Efficient and reliable virtual backbone network construction method for heterogeneous wireless sensor network
By constructing the minimum strong connectivity dominant suction collection with (k, m)-α times routing constraints in heterogeneous wireless sensor networks, the problem of insufficient routing path length and backbone fault tolerance is solved, and network energy consumption saving and robustness are achieved promote.
Patent Information
- Application Number
- CN202410674415.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-28
- Publication Date
- 2024-11-12
AI Technical Summary
In heterogeneous wireless sensor networks, it is difficult for the prior art to optimize the routing path length and backbone fault tolerance at the same time, resulting in increased network transmission delay and energy consumption, and insufficient network robustness.
A minimum strong connectivity dominant absorption collection (α-MOC-SCDAS) method is proposed to construct (k, m)-α-fold routing constraints in heterogeneous wireless sensor networks, shorten the routing path length through routing constraints, and build m- Dominant absorbing collection improves the fault tolerance of the virtual backbone network and uses a three-stage algorithm to generate target sets.
It effectively reduces the length of routing paths, reduces redundant paths, saves network energy consumption, significantly improves the robustness and stability of the network, and extends the network life cycle.
Smart Images

Figure FT_1 
Figure FT_2 
Figure FT_3
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of network communication, and specifically proposes a virtual backbone network construction method that simultaneously considers routing path length and backbone fault tolerance in a heterogeneous wireless sensor network. The method effectively shortens the routing path length between network nodes through routing constraints, and improves the fault tolerance of the virtual backbone network by constructing an m-dominated absorption set. When the virtual backbone network constructed by the method is used for communication routing, it can effectively reduce the redundant paths of message transmission between sensor nodes in the heterogeneous wireless sensor network, reduce transmission delay and network energy consumption, and can also significantly enhance the robustness and stability of the entire network when performing communication work. Background Art
[0002] Different from traditional wired networks, wireless sensor networks (WSNs) have many characteristics, such as self-organization, multi-hop routing, dynamic topology, and node data fusion capabilities. They have been widely used in environmental monitoring, health applications, disaster relief, military operations, traffic control, mobile computing and other application fields where it is difficult to install base stations in the network. Since there is no fixed infrastructure, in order to effectively solve the high variability of network topology and reduce unnecessary energy consumption caused by redundant routing to avoid information conflicts and broadcast storms, Ephremides et al. [1] noticed that previous wireless network routing algorithms did not try to use a structure similar to the physical backbone infrastructure of packet cellular networks, so they introduced the concept of virtual backbone network. The virtual backbone network is a subset of wireless network nodes. The information interaction between any nodes in the network can be converted into homologous operations on the virtual backbone network, which can greatly reduce the message forwarding process, shorten the routing path search time, reduce the routing table and simplify routing maintenance to a certain extent. In wireless sensor networks, there are many methods to build a virtual backbone network, among which the construction method based on connected dominating set is a common and competitive method.
[0003] In actual situations, sensors in WSNs will have different transmission ranges due to the influence of power and transmission media, and the network formed is called a heterogeneous wireless sensor network. At this time, the network can be modeled as a directed graph with different node transmission radius. The construction of a virtual backbone network in a heterogeneous WSN can be transformed into the construction problem of a strongly connected dominant absorption set in graph theory. With the continuous development of practical applications, the performance requirements for virtual backbone networks have gradually increased. Building an efficient virtual backbone has become a research hotspot in the current academic community. The optimization goal is no longer limited to minimizing the size of the backbone. Related research has begun to focus on routing path length, fault tolerance, node energy efficiency, network life cycle, etc.
[0004] In WSNs with virtual backbones, the Average Backbone Path Length (ABRL) [2] is usually used to measure the length of the transmission path between nodes in the network, which indirectly reflects the cost or energy consumption of message transmission in the network. However, studies have found that the ABRL of networks using virtual backbones is often much larger than the original path length, which greatly increases communication delay and routing overhead. In order to make the routing path between any two points as short as possible in VBN-based communication routing, researchers have proposed a virtual backbone network with routing constraints. By adding some backbone nodes, the routing path length between any pair of nodes in the network is guaranteed not to exceed a constant multiple (α) of the shortest path length in the original network, thereby reducing energy consumption and transmission delay. Studies have found that this method can achieve a trade-off between routing constraints and backbone network scale to a certain extent.
[0005] Since sensor networks are often deployed in harsh or hostile environments, channel bandwidths often change and are sometimes connected and sometimes disconnected, and sensor failures caused by energy exhaustion and damage frequently occur. The theoretically constructed energy-saving topology is often not applicable in practical applications. In practical applications, a reasonable topology should retain some redundancy to deal with possible channel and node failures, that is, it should have a certain degree of fault tolerance. Therefore, the research on fault-tolerant virtual backbones has attracted much attention. The fault-tolerant virtual backbone retains some redundancy on the basis of the original backbone structure to form a target virtual backbone with reasonable coverage and connectivity to deal with the occurrence of failure behaviors, improve the robustness, robustness and availability of the network, and provide a more reliable underlying topology for upper-layer applications. In graph theory, this problem can be solved by constructing a (k, m)-CDS, that is, it is required that all nodes outside the VB are dominated by at least m points in the VB, and there are at least k non-intersecting paths between any backbone node pairs. At this time, even if min{k-1, m-1} nodes fail, the virtual backbone can still work normally.
