities Based on Information Security Technology

In an undirected network, the degree is defined as the number of connected edges of the node with other nodes in the network, i.e., the number of nodes with path length 1 to this node. Denote the degree of node i as k_i and the average degree as <k>, then: (1)<k>=1N∑i=1Nki=2MN$$ < k > = \frac{1}{N}\sum\limits_{i = 1}^N {{k_i}} = \frac{{2M}}{N}$$

Where M is the number of connected edges and N is the number of nodes.

In directed networks, edges are considered as vectors with starting and ending points, and the sum of edges ending at a node is called the in-degree of the node, and vice versa, the sum of edges starting at this node is called the out-degree of the node, which are denoted by kiin$$k_i^{in}$$ and kiout$$k_i^{out}$$, respectively. In various complex directed graphs, although the in-out degree of each node may not be the same, a vector edge must have a starting point and an ending point, so as a whole, the average in-degree of the network <kim>$$ < k_i^m >$$ and the average out-degree of the network <kiout>$$ < k_i^{out} >$$ are the same, and its expression is: (2)<kiin>=<kiout>=MN$$ < k_i^{in} > = < k_i^{out} > = \frac{M}{N}$$

In a weighted undirected network, the degree is defined as the strength of a node, W = (w_ij) denotes the weight matrix of the network, and the strength of node i is: (3)si=∑j=1Nwij$${s_i} = \sum\limits_{j = 1}^N {{w_{ij}}}$$

Similarly, in a weighted directed network, the intensity of node i consists of the outgoing intensity and the incoming intensity, two components, i.e.: (4)siout=∑j=1nwij siin=∑j=1mwij$$\begin{array}{l} s_i^{out} = \sum\limits_{j = 1}^n {{w_{ij}}} \\ s_i^{in} = \sum\limits_{j = 1}^m {{w_{ij}}} \\ \end{array}$$

Due to the large size of the actual network, the degree of individual nodes does not portray some of the properties of the network well, so the degree distribution is introduced to capture the overall nature of the degree. The degree distribution is the number of nodes in a specific network with a particular degree value of k as a percentage of the total, denoted as P(k) [40].

The length of the shortest path between two points in a network is called the distance between these two nodes. However the number of shortest paths between two points is estimated to be multiple, but the distance is certain. The mean of the sum of the distances between all pairs of nodes is the average path length of the network L, which is expressed as: (5)L=1N(N−1)/2∑i≥jdij$$L = \frac{1}{{N(N - 1)/2}}\sum\limits_{i \geq j} {{d_{ij}}}$$

Network connectivity is a necessary condition for the above equation to hold. If the network is not connected, the result obtained from the above equation will become infinite, and in order to more conveniently represent the average path length L of the network, one defines it as the simple harmonic mean between two points, i.e.: (6)L=1GE,GE=1N(N−1)/2∑i≥j1dij$$L = \frac{1}{{GE}},GE = \frac{1}{{N(N - 1)/2}}\sum\limits_{i \geq j} {\frac{1}{{{d_{ij}}}}}$$

Although many of the actual network nodes are numerous and structurally complex, their L is generally small.

In a real network, there may also be connections between the neighbors of node i, and the magnitude of this possibility can be expressed by the clustering coefficient, which reflects the degree of connectedness between the neighbors of node i. The maximum number of edges obtained between the neighbors of node i with degree k_i by connecting them two by two is (k_i(k_i − 1))/2. Assuming that the real number of connected edges between its neighbors is Ei(Ei≤ki(ki−1)/2)$${E_i}({E_i} \leq {k_i}\left( {{k_i} - 1} \right)/2)$$, the clustering coefficient C_i of node i can be expressed as: (7)Ci=Eiki(ki−1)/2=2Eiki(ki−1)$${C_i} = \frac{{{E_i}}}{{{k_i}({k_i} - 1)/2}} = \frac{{2{E_i}}}{{{k_i}({k_i} - 1)}}$$

Where k_i > 1. The clustering coefficient C_i = 0 when node i has degree k_i < 1, i.e., it is an isolated node or has only one node connected to it. The clustering coefficient of the network c refers to the ratio of the clustering coefficient sum of each node individually to the total number of nodes in the network in the overall network, viz: (8)C=1N∑i=1NCi$$C = \frac{1}{N}\sum\limits_{i = 1}^N {{C_i}}$$

0 ≤ C ≤ 1 of them.

