A sensor node based end-to-end path data collection optimization method

Abstract

The application discloses an optimization method for CP selection and path planning based on ACO, and comprises the following steps: in the initial stage, a data forwarding tree is constructed by using Prim algorithm; all nodes in the data forwarding tree can only send their data packets to their parent nodes with the base station as the root; when a node is selected as a CP, the edge between the CP and its parent node is broken, and the CP can directly upload the collected data packets to a mobile receiver; the optimization method selects a suitable sensor node as the CP, and plans a path for the mobile receiver to access all the CPs, the proposed strategy is based on ACO, and can be extended to be combined with other data collection methods, extended to be combined with other methods for solving the CP selection and path planning independently, and fully utilize the maximum path length to solve the problems in the network nodes, and the proposed end-to-end strategy can achieve better performance in the network life, total energy consumption and fairness index.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of sensor network data collection, and particularly relates to an end-to-end path data collection optimization method based on a sensor node. BACKGROUND

[0002] Wireless sensor networks (WSNs) are a distributed network for sensing the environment. Due to their flexibility, WSNs have been widely applied in smart homes, healthcare, environmental monitoring, disaster prediction, military fields, and many other applications. A WSN is composed of a set of sensor nodes deployed in a specific area. Each sensor has a limited battery capacity and a fixed size data buffer, which leads to energy efficiency problems. Therefore, some protocols for improving energy efficiency in WSNs have been proposed. These protocols are mainly divided into four categories: ratio module, data reduction, sleep / wake-up scheme, and battery charging 6. The ratio module considers four states for each sensor node: idle, sleep, transmission, and reception. Most of the time, the sensor is in the idle state, which leads to the sensor node being exhausted. The ratio module minimizes energy consumption by adjusting the optimization parameters. However, the ratio module has to manage four states, which introduces additional energy consumption. The next category is data reduction, which maximizes the reduction of the number of transmitted data packets to achieve energy efficiency and reduce traffic and latency. However, it inevitably loses the accuracy and accuracy of the original data. The sleep / wake-up scheme adopts a duty cycle mechanism to reduce the idle state and support sleep mode. However, the working cycle causes sleepy latency and wastes energy for redundant wake-up. Battery charging obtains energy from wind, solar, and other new energy sources. It considers nodes with uneven residual energy when designing the protocol.

[0003] Sensor nodes periodically generate sensing data and transmit the data to the base station through multi-hop forwarding. Due to the characteristics of multi-hop forwarding, nodes close to the base station will receive and transmit more data packets, thus consuming more energy. In this case, the nodes closest to the base station are easily the first to be exhausted. When those nodes die, the data of other nodes cannot reach the base station, and the network cannot work until the dead nodes are charged. The problem of uneven energy consumption will significantly shorten the lifetime of the network.

[0004] To alleviate the problem of uneven energy consumption, some research considers a new data collection strategy with a mobile sink, and we are currently focusing on WSNs in which sensor nodes generate data packets periodically. All data packets must be transmitted to the base station within a given delay. There is a set of collection points responsible for collecting and storing data packets generated by a certain set of sensor nodes. The mobile sink can move around the WSN to collect data from the CPs. The goal is to select a set of CPs and plan a tour of the mobile sink through all the CPs within a given delay. SUMMARY

[0005] At present, for the content mentioned in the background art, we currently provide a sensor node based end-to-end path data collection optimization method, wherein the data collection strategy proposed first constructs a data forwarding tree, and then uses an ant colony optimization (ACO) based algorithm to jointly optimize CP selection and path planning, which is extended to combine other methods that independently solve CP selection and path planning, and fully utilizes the maximum path length to solve the problem in the network node.