[0006] Based on the demand for optimizing the routing path length in heterogeneous networks and improving the robustness of virtual backbones, the present invention proposes a virtual backbone network construction problem that considers both routing path length and backbone fault tolerance in heterogeneous wireless sensor networks, and transforms the problem into a (k, m)-α times routing-constrained minimum strongly connected dominating and absorbing set (α-MOC-SCDAS) problem. At the same time, an approximate construction algorithm is given, which uses the routing constraint method to reduce the path length of communication between nodes, and improves the fault tolerance of the virtual backbone network by constructing a (k, m)-SCDAS. After theoretical analysis and simulation experimental verification, the target set finally generated by the algorithm can effectively reduce the routing path length, reduce redundant paths, save network energy consumption, and also improve the backbone fault tolerance to a certain extent, cope with node failure, enhance the robustness of the network, and extend the network life cycle. Summary of the invention
[0007] [Purpose of the invention]: Based on the need to optimize the routing path length in heterogeneous networks and to improve the robustness of the virtual backbone, the present invention proposes a method for constructing a virtual backbone network that simultaneously considers the routing path length and backbone fault tolerance in heterogeneous wireless sensor networks, which is modeled as constructing a (k, m)-α times routing-constrained minimum strongly connected dominating and absorbing set (α-MOC-SCDAS). At the same time, an approximate construction algorithm is given, which uses the routing constraint method to reduce the path length of communication between nodes, and improves the fault tolerance of the virtual backbone network by constructing a (k, m)-SCDAS. After theoretical analysis and simulation experiments, the target set finally generated by the present invention can effectively reduce the routing path length, reduce redundant paths, save network energy consumption, and also improve the backbone fault tolerance to a certain extent, cope with node failures, enhance the robustness of the network, and extend the network life cycle.
[0008] [Technical solution]: The technical solution of the present invention is:
[0009] 1. Network model
[0010] In a wireless sensor network, the transmission range of all sensor nodes is uncertain. For any node pair u and v, if v is within the transmission range of u, it is assumed that the message from u can be received. The above network can usually be described by a directed graph G = (V, E), where V represents the sensor node and E represents whether the sensor nodes can communicate directly. For any node u, v∈V, if (u, v)∈E, it means that v can receive the message from u. If (u, v)∈E and (v, u)∈E, it means that u and v can transmit messages to each other.
[0011] 2. Problem Definition
[0012] Given a directed graph G = (V, E) and a set of nodes For any node u∈V, either u∈S, or there exist v, w∈S, such that (u, v)∈E and (w, u)∈E, then the set S is called a dominated absorbing set of G. Let G[S] represent the induced subgraph of G. If for any pair of nodes u, v∈S, there exists a directed path from u to v in G[S], then G[S] is called strongly connected. In order to further solve the problem of network topology instability caused by node failure and other reasons in sensor networks, we use k-connected m-dominated sets in graph theory to solve the fault tolerance problem of the network. The survey found that some studies have used α-fold routing constraints to save routing overhead between sensor nodes, thereby reflecting the high efficiency of the virtual backbone network. To this end, we introduce the following definition:
[0013] Definition 1. Strongly Connected Dominated Absorbing Set (SCDAS). Consider a directed graph G = (V, E), G[S] is a subset of G induced subgraph of G. S is a strongly connected dominated absorbing set of G if and only if:
[0014] 1) G[S] is strongly connected.
[0015] 2) S is a dominating absorbing set DAS of G.
[0016] For a CDS S, for any node pair u, v∈V, let p(u,v) denote the shortest path from u to v in G, d(u,v) denotes the length of the shortest path p(u,v), where d(u,v) = |p(u,v)|-1; let p s (u, v) represents the shortest path from u to v in G[S], d S (u, v) represents the shortest path p S The length of (u, v), where d S (u, v) = |p S (u, v)|-1.
[0017] Definition 2. Minimum Routing Constrained Strongly Connected Dominant Absorbing Set (MOC-SCDAS). MOC-SCDAS is to find a minimum set of points in a directed graph G = (V, E) This set D has the following properties:
[0018] 1) w∈D, let (u, v)∈E and (w, u)∈E.
[0019] 2) The induced subgraph G[D] is strongly connected.
[0020] 3) If d(u,v)>1, then where p i (u, v) = {u, w1, w2, ..., w k , v} represents the i-th shortest path between points u and v, P i (u, v) represents the set of all shortest paths between u and v.