There are three main types of common regular networks: global coupling networks, nearest neighbor coupling networks, and star coupling networks [41]. 1)

Globally coupled network: every node has a connecting edge to the remaining N − 1 node (where N is the total number of nodes in the network), and a network that is saturated with such a number of edges is called a globally coupled network. From the properties of such a network, it has the closest average path length L = 1, the highest clustering coefficient C = 1, and the largest number of edges of any network N(N − 1)/2. However, many real networks are sparse, and the order of magnitude of the total number of edges in the network is N rather than N².

In order to model the properties of real networks, the ER stochastic model has been proposed with the construction algorithm: 1)

Given N node and a connected edge probability P(P ∈ [0, 1]).

The average path length of the ER network is LER∞lnNln<k>$${L_{ER}}\infty \frac{{\ln }}{N}\ln < k >$$, which shows that the ER model conforms to the characteristics of a small world. Its clustering coefficient is C=p=<k>N−1$$C = p = \frac{{ < k > }}{{N - 1}}$$, which indicates that when the size of the random network is large, its clustering coefficient is small and there is no obvious clustering characteristics. The average degree of the ER model network is <k > = p(N − 1) ≈ pN, and it is also a homogeneous network whose overall degree distribution obeys the Poission distribution, whose expression is: (9)P(k)=( N−1 k)pk(1−p)N−1−k≈<k>kk!e−<k>$$P(k) = \left( {\begin{array}{*{20}{c}} {N - 1} \\ k \end{array}} \right){p^k}{(1 - p)^{N - 1 - k}} \approx \frac{{ < k{ >^k}}}{{k!}}{e^{ - < k > }}$$

Many actual networks, despite their large number of nodes and complex structure, not only do not have large average path lengths, but also have relatively small clustering coefficients. Both ER random networks and nearest-neighbor coupling networks do not have the “small average path length” and “large clustering coefficient” characteristics of real networks.

The WS model improves on the shortcomings of the above model with the following construction algorithm: 1)

Generate a regular graph: Generate a regular network with the number of nodes N by the generation rule of the nearest neighbor coupling network.

In this paper, we improve the WS model based on the WS model to construct the NW small-world network model, and its construction algorithm is as follows: 1)

Given a rule graph: generate a rule network with the number of nodes N by the generation rule of nearest neighbor coupling network.

The WS small world network clustering coefficient is: (10)C(p)=3(K−2)4(K−1)(1−p)3$$C(p) = \frac{{3(K - 2)}}{{4(K - 1)}}{(1 - p)^3}$$

The NW small world network clustering coefficient is: (11)C(p)=3(K−2)4(K−1)+4Kp(p+2)$$C(p) = \frac{{3(K - 2)}}{{4(K - 1) + 4Kp(p + 2)}}$$

The average path length for both WS and NW models can be approximated as: (12)L = 1n(NKp)K2p$$\begin{array}{*{20}{c}} L& = &{\frac{{1{\text{n}}(NKp)}}{{{K^2}p}}} \end{array}$$

The WS small world network degree distribution is:

If k ≥ K/2, then: (13)P(k)=∑n=0min(k−K/2,K/2)( K/2 n)(1−p)np(K/2)−n(pK/2)k−(K/2)−n(k−(K/2)−n)!e−pK/2$$P(k) = \sum\limits_{n = 0}^{\min (k - K/2,K/2)} {\left( \begin{array}{c} K/2 \\ n \\ \end{array} \right)} {(1 - p)^n}{p^{(K/2) - n}}\frac{{{{(pK/2)}^{k - (K/2) - n}}}}{{(k - (K/2) - n)!}}{e^{ - pK/2}}$$

If k ≤ K/2, then P(k) = 0. The NW small-world network degree distribution is: (14)P(k)=( N k−K)(KpN)k=K(1−KpN)N−k+K$$P(k) = \left( {\begin{array}{*{20}{c}} N \\ {k - K} \end{array}} \right){\left( {\frac{{Kp}}{N}} \right)^{k = K}}{\left( {1 - \frac{{Kp}}{N}} \right)^{N - k + K}}$$

The improved model discards the original rule of “randomized reconnection”, and replaces it with fixing the connecting edges in the original regular network, and directly adds “randomized edges” in it, which can simultaneously satisfy the two characteristics of the real network, and realize the attack and defense simulation of the student network [43].