[0006] A sensor node based end-to-end path data collection optimization method, comprising an initial stage, ACO based algorithm optimization of CP selection and path planning, characterized by:

[0007] I. In the initial stage, Prim algorithm is used to construct a data forwarding tree; the base station is taken as the root, and all nodes in the data forwarding tree can only send their data packets to their parent nodes. When a node is selected as a CP, the edge between the CP and its parent node is broken, and the CP can directly upload the collected data packets to the mobile receiver. In addition, a new data forwarding tree with the CP as the root is generated, and at the same time, the isolated data forwarding tree is regarded as a group, and all nodes forward data to the CP through the parent node in their respective groups;

[0008] In WSN, the topology is represented as an undirected graph G = 〈V, E), where V is the set of sensor nodes in WSN, and E is the edge between two sensor nodes. In addition, a complete graph G' = <V, E'> is constructed based on the original input graph G, where E' represents the set of feasible motion paths of the mobile receiver.

[0009] II. In the stage of ACO based algorithm optimization of CP selection and path planning, in the WSN topology, sending and receiving a data packet will consume a certain amount of energy, and the energy consumption is proportional to the size of the data packet; the sensor node generates a fixed size data packet in a round, at this time the energy consumption of sensor i in each round is formulated as follows:

[0010] E i =E t ×(nd i +1)+E r ×nd i

[0011] Where E t and E r represent the energy consumption of sending and receiving a data packet respectively, nd iis the number of data packets sent to sensor i, the lifetime of WSN depends on the node that consumes energy first, while CP consumes the most energy, in order to prolong the lifetime of WSN, we should minimize the energy consumption of CP with the most energy consumption; at the same time, nd i represents the number of nodes in the subtree rooted at node i , we update pheromone in each iteration using the best result of elite ants, in each iteration, several ants search feasible solutions independently according to a probability function, and then update pheromone using the best solution with the minimum max nd i , the probability function is defined as follows:

[0012]

[0013] where P ij represents the probability of the ant moving from node i to node j , J(i) represents the subset of nodes in the neighbor node set G' of node node i that have not been accessed by ants, τ ij is the quantity of pheromone, η ij is the heuristic value, which is defined as follows:

[0014]

[0015] where dist(i, j) is the Euclidean distance, w j represents the gain value of selecting node j as CP, which is defined as follows:

[0016] w j = nd j × H(j, root j )

[0017] where H(j, i) is the hop distance from node j to node i in the data forwarding tree, root j is the root of the subtree containing node i , when node j is selected as CP, it does not need to send data packets to its parent node, and all nodes in the path from node j to the root save energy consumption;

[0018] Let T = <v'0 = v0, v'1, v' m , v' m+1 = v0> represent the best solution after the kth iteration, the elements in T represent CP, and the order corresponds to the patrol order of the moving receiver, the following rules are used to update pheromone:

[0019] τ k+1 (v i , v i ) = p x τ k (v i , v i ) + Q x Δτ

[0020]

[0021] where p is the pheromone evaporation rate, Q and σ are adjustable control scalars, Δτ represents the pheromone increment, M nd is the maximum value in all nodes i , F is the fairness index, representing the load balancing degree of all CPs;

[0022] An algorithm based on convex polygon is adopted for path optimization, specifically including:

[0023] (1) Construction of convex polygon, this step aims to build the bounding box of all CPs, where the box is a convex polygon and has the minimum area;

[0024] (2) After the construction of convex polygon, we iteratively connect the remaining disconnected points to the polygon;

[0025] (3) Estimation of the run length, we use the total length of the polygon to estimate the minimum length of the tour path.

[0026] Further, the time complexity of the ACO algorithm in the CP selection and path planning stage is O(n 2 ), where n represents the number of sensors, and the path estimation complexity of the ACO algorithm in the CP selection and path planning stage is O(m 2 ), where m represents the number of CPs and is less than n, therefore, the time complexity of each ant is:

[0027] O(m x (O(n 2 ) + O(m 2 ))) = O(mn 2 )

[0028] Given t1 iterations, each iteration contains t2 ants, the overall time complexity of the proposed method is

[0029] O(n 2 ) + O(mn 2 ) x t1 x t2 = O(t1t2mn 2 ).

[0030] The optimization method of the application selects a suitable sensor node as a CP, and plans a path for a mobile receiver to access all CPs simultaneously, the proposed strategy is based on ACO, and can be extended to be combined with other data collection methods, extended to be combined with other methods of independently solving CP selection and path planning, and make full use of the maximum path length to solve the problem in the network node, and the proposed end-to-end strategy can achieve better performance in network lifetime, total energy consumption and fairness index. BRIEF DESCRIPTION OF DRAWINGS

[0031] The specific embodiments of the application will be further described in detail below with reference to the accompanying drawings.