[0021] The present invention considers the α-MOC-SCDAS problem in a directed disk graph. Obviously, α-MOC-SCDAS is a SCDAS under special constraints. Therefore, the α-MOC-SCDAS problem can be formally defined as:
[0022] Definition 3. α-MOC-SCDAS. Consider a directed graph G = (V, E), find a subset with the smallest size S is a minimal strongly connected dominating absorbing set with an α-fold routing constraint if and only if:
[0023] 1) S is an SCDAS.
[0024] 2) d S (u, v) ≤ αd (u, v), where d S (u, v) and d(u, v) are p S The lengths of (u, v) and p(u, v).
[0025] Definition 4. k-connected m-dominated set. A connected dominating set D in a given graph G = (V, E) is k-connected m-dominated if and only if any subset of size at most k-1 Make G D-D’ is still connected, and for any node u∈V, either u∈D or u has at least m neighbors in D.
[0026] Definition 5. (1, m)-αMOC-SCDAS. Given a directed disk graph G = (V, E), a subset S is called an α-routing constrained strongly connected dominating absorbing set of G based on an m-dominating absorbing set with fault tolerance. Then S must satisfy the following conditions:
[0027] 1) S is a dominating absorbing set of G.
[0028] 2) The subgraph G[S] derived from S is strongly connected, that is, there is at least one path between any pair of nodes in G[S], so that the nodes can transmit information to each other.
[0029] 3) For a given constant α≥1, any node pair u and v satisfies d S (u, v)≤αd(u, v).
[0030] 4) All non-backbone nodes in G are dominated and absorbed by at least m backbone nodes.
[0031] 3. Strategy and algorithm design
[0032] First, in order to solve the problem of too long paths when routing through virtual backbones, we use routing constraints to reduce redundant paths. The virtual backbone construction problem combined with routing constraints is solved in graph theory by selecting α-MOC-CDS. Existing research has confirmed that finding α-MOC-CDS is a non-deterministic polynomial (NP) hard problem. Some literatures solve this problem by proposing approximate algorithms for finding MOC-CDS. The following provides theoretical support for this method through some lemmas.
[0033] Lemma 1. Let G be a connected graph and S be a connected dominating set of G. If for any pair of nodes u and v that satisfy the condition d(u, v) = 2, d S (u, v)-1≤α. Then, for any pair of different nodes u and v, there are d S (u, v)-1≤α(d(u, v)-1).
[0034] Lemma 2. Let G be a connected graph and S be a connected dominating set of G. For any pair of distinct nodes u and v with d(u, v) = 2 and ds(u, v)-1 ≤ α, then for any pair of distinct nodes u and v, there is d S (u, v)-1≤αd(u, v).
[0035] Lemma 3. Set C is a maximal independent set of graph G, and S is a connected dominating set containing C if for every pair of nodes u, v∈C such that d(u, v)≤4, there is dS (u, v)≤4, then d S (u, v)≤5d(u, v).
[0036] Lemma 4. Given that set C is a minimal dominating set of graph G, and S is a strongly connected bidirectional dominating set containing C, for any pair of nodes u, v∈C, if d(u, v)≤3 is satisfied, then all the points on its shortest path are added to set S. Then the induced subgraph G[S] is strongly connected, and for any pair of nodes u, v in the graph, d S (u, v)≤5d(u, v).
[0037] According to Lemmas 1 to 4, we introduce the construction strategy of GOC-CDS as follows:
[0038] GOC.-CDS construction strategy: In a connected graph G, after finding the dominating set C of G, put all the nodes in C into S. If any pair of nodes u, v∈S satisfies d(u, v)≤3, add all the nodes of the shortest path between u and v into the set S, and you can get d S (u, v)≤5d(u, v).
[0039] Du et al.
[13] theoretically analyzed and proved Lemmas 1 to 3, and Putwattana et al.
[15] theoretically analyzed the correctness of Lemma 4. By constructing the GOC-CDS strategy, a connected dominating set with guaranteed routing costs within a certain range can be obtained.
[0040] In order to improve the fault tolerance of the virtual backbone network, we introduce the m-dominated absorption set to construct αMOC-SCDAS. The specific strategy is as follows:
[0041] αMOC-SCDAS construction strategy: Based on the GOC-CDS construction strategy, the minimal dominating absorbing set I2, I3, …, I m , and added to the GOC-CDS set. After m-1 iterations, the m-dominated αMOC-SCDAS set can be obtained, ensuring that any ordinary node is dominated and absorbed by at least m backbone nodes.
[0042] Based on the GOC-CDS construction strategy and the αMOC-SCDAS construction strategy, the present invention designs a three-stage (1, m)-αMOC-SCDAS construction algorithm:
[0043] 1) The first stage: construct the minimal dominant absorption set C as the initial backbone node set;
[0044] 2) The second stage: Generate a strongly connected dominated absorption set S based on the GOC-CDS construction strategy.