In order to verify the protection effectiveness of the information security protection system against multi-layer network simulation attacks, the complex student information network coupled with the physical network (including servers, routers, switches, etc.) and the information network (including the student management system, library system, campus social platforms, etc.) of a university is selected as an experimental object, and the attack and defense simulation is carried out on the complex student network, and the changes of the complex network and sub-network are observed. In order to compare and analyze the structural characteristics and anti-destruction characteristics of the complex student information network of the university, this paper mainly uses the complex student information network and the library system to simulate the attack and defense.

In this paper, the attack and defense simulation mainly adopts the complex network attack mode, which is to simulate the attack on the nodes of the complex network as a whole, and the attack parameters are set as follows: the number of preset attack rounds is set to 50, the attack mode is selected as the complex network attack mode, the target of attack is selected as the complex network, the number of attacks is set to 3, and the attack strategy is selected as the degree of centrality.

In the information security protection system for multi-layer network attack static experiments in this information security protection system operating steps are as follows: 1)

Enter the multilayer network experiment interface, select the experimental network, click the load network button, after loading the completion of each sub-network, click the fusion network button, the visualization display module displays each sub-network and complex student network topology.

In this section, maximum connectivity, network efficiency, number of connected slices and maximum connected slice size are selected as network evaluation metrics to attack the complex network through the complex network attack pattern, and the changes of network evaluation metrics are analyzed graphically.

Figure 2 demonstrates the change of network efficiency under the degree centrality strategy attack, where the horizontal coordinate is the number of attack rounds and the vertical coordinate is the network evaluation index value. It can be seen that under the degree centrality descending attack, the network efficiency of complex student network, information network and physical network decreases with the increase of the number of attack rounds, which is due to the fact that with the increase of the number of attack rounds, the connecting edges between nodes and nodes are reduced, and the shortest paths between the nodes will become longer, which leads to the decrease of network efficiency. The decreasing trend of network efficiency of physical network is more obvious than the decreasing trend of complex student network and information network, this is due to the fact that the physical network basically disintegrates in the early stage of the attack, while the decreasing trend of network efficiency of complex student network and information network is similar, and their resistance to the attack is similar, which indicates that the information network is dominant in the complex student network.

Figure 2.

Variation of net work efficiency under betweenness centrality strategy attacks

The variation of maximum connectivity degree under degree centrality strategy attack is shown in Fig. 3. Under the degree centrality descending attack, the maximum connectivity of the complex student network, information network and physical network decreases with the increase of the number of attack rounds, and after the 1st round of attack, the maximum connectivity of the physical network decreases from 1 to about 0.68, while the complex student network and information network decrease from 1 to about 0.98. After the 7th round of attack, the maximum connectivity of the physical network decreases to 0.09, at this time, there are only a few isolated nodes and small connectivity slices in the physical network, and the physical network is basically disintegrated. After the 15th round of attack, the maximum connectivity of complex student network and information network drops to about 0.09, and the change of maximum connectivity of complex student network and information network tends to flatten out in the subsequent attacks, and the complex student network, information network, and physical network basically disintegrate. Overall the information network’s maximum connectivity decreasing trend is similar to that of the complex student network, and its ability to resist attacks is similar, reflecting the dominant position of the information network in the student network.

Figure 3.