[0032] Figure 1 The path length constraint is fully utilized in the application.

[0033] Figure 2 The CP selection and path planning are independent in the application.

[0034] Figure 3 The WSN total energy consumption line graph of different algorithms in two scenarios in the embodiment of the application.

[0035] Figure 4 The FI (fairness index) line graph comparison of different methods in the embodiment of the application.

[0036] Figure 5 The EI (effective index) line graph comparison of different methods in the embodiment of the application.

[0037] Figure 6 The running time line graph comparison of different methods of different node numbers in the embodiment of the application. DETAILED DESCRIPTION

[0038] In order to make the purpose and advantages of the application more clear and apparent, the application will be further described below with reference to the embodiments; it should be understood that the specific embodiments described herein are only used to explain the application, and are not used to limit the protection scope of the application.

[0039] The preferred embodiments of the application will be described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the application, and are not used to limit the protection scope of the application.

[0040] The application relates to an end-to-end path data collection optimization method based on sensor nodes, based on the observed fact that CP selection and travel planning are not independent. The proposed data collection strategy first constructs a data forwarding tree, and then jointly optimizes CP selection and path planning using an ant colony optimization (ACO) based algorithm.

[0041] In a WSN, it consists of a set of sensor nodes, a communication topology network and a base station. Each node is responsible for sensing the environment state and generating data packets periodically. Those data packets must be delivered to the base station within a delay T. To reduce the energy consumption and prolong the lifetime of the WSN, mobile sinks are used to collect the data packets. Since the mobile sink has a fixed moving speed, the moving time is proportional to the tour length. Given a delay T, there is a length L max corresponding to it, where L max denotes the maximum path length. To satisfy the length constraint, all sensor nodes are divided into several groups. Each group selects

[0042] a node as the collection point to collect data packets from other nodes and upload these data packets to the mobile sink. Other nodes will transmit their data packets to the CP through multi-hop. The mobile sink visits all CPs and collects all data packets. Transmitting and receiving a data packet consumes some energy. The energy consumption is proportional to the data packet size. In this paper, we assume that a sensor node generates a fixed size data packet in a round. We can formulate the energy consumption of sensor i in each round as follows

[0043] E i = E t × (nd i + 1) + E r × nd i

[0044] where E t and E r denote the energy consumption of transmitting and receiving a data packet respectively, nd i is the number of data packets transmitted to sensor i, for prolonging the lifetime of the WSN, we need to minimize the energy consumption of the CP with the largest energy consumption; the objective is defined as follows

[0045]

[0046] where m is the number of CPs. From (1), we can find that E i ∝ nd i For CP i, (nd i + 1) is equal to the number of nodes in the group containing CP i. Minimizing the energy consumption of the CP is equivalent to minimizing the size of the group containing the CP

[0047] At this time, we make the following assumptions:

[0048] 1. The mobile sink moves at a fixed speed.

[0049] 2. The mobile sink collects data packets from CPs only when it arrives at the location of the CP.

[0050] 3. Compared to the circumvention time of the mobile destination, the communication time between the mobile destination and the CP is negligible. Similarly, the transmission and reception time between sensor nodes is negligible compared to the circumvention time.

[0051] 4. Each CP has sufficient capacity to store all detected data.

[0052] 5. Sensor nodes have a fixed communication range and initial energy.

[0053] 6. There are no isolated nodes; each sensor node can communicate with other nodes in a multi-hop manner.

[0054] 7. Each sensor node generates a data packet per round.

[0055] 8. The energy of the moving sink is sufficient to access all CPs within a given delay and collect packets in one round.

[0056] Path estimation

[0057] Due to the randomness of ACO-based algorithms, it's easy to obtain a path with multiple intersecting segments. For the same CP (Cost Per Journey), a travel path with intersecting segments implies a longer travel length. The problem of intersecting segments also leads to suboptimal solutions. To alleviate this problem, we utilize an efficient path estimation algorithm to calculate the travel length LT and refine the travel path.