[0045] 3) The third stage: Generate the m-dominated set αMOC-SCDAS based on the αMOC-SCDAS construction strategy.
[0046] 4. Detailed algorithm design and theoretical analysis
[0047] The present invention first designs an algorithm for constructing a minimal dominating absorption set of a graph. The algorithm draws on the idea of the three-color marking algorithm, in which all nodes are marked white in the initial state, the backbone nodes selected during the algorithm execution are marked black, and the neighbor nodes of the backbone nodes are marked gray. The specific algorithm process is described in Algorithm 1. The algorithm uses a greedy strategy to greedily search for the node with the largest transmission range in the node set as the backbone node, and finally forms a minimal dominating absorption set. For a directed graph G, the directions of all directed edges are reversed to obtain a directed graph G′ with the opposite direction of the edge set, as shown in the first line of Algorithm 1. In the first stage, the greedy strategy is used twice to execute the algorithm process in graphs G and G′ respectively, generating a minimal dominating absorption set C of the original directed graph G.
[0048]
[0049] In the second phase of the (1, m)-αMOC-SCDAS algorithm, a connected node set needs to be constructed to connect the minimal dominating absorption set in the first phase of the algorithm. In this phase, the relevant strategies of routing constraints are executed. The specific description of the second phase of the algorithm is shown in Algorithm 2:
[0050]
[0051] Algorithm 2 generates a strongly connected dominating absorption set based on the αMOC-SCDAS construction strategy. Create an initial node set D, and for any pair of nodes u and v in the minimal dominating absorption set that satisfy d(u, v)≤3, find the shortest path of the node pair in graph G, and add all the intermediate nodes on the path to set D. The final set D is a connected node set. Then, the union of the minimal dominating absorption set C constructed in the first stage and the connected node set D constitutes a strongly connected dominating absorption set S of graph G.
[0052] In the third stage, the dominance / absorption of the current SCDAS is improved through iteration to achieve m-dominance absorption. Use the FOR loop for m-1 iterations, and each iteration needs to call Algorithm 1 to generate the minimum dominance absorption set I of the current graph. i , where i = 2, 3, ..., m. Finally, we take S = S∪I2∪I3∪...∪I m This constitutes a (1, m)-α times routing-constrained strongly connected dominating absorption set of graph G. The detailed description of the third stage of the algorithm is shown in Algorithm 3:
[0053]
[0054] By improving the dominance absorption of the target node set through iteration, the final selected node set still satisfies the routing constraints described in Lemma 4. The specific theoretical analysis process is as shown in Theorem 1.
[0055] Theorem 1. Set c is the minimal dominating absorption set constructed in the first phase of the algorithm, and set S is the strongly connected m-dominated absorption set with α-fold routing constraints constructed in the third phase of the algorithm. Either it belongs to the set S, or it is dominated by at least m nodes in the set S: and All meet d S (u, v)≤5d(u, v).
[0056] prove:
[0057] We first prove that the set constructed in the third phase of the algorithm is still strongly connected. After the third phase of the algorithm is executed, since the strongly connected (m-1)-dominated absorbing set before iteration m-1 must be strongly connected, the remaining nodes are all dominated and absorbed at this time, so adding any new backbone node can still guarantee strong connectivity. Back to before the first iteration, the first and second phases of the algorithm have constructed a strongly connected dominated absorbing set (1, 1)-SCDAS, so the remaining nodes must have dominating neighbor nodes and absorbing neighbor nodes. After executing the first round of iterations, some backbone nodes are added to satisfy 2-dominated absorption at this time. These newly added nodes were previously dominated and absorbed nodes, so the current (1, 2)-SCDAS can still guarantee strong connectivity.
[0058] Then we prove that the virtual backbone constructed after the third stage of the algorithm satisfies the m-dominance property. This is obvious because after each iteration of the dominating set construction algorithm, the ordinary nodes in the entire network will increase at least one backbone node that dominates and absorbs it. After m-1 iterations, the remaining non-backbone nodes in the network will be dominated and absorbed by at least m backbone nodes.
[0059] We further prove that the set finally constructed by the algorithm still has the α-fold routing constraint. From Lemma 4, we can see that the strongly connected dominant absorption set (1, 1)-αMOC-SCDAS constructed in the second stage of the algorithm satisfies d S (u, v)≤5d(u, v). Since the newly added backbone nodes are dominated and absorbed no matter how many times the algorithm is iterated, the SCDAS S constructed in the first and second stages of the algorithm meets the α-fold routing constraint, so any node pair can guarantee d S(u, v)≤αd(u, v), so the iterative method to construct the multi-dominated absorption set only adds some backbone nodes to improve the dominance of the virtual backbone and does not affect the routing constraints.
[0060] 5. Algorithm approximation ratio analysis
[0061] Lemma 5. For any two nodes u and v belonging to C constructed by Algorithm 1, d(u, v) ≥ r min .