Variation of max connectivity under betweenness centrality strategy attacks

Fig. 4 reflects the variation of the number of connectivity slices under the degree centrality strategy attack. Under the degree centrality descending attack, the number of connectivity slices of complex network, information network and physical network shows a trend of increasing and then decreasing with the increase in the number of attack rounds, and as the number of attack rounds increases, the complex student network, the information network, and the physical network generates many isolated nodes and connectivity slices, which leads to the increase in the number of connectivity slices. The physical network only has isolated nodes and small connected slices left in the physical network after the 6th round of attack, and the physical network is basically disintegrated, while the maximum number of connected slices of the information network is more similar to the trend of the complex network changes, showing the dominant position occupied by the information network.

Figure 4.

Variation of the number of connected components under attacks

Fig. 5 depicts the variation of maximum connectivity slice size under degree centrality strategy attack. Under degree centrality descending attack, the maximum connectivity slice size of complex, information and physical networks is negatively correlated with the number of attack rounds, which is due to the fact that as the number of attack rounds increases, the maximum connectivity slice disintegrates to produce isolated nodes and small connectivity slices, and the number of nodes in the maximum connectivity slice is decreasing until the end of the attack when the complex student network disintegrates, and the physical and information networks disintegrate with it. Overall, the size of the maximum connectivity slice of the information network is similar to the decreasing trend of the complex student network, and its ability to resist attacks is similar, again indicating the dominance of the information network.

Figure 5.

Variation of size of the largest connected components under attacks

Based on the above complex student network attack pattern, it can be shown that the change trend of network evaluation indexes of complex student network and information network is more similar to that of physical network than that of physical network based on the degree centrality strategy attack of complex network attack pattern, and the results show that information network dominates in the composition of complex student network. The structure and function of the complex student network are still maintained to a certain extent after the information network is disintegrated, and the structure and function of the complex student network decreases to a smaller extent after the physical network is disintegrated, which indicates that the complex student network is more capable of resisting the attack than the single information network and physical network.

Setting the attack content, attacking the internal network information, comparing the protection ability of the two network information security protection systems, the attack content in the experiment is shown in Table 1.

Each attack content contains 220 network information, compare this paper protection system and the traditional system in these five kinds of attacks, the network information in the system was successfully attacked information. The more successful information is attacked, the worse the security protection function of the system is, and vice versa, the better the protection ability of the system is. The experimental process lasts for a total of 10h, and the experimental data collection is carried out every 2h, and the network defense strength is gradually enhanced.

The comparison results of this designed network information security protection system based on attack and defense game model with the traditional system are shown in Table 2. As can be seen from Table 2, the number of attacked information of the system designed in this paper is 198, 106, 126, 182 and 168 less than the traditional system in account stealing attack, planting Trojan horse attack, rebound attack, web page attack and data listening attack, respectively.

The simulation and comparison experiment software is further used to simulate the network information system environment, and add strong attack strategy, weak attack strategy, strong defense strategy and weak defense strategy in this environment to defend the network information by using the method of this paper and the traditional defense method respectively, and set them as the experimental group and the control group respectively. In the experimental process, the simulation step size is 0.05 to simulate the strategy evolution process of both attack and defense under different conditions, and record the experimental results, which are plotted into the experimental results comparison diagram shown in Fig. 6.

Figure 6.

The comparison of experimental results

From the experimental results in Figure 6, it can be seen that in 50 samples, the experimental group’s effective defense of network information is significantly higher than that of the control group, and at the same time, it can be seen from the magnitude of the change of the two curves, the fluctuation of the experimental group is basically the same as the amount of network information that is attacked by the attacking strategy added in the experimental process of this paper, which indicates that the experimental group can choose different strengths of the defense strategy based on the attacking party’s attacking strategy strength, the Targeting is stronger. Therefore, through comparative experiments, it is proved that the security protection protection method proposed in this paper can effectively increase the amount of effective defense information, higher defense rate, and more in line with the needs of network information security transmission.

Comprehensive results of the above comparisons show that the network information security protection system designed in this paper has a stronger protection ability for students’ network information, which significantly reduces the number of attacks on information than the traditional system, and is more targeted in the face of different attack strategies. Therefore, it can be proved that the protection ability of the network information security protection system based on the attack and defense game model designed in this study is better than that of the traditional system, which enhances the security of students’ network information in colleges and universities.