[0058] The algorithm consists of three steps: constructing the convex polygon, connecting the interior points, and estimating the run length.

[0059] 1. Construction of the convex polygon: This step aims to construct the bounding boxes of all CPs, where the boxes are convex polygons with the minimum area. First, choose the bottom left corner as the starting point p0. Let line l pass through p0. Then, using p0 as the center, rotate l counterclockwise until this line touches any other CP; it will be marked as p1. Then, using the touch point as the new center, repeat the rotation operation. The touch points will be labeled in order as {p1, p2, p3, ...}. This operation will be repeated until the touch point becomes the starting point p0. Finally, a convex polygon is constructed. <p0,p1,p2,...,p k ,p0> where (k+1) represents the number of vertices of the convex polygon.

[0060] 2. Connecting Internal Points: After constructing a convex polygon, there may still be some disconnected points. Clearly, these points are all located inside the convex polygon. We iteratively connect the remaining points to the polygon. Let P... CP and P R Let V and P represent the set of vertices and the set of remaining points in the convex polygon, respectively. In each iteration, the point V∈P with the minimum selection cost will be selected.R and inserted into the polygon. The selection cost of a point V is defined as follows:

[0061]

[0062] dist incre (V, P i , P i+1 ) = dist(V, P i ) + dist(V, P i+1 ) - dist(P i , P i+1 )

[0063] cost V evaluates the length increment after inserting a point V into a polygon for the construction of energy-aware paths using mobile sink data. The position of the inserted point V' is at pos, where:

[0064]

[0065] P i ∈ P CP After inserting the point V' into the polygon, the polygon has several changes.

[0066] V' will be labeled as P pos and the indices of the original points' labels {P pos , P pos+1 ,...} will be incremented by one, i.e., their new labels are {P pos+1 , P pos+2 ,...} respectively. In addition, the point V' will be removed from P R . Then the above procedure is repeated until P R is empty.

[0067] The estimate of the tour length, when P R is empty, P CP presents a tour path of the polygon. We use the total length of the polygon to estimate the minimum length of the tour path.

[0068] To verify the effectiveness of the proposed method, we conduct extensive experiments. The Python simulator is used to simulate the network environment. We consider a 300 million by 300 million network area. Several sensor nodes are randomly deployed on the network and can communicate with each other through multi-hop. The communication range is 30 m. In general, we assume that each sensor node generates one data packet per round of data collection, where the data packet has a fixed size. The energy consumed to send and receive one data packet is 0.021 & 3 (J) and 0.015 J, respectively. The initial battery energy of each sensor is 20 J. In addition, the moving speed of the mobile sink is 1 m / s. All the simulation settings are described in Table 1. The deployment of sensor nodes depends on its specific application. Different deployment characteristics will lead to different performance evaluations. We conduct extensive experiments for two cases:

[0069] (1) Unbalanced deployment (UD) scenario: All sensor nodes are uniformly deployed in the network area.

[0070] (2) Balanced deployment (BD) scenario: The network area is divided into a 3x3 grid, with each grid having a size of 100 m x 100 m. Then, we select 5 grids and deploy the same number of sensors in each grid. The grids do not share any edges with each other.

[0071] To benchmark our method, we compare it with the "stop-and-collect" version of the Energy-Aware Path Construction (EAPC-SC) algorithm 8 in terms of network lifetime, total energy consumption, fairness index, efficiency index, and running time. The EAPC-SC algorithm has more efficient performance than existing data collection mechanisms for solving the same problem as considered in this patent. EAPC-SC first constructs a minimum spanning tree (MST), selects a set of CPs, and then constructs data collection paths. The selection of CPs is based on a heuristic function, which is the same as η ij in our proposed algorithm. In addition, the path estimation is a modified version of the EAPC-SC path construction. To some extent, our proposed algorithm is an evolution of EAPC-SC. Our proposed method addresses the problem of suboptimal solutions caused by the independence between CP selection and path planning. We compare with the EAPC-SC algorithm to verify the feasibility of the proposed method.