[0062] Proof. Without loss of generality, assume that r u ≥r v ≥r min u is added to C earlier than v in the algorithm. Assume d(u, v) ≤ r min , then v∈N + (u). Therefore, v cannot be added to the dominating set C, which contradicts the previous assumption.
[0063] Next, we use the idea of circle packing to fill a circle with a radius of r. min A virtual disk of / 2 is introduced into the directed graph G(V, E), with the center of the circle being the sensor node and the radius r min / 2 is used as the transmission range. Since there is no transmission radius r in G(V, E) min / 2 of the real nodes, so it is virtual. However, this virtual disk plays an important role in the proof of the theorem of the approximation ratio of the calculation algorithm below.
[0064] Lemma 6. If C is a minimal dominating absorbing set, then the upper bound of the node set C generated in Algorithm 1 is (2ρ+1) 2 opt, where ρ = r max / r min , opt is the optimal solution of the Minimum Strongly Connected Dominated Absorbing Set Problem (MSCDAS).
[0065] Proof. Lemma 5 shows that in C, r min All disks with radius / 2 and nodes as the center do not intersect each other. Therefore, considering the maximum coverage, all nodes in the optimal DAS are r max +r min / 2 is the transmission radius, then within the area covered by the optimal DAS, the size of any DAS is r min / 2 is the upper bound. It can be calculated that the maximum number of virtual disks filled within the node coverage range in DAS is π(r max +r min / 2) 2 / π(r min / 2) 2 =(2ρ+1) 2 , where ρ = r max / r min That is, for each node v∈C, it can dominate at most (2ρ+1) 2 independent nodes, so the upper bound of C is (2ρ+1) 2 opt.
[0066] Lemma 7. If we construct a strongly connected dominating absorbing set S by connecting u and v∈C, so that the intermediate nodes on the shortest path of the node pair u and v with d(u, v)≤3 are added to the set S, that is, they are connected through a path with a length of at most 3 hops. Then the size of S is at most 2(6ρ+1) 2 (2ρ+1) 2 opt.
[0067] Proof. Consider any node u∈C. Since the distance between the dominant nodes within 3 hops is [r min , 3r max ], so the upper bound of the number of nodes v∈C within 3 hops from node u is π(3r max +r min / 2) 2 / π(r min / 2) 2 =(6ρ+1) 2 We connect u and v through paths of length at most 3, with two intermediate nodes in each path. Therefore, each u connects to v, making d(u, v) ≤ 3, using at most 2(6ρ+1) 2 nodes. Then the final scale of S is |S|≤2(6ρ+1) 2 |C|≤2(6ρ+1) 2 (2ρ+1) 2 opt is the upper bound.
[0068] Theorem 2. The upper bound of the strongly connected dominating absorption set S with (1, m)-α times routing constraints finally constructed by the (1, m)-αMOC-SCDAS algorithm is (72ρ 2 +24ρ+m+1)(2ρ+1) 2 opt, where ρ = r max / r min , m is the dominance / absorption degree, opt is the optimal solution of the Minimum Strongly Connected Dominated Absorbing Set Problem (MSCDAS).
[0069] Proof. According to the basic idea of the third stage of the algorithm, the (1, m)-αMOC-SCDAS finally constructed is obtained by adding the (1, 1)-αMOC-SCDAS generated in the second stage to the m-1 dominating absorption sets obtained by the iteration method. From Lemma 7, it is known that the upper bound of (1, 1)-αMOC-SCDAS is approximately 2(6ρ+1) 2 (2ρ+1) 2 opt, combined with Lemma 6 and Lemma 7, we can calculate that the upper bound of (1, m)-αMOC-SCDAS is 2(6ρ+1) 2 (2ρ+1) 2 opt+(m-1)(2ρ+1) 2 opt=(72ρ 2 +24ρ+m+1)(2ρ+1) 2 opt.
[0070] [Beneficial Effects]: The significance of the present invention lies in that a method for constructing a virtual backbone network that considers both routing path length and backbone fault tolerance in a heterogeneous wireless sensor network is proposed. When the virtual backbone network constructed using this method is used for communication routing, it can not only significantly control the scope of message forwarding and save network energy consumption, but also effectively reduce the redundant paths for message transmission between sensor nodes in a heterogeneous wireless sensor network, reduce transmission delay, and significantly enhance the robustness and stability of the entire network during communication work, thereby extending the network life cycle. BRIEF DESCRIPTION OF THE DRAWINGS
[0071] Figure 1 r min ABPL of SCDAS generated by each algorithm in networks with different node numbers when =15
[0072] Figure 2 r min ABPL of SCDAS generated by each algorithm in networks with different node numbers when =25
[0073] Figure 3 r min =15 The size of SCDAS generated by each algorithm in networks with different node numbers
[0074] Figure 4 r min =25 The size of SCDAS generated by each algorithm in networks with different node numbers
[0075] Figure 5 The success rate of maintaining backbone performance by generating the remaining backbone using different algorithms after randomly destroying some network nodes DETAILED DESCRIPTION
[0076] The present invention is described in detail below with reference to the accompanying drawings and specific examples. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and do not limit the present invention.