[0072] Table 1. Simulation experiment settings

[0073] (Table 1)

[0074] In the experiment, the distribution of sensors is randomly initialized, and the proposed algorithm is a random-based ant colony optimization algorithm. Therefore, we conduct ten repeated experiments to mitigate the impact of randomness. The following experiments are conducted.

[0075] Network lifetime

[0076] Network lifetime describes the maximum working time of the whole WSN, which is an important indicator of a well-designed WSN. It uses rounds as the measurement unit. The higher the value, the better the performance. In each round, the mobile sink visits all the CPs according to the scheduled route and collects all the data packets from the CPs. The lifetime is defined as the time before a node runs out of energy. Very simply, we assume that the sensor consumes energy only for transmitting and receiving data packets. The lifetime of the network depends on the total number of sensor nodes and the mobile path of the mobile sink. We compare the lifetime of different numbers of sensor nodes in two scenarios, where the number of sensor nodes ranges from 10 to 300. The results are shown in Tables 2 and 3, where ACO represents our proposed strategy. We conduct experiments in both scenarios under different time delay constraints. It is clear that our proposed end-to-end strategy can further prolong the network lifetime.

[0077] Table 2. Network lifetime (rounds): BD scenarios.

[0078]

[0079] (Table 2)

[0080] Table 3. Network lifetime (rounds): UD scenarios.

[0081]

[0082] (Table 3)

[0083] Please refer to the attached Figure 3 Figure 2 shows the total energy consumption, which is defined as the sum of the energy consumption of all sensors in each round of the WSN. We compare the total energy consumption of WSNs with different algorithms in two scenarios.

[0084] Please refer to the attached Figure 4 Figure 3 shows the fairness index, which should quantify the uniformity since the introduction of the mobile sink-based data collection strategy to alleviate the problem of uneven energy consumption. Higher energy consumption uniformity will result in lower total energy consumption and longer network lifetime. The fairness index (FI) reflects the balance between nodes and is defined as follows:

[0085]

[0086] where n is the number of sensor nodes, E i is the energy consumption of sensor i per round.

[0087] Figure 4The FI (fairness index) of different methods are compared in Fig. 6, in which it is revealed that the proposed end-to-end strategy can also achieve a higher upper bound of the fairness index.

[0088] Effective index

[0089] The given delay constraint corresponds to the maximum tour length of the mobile sink. Making full use of the maximum length can effectively improve the network lifetime and the fairness index. The effective index (EI) describes the utilization of the given maximum length, which is defined as follows:

[0090] where m is the number of CPs, N is the total number of sensors, L act and L max denote the actual tour length and the maximum length of the mobile sink, respectively

[0091] Appendix Figure 5 The EI (effective index) of different methods are compared in Fig. 6. The results show that our method produces an effective index comparable to EAPC. Since the DDR protocol uses static sinks, the actual tour path length is zero.

[0092] Figure 6 The running time of different methods with different numbers of nodes are compared.

[0093] The time complexity of EAPC and DRR methods are the same, and the running time of the two methods plotted in the figure almost overlap. In order to better distinguish, we change the scale of the vertical axis to logarithm. Our proposed method is more time-consuming and less responsive, which makes it difficult to apply to online cases. However, there are many applications that run the algorithm to generate data collection strategies offline, in which the structure of the WSN is almost unchanged, and the proposed method is only executed when the structure changes

[0094] . Therefore, the running time is acceptable in the offline case.

[0095] Data collection is an important topic in WSNs, as it consumes energy and affects network lifetime. Using a mobile sink can significantly reduce

[0096] energy consumption, extending network lifetime. In this paper, we propose an end-to-end data collection strategy that selects appropriate sensor nodes as CPs and plans a path for the mobile sink that visits all CPs at the same time. The proposed strategy is based on ACO and can be extended to be combined with other data collection methods. Experimental results show that the proposed end-to-end strategy can achieve better performance in terms of network lifetime, total energy consumption, and fairness index, especially in the UD scenario.

[0097] The technical scheme of the present application has been described in combination with the preferred embodiments shown in the drawings, but it is easy for those skilled in the art to understand that the protection scope of the present application is obviously not limited to these specific embodiments. Those skilled in the art can make equivalent changes or replacements to the related technical features without departing from the principles of the present application, and the technical schemes after the changes or replacements will all fall within the protection scope of the present application.

Claims

1. A sensor node based end-to-end path data collection optimization method, comprising an initial stage, ACO based algorithm optimization CP selection and path planning, characterized in that: I. In the initial stage, Prim algorithm is used to construct a data forwarding tree; Taking the base station as the root, all nodes in the data forwarding tree can only send their data packets to their parent nodes, when a node is selected as the CP, the edge between the CP and its parent node is broken, the CP can directly upload the collected data packets to the mobile receiver; In addition, a new data forwarding tree with the CP as the root is generated, at the same time, the isolated data forwarding tree is regarded as a group, and all nodes forward data to the CP through the parent node in their respective groups; In the WSN, the topology is represented as an undirected graph G = (V, E), where V is the set of sensor nodes in the WSN, and E is the edge between two sensor nodes, in addition, a complete graph G′ = (V, E′) is constructed on the basis of the original input graph G, where E′ represents the set of feasible motion paths of the mobile receiver; II. In the ACO based algorithm optimization CP selection and path planning stage, in the WSN topology, sending and receiving a data packet will consume a certain amount of energy, and the energy consumption is proportional to the size of the data packet; Sensor nodes generate a fixed size data packet in a round, at this time, the energy consumption of sensor i in each round is formulated as follows: E i = E t × (nd i + 1) + E r × nd i where E t and E r represent the energy consumption of sending and receiving a data packet, nd i is the number of data packets sent to sensor i, to extend the life cycle of WSN, the energy consumption of CP with the largest energy consumption needs to be minimized; at the same time, nd i represents the number of nodes in the subtree rooted at node i , we use the best result of elite ants to update pheromone in each iteration, in each iteration, several ants search for feasible solutions independently according to a probability function, and then use the best solution with the minimum max nd i to update the pheromone, and the probability function is defined as follows: where P ij represents the probability of an ant moving from node i to node j , J(i) represents a subset of the set of neighboring nodes G' of node node i that has not been visited by an ant, τ ij is the quantity of pheromone, and η ij is a heuristic value, defined as follows: where dist(i, j) is the Euclidean distance, w j represents selecting node j As the gain value of the CP, it is defined as follows: w j = nd j × H(j, root j ) where H(j, i) is the hop distance from node j to node i , root j is the root of the sub-tree containing node i , when node i is elected as a CP, it does not need to send data packets to its parent node, and all the nodes in the path from node j to the root save energy consumption; Let T = <v′0=v0,v′1,v′ m ,v′ m+1 =v0> represents the optimal solution after the k-th iteration. The elements in T represent CP, and their order corresponds to the inspection order of the moving receiver. The following rules are used to update pheromones: τ k+1 (v i ,v i ) = p x τ k (v i ,v i ) + Q x Δτ where p is pheromone evaporation rate, Q and s are adjustable control scalars, AT represents the increment of pheromone, M nd is the maximum value among all nodes i m represents the number of CPs, F is the fairness index, which represents the load balancing degree of all CPs; In terms of path optimization, a convex polygon based path estimation algorithm is used, which specifically includes: (1) Construction of convex polygon, this step aims to construct the bounding box of all CPs, where the box is a convex polygon and has the minimum area; (2) After the construction of the convex polygon, we iteratively connect the remaining disconnected points to the polygon; (3) Estimation of run length, we use the total length of the polygon to estimate the minimum length of the tour path.

2. The sensor node based end-to-end path data collection optimization method of claim 1, wherein, The algorithm of the ACO optimizes the time complexity in the CP selection and path planning phase as O(n 2 ), where n represents the number of sensors, and the algorithm based on ACO optimizes the path estimation complexity in the CP selection and path planning phase as O(m 2 ), where m is less than n, and therefore the time complexity of each ant is: O(m x (O(n 2 )+O(m 2 ))) = O(mn 2 ) Given t1 iterations, in each iteration contains t2 ants, the overall time complexity of the proposed method is O(n 2 )+O(mn 2 )×t1×t2=O(t1t2mn 2 )

Images