[0077] We conducted simulation experiments on the three-stage algorithm (1, m)-αMOC-SCDAS proposed in the present invention for constructing a virtual backbone network that takes into account fault tolerance and routing constraints, and selected several classic and effective virtual backbone network construction algorithms for performance comparison. The experiments proved that the algorithm of the present invention has better results.
[0078] 1. Build an experimental platform
[0079] The experimental platform was built on a personal computer. The computer on which the experiment was based had a 13th Gen Intel(R) Core(TM) i5-13600KF processor with a main frequency of 3.50GHz, 32.0GB of RAM (31.8GB available), and a 64-bit Windows 10 Professional operating system. Matlab was used as the programming IDE.
[0080] 2. Network parameter settings
[0081] The simulation experiment uses Dr. Salama's network topology random generation algorithm to generate a random network topology. First, the random network generation algorithm is modified as needed, and then a random network topology that meets the needs of this experiment is generated.
[0082] In the simulation experiment, the parameters of the heterogeneous wireless sensor network composed of different numbers of sensors with different transmission ranges are set as follows. The sensors in the network are simulated by a group of nodes with different transmission radius randomly deployed on the Euclidean plane with a boundary of 100×100. The total number of nodes in the network varies between 100, 150, 200, 250, 300, 350 and 400. Each node has a fixed transmission range of [r min , r max ], the minimum transmission radius is set to 15 and 25, where is the transmission radius ratio, and the variation range is 1.25, 1.50, and 1.75. In each simulated network, if and only if d(u, v)≤r u , there is an edge from u to v, and then the directed disk graph corresponding to the simulated network is constructed according to the method in Section 2, and it is determined whether it is strongly connected. If it is a strongly connected graph, it is retained as a candidate network graph; otherwise, the graph is discarded and the network and graph are regenerated. To avoid contingency, for each set of parameter settings (for example, the number of nodes is 100, r min=15, ρ=1.25 to construct a set of parameter settings) to randomly generate 20 simulated networks. Each algorithm conducts 10 experiments in each simulated network and takes the average value as the operation result of the algorithm under this set of parameter settings. Among them, for the comparison of the fault tolerance of constructing virtual backbones by different algorithms, it is necessary to randomly destroy 10%, 20% and 30% of the nodes in the simulated network under each set of parameter settings. Finally, the size of the virtual backbone, the average backbone path length ABPL and the success rate of the remaining backbone maintaining performance after some nodes fail are used to compare the simulation results.
[0083] In the experiment, the (1, m)-αMOC-SCDAS algorithm was used to compare and analyze with the SCDAS-BD-C2 algorithm, the SCDAS-BFS algorithm and the ShortestPath-SCDAS algorithm to prove its effectiveness. For the SCDAS generated by various algorithms, three evaluation indicators were set: (1) average backbone path length ABPL (average backbone path length); (2) SCDAS scale; (3) SCDAS fault tolerance. Obviously, for the strongly connected dominated absorbing set problem that considers both routing constraints and backbone fault tolerance, first of all, it is expected that the number of nodes in the node set solved by the algorithm is as small as possible, and secondly, it is expected that the path length of the communication routing only through the solution set solved by the algorithm is as short as possible. At the same time, it is also expected that the backbone network generated by the algorithm has high fault tolerance.
[0084] 3. Simulation Experiment Results and Analysis
[0085] The first evaluation metric is the average backbone path length ABPL of the virtual backbones generated by different algorithms. Figure 1 The minimum transmission radius r is shown min When ρ is 15, the performance of (1, m)-αMOC-SCDAS algorithm, SCDAS-BD-C2 algorithm, SCDAS-BFS algorithm and ShortestPath-SCDAS algorithm in terms of average backbone path length ABPL, where ρ is 1.25, 1.50 and 1.75 respectively. Similarly, Figure 2 For r min The experimental results are as follows: Figure 1 and Figure 2It can be seen that when the node transmission radius is generally small, as the number of sensor nodes increases from 100 to 400, the average backbone path length corresponding to the four algorithms decreases slightly. This is because when the number of nodes is small, the shortest path between nodes in the graph is likely to be a very long curve, requiring more intermediate nodes to form a path, which means that the shortest path length value is likely to be large; if the number of nodes reaches a certain number, that is, when the network density increases to a certain extent, the shortest path between nodes can be close to a straight line. At this time, the shortest path length value is the smallest: However, when the number of nodes exceeds a certain range, the shortest path length value between nodes will no longer change with the increase in the number of nodes. In addition, from Figure 1 and Figure 2 It can be seen that the ABPL of the backbone generated by the (1, m)-αMOC-SCDAS algorithm is slightly larger than that of the ShortestPath-SCDAS algorithm using the shortest path of the original graph, but smaller than the two algorithms that do not consider routing constraints.
[0086] The second evaluation metric is the size of the virtual backbone generated by different algorithms. Figure 3 The minimum transmission radius r is shown min When ρ is 25, the performance of the four algorithms in terms of the size of the constructed SCDAS, where ρ is 1.25, 1.50 and 1.75 respectively. Similarly, Figure 4 For r min The experimental results are as follows: Figure 3 and Figure 4 It can be seen that as the total number of sensor nodes increases from 100 to 400, the average size of the SCDAS generated by all algorithms increases slightly. This is because as the total number of nodes increases, more backbone nodes need to be added to build the SCDAS during the execution of any algorithm. The scale of the SCDAS generated by the (1, m)-αMOC-SCDAS algorithm is larger than that of the SCDAS-BD algorithm and the SCDAS-BFS-C2 algorithm, but its average backbone path length ABPL is smaller than both algorithms, and is significantly smaller than the ShortestPath-SCDAS algorithm that directly uses the shortest path to connect backbone nodes. More importantly, the scale of the virtual backbone constructed by the algorithm designed by the present invention will not exceed the upper bound value proved by Theorem 1.
[0087] The third evaluation indicator is the fault tolerance performance of virtual backbones generated by different algorithms. The node transmission radius is set to a random interval of [15, 30], and then 10%, 20%, and 30% of the nodes are randomly destroyed for each simulated network, and the probability that the virtual backbones generated by different algorithms can ensure connectivity is counted. Figure 5The fault tolerance performance of the virtual backbone constructed by the (1,3)-αMOC-SCDAS algorithm, (1,2)-αMOC-SCDAS algorithm, (1,1)-αMOC-SCDAS algorithm, SCDAS-BD-C2 algorithm and SCDAS-BFS algorithm is shown when the number of randomly destroyed nodes is 10%, 20% and 30% respectively. It is intuitively reflected in the success rate of the remaining backbone to maintain the backbone performance after the network nodes are randomly destroyed. There are two requirements for setting the backbone performance detection: (1) Whether the remaining backbone network is strongly connected; (2) The non-backbone node is dominated by at least one backbone node and absorbed by one backbone node at the same time. Figure 5 It can be seen that in the random input graph with a total number of sensor nodes between 100 and 400, as the number of damaged nodes increases from 10% to 30%, the success rate of the remaining parts of the SCDAS generated by various algorithms to maintain the backbone performance has decreased. This trend is obvious, because the fault tolerance of the virtual backbone constructed by the algorithm is limited. As the number of randomly damaged nodes increases, the number of original backbone nodes also decreases, and then there will be a situation where non-backbone nodes lose the control of backbone nodes. For the construction algorithm that does not consider the fault tolerance of the backbone, it can only ensure that the non-backbone nodes are controlled or absorbed by at least one backbone node. When a backbone node fails due to various factors, the non-backbone nodes of its only backbone neighbor lose communication with the network. At this time, the current backbone cannot maintain the virtual backbone characteristics, and the remaining backbone network cannot work normally. Therefore, it is very necessary to consider the fault-tolerant virtual backbone network construction strategy. The (1, m)-αMOC-SCDAS algorithm designed by the present invention optimizes the fault tolerance of backbone construction.
[0088] For the (1, m)-αMOC-SCDAS algorithm, Figure 5 The experimental results when m is 1, 2 or 3 are shown in Figure 1, and compared with the effects of the SCDAS-BD-C2 algorithm and the SCDAS-BFS algorithm. In this figure, m is the domination absorption number, which is related to the fault tolerance between backbone nodes and non-backbone nodes. That is, the non-backbone nodes in the figure are dominated by at least m backbone nodes and absorbed by at least m backbone nodes. When the number of randomly destroyed nodes in the network is 10%, 20% and 30% respectively, the success rate of the remaining backbone in maintaining backbone performance is significantly higher than that of the other two algorithms. As m increases, the success rate of backbone network performance maintenance is significantly improved, because the increase in the domination absorption number m means that any non-backbone node is dominated and absorbed by at least one backbone node. If one backbone neighbor node fails, the other m-1 backbone neighbor nodes can still maintain the backbone performance, allowing the entire virtual backbone to work normally.
[0089] 5. Experimental Conclusion
[0090] In a heterogeneous wireless sensor network formed by different transmission radius of sensor nodes, the (1, m)-αMOC-SCDAS algorithm proposed in the present invention can effectively solve the problem of constructing a virtual backbone network considering fault tolerance and α-fold routing constraints, and construct a virtual backbone network that simultaneously improves network robustness and optimizes node routing path length, so that the node routing path length when communicating only through the backbone network is significantly reduced, and the fault tolerance of the virtual backbone can be improved to a certain extent. Compared with several comparative algorithms, the SCDAS generated by the (1, m)-αMOC-SCDAS algorithm has a smaller backbone scale, a shorter average backbone path length, and a higher backbone fault tolerance. Overall, its comprehensive performance is the best.
[0091] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A method for constructing an efficient and reliable virtual backbone network in heterogeneous wireless sensor networks Aiming at heterogeneous wireless sensor networks, the present invention proposes for the first time a virtual backbone network construction problem that takes both routing path length and network fault tolerance into consideration, and proposes a strategy for designing an approximate backbone network construction algorithm to solve the problem. The network formed by the difference in transmission range caused by the influence of power and transmission medium is called heterogeneous wireless sensor network. The construction of virtual backbone network in heterogeneous wireless sensor network can be transformed into the construction problem of strongly connected dominating and absorbing set (SCDAS) in graph theory. The average backbone path length (ABPL) is usually used to measure the length of the transmission path between nodes in the network, which indirectly reflects the cost or energy consumption of message transmission in the network. However, studies have found that the average backbone path length of the network using virtual backbone network is often much longer than the original path length, which greatly increases the communication delay and routing overhead. In order to make the routing path between any two points as short as possible in the communication routing based on virtual backbone network, researchers have proposed a virtual backbone network with routing constraints. By adding some backbone nodes, it is ensured that the routing path length between any pair of nodes in the network does not exceed a constant multiple (α) of the shortest path length in the original network, thereby reducing energy consumption and transmission delay. Since sensor networks are often deployed in harsh or hostile environments, node failures frequently occur due to a variety of internal and external factors. Therefore, a reasonable backbone structure should retain some redundancy to deal with possible channel and node failures, that is, it should have a certain degree of fault tolerance and improve the robustness, robustness and availability of the network. Therefore, the research on fault-tolerant virtual backbones has attracted much attention. In graph theory, this problem can be solved by constructing a (k, m)-connected dominating set, that is, it is required that non-backbone nodes are dominated by at least m backbone nodes, and the connectivity of the derived subgraph of the backbone node set is at least k, that is, there are at least k non-intersecting paths between any backbone node pairs. Therefore, the research problem proposed by the present invention is to optimize both the routing path length and the backbone fault tolerance in heterogeneous wireless sensors.
2. An approximate algorithm for constructing a virtual backbone network for heterogeneous networks considering efficient transmission and backbone fault tolerance This paper proposes a virtual backbone network construction algorithm that considers both routing path length and backbone fault tolerance in heterogeneous wireless sensor networks, named (1, m)-αMOC-SCDAS. In this strategy, routing constraints are used to reduce the routing path length between nodes, and the fault tolerance of the virtual backbone network is improved by constructing an m-dominated absorption set. Finally, simulation experiments are conducted to verify its advantages and effectiveness, and the approximation ratio of the algorithm is calculated through theoretical analysis.
3. Approximation ratio of the target result generated by the (1, m)-αMOC-SCDAS algorithm of the present invention The upper bound of the strongly connected dominant absorption set S constructed by the (1, m)-αMOC-SCDAS algorithm of the present invention is (72ρ 2 +24ρ+m+1)(2ρ+1) 2 opt, where ρ is the ratio of node transmission radius and opt is the optimal solution to the minimum strongly connected dominated absorbing set problem.
4. Algorithm for constructing virtual backbone network with α-fold routing constraints based on (1, m)-connected dominating sets The (1, m)-αMOC-SCDAS algorithm proposed in this invention is mainly divided into three stages: the first stage, constructing a minimal dominating absorption set C as the initial backbone node set; the second stage, generating a strongly connected dominating absorption set S based on the GOC-CDS construction strategy; the third stage, generating an m-dominated set αMOC-SCDAS based on the αMOC-SCDAS construction strategy. The target node set of the algorithm is a special case of (k, m)-αMOC-SCDAS, where k=1. Firstly, αMOC-SCDAS is constructed, and an algorithm for constructing the minimal dominating / absorbing set of the graph is designed. This algorithm draws on the idea of the three-color marking algorithm and uses a greedy strategy to greedily search for nodes with the largest transmission range between nodes as backbone nodes, and finally forms a minimal dominating / absorbing set. In the second stage of the (1, m)-αMOC-SCDAS algorithm, a connected node set needs to be constructed to connect the minimal dominating absorption set of the first stage of the algorithm. In this stage, the relevant strategies of routing constraints are executed. In the third stage, the dominance / absorption of the current SCDAS is improved through iteration to achieve m-dominance absorption. Use the FOR loop for m-1 iterations, and each iteration needs to call Algorithm 1 to generate the minimum dominance absorption set I of the current graph. i , where i = 2, 3, ..., m. Finally, we take S = S∪I 2 ∪I 3 ∪…∪I m This constitutes a (1, m)-α times routing constrained strongly connected dominated absorption set of graph G.
