Research on rechargeable agricultural wireless sensor network based on ZigBee immune routing repair algorithm | Scientific Reports

Scientific Reports volume 15, Article number: 5756 (2025) Cite this article

Metrics details

WSN (wireless sensor network) plays a very important role in the agricultural environment monitoring. Although solar energy and other power supply methods are used to solve the node energy problem, the monitoring equipment works outdoors for a long time, which is easily affected by the environment. The supply is unstable to cause abnormalities in some nodes. So this study proposes a ZIRRA algorithm (ZigBee immune routing repair algorithm) for the rechargeable agricultural WSN. It simulates the working mechanism of the immune system and designs modules such as identification, processing, cloning and storage, which can provide a better repair strategy for abnormal nodes. Then it compares the quality of the backup nodes and replaces the backup nodes with poor quality, so that the optimal paths are maintained between source nodes and middle relay nodes, which increases the optimization ability of the algorithm. The experimental results show that the ZIRRA algorithm shows significant advantages in routing node repair mechanism. Compared with the LFRA, AR-TORA and ICCO algorithms, the average routing energy consumption of the ZIRRA algorithm reduced 35.33%, 58.37% and 45.15% , the data transmission delay reduced by 23.72%, 36.74% and 16.28%, and the average node survival time extended 25.08%, 33.55% and 13.88%. In addition, the maximum communication time and network throughput of the ZIRRA algorithm increased 44.49% and 13.03% at the scale of 1000 to 2000 nodes. These quantitative results show that the ZIRRA algorithm can improve the energy efficiency, transmission reliability and stability. The ZIRRA algorithm draws on the working principle of the immune system and repairs abnormal nodes through identification, processing, cloning and storage modules. Unlike the traditional node repair algorithms, the ZIRRA algorithm has higher efficiency and accuracy in identifying and processing abnormal nodes through the improved clone tracking algorithm. It uses an improved clone tracking algorithm in the learning module, improves the cloning and mutation mechanisms, and generates the optimal antibodies for repairing abnormal nodes. It also integrates an adaptive energy management strategy to cope with fluctuations in energy levels by prioritizing the transmission of critical data and reducing the frequency of non-essential communications, which improves the network stability and data transmission reliability.

In agricultural production, the growth of crops is affected by different environmental parameters. Due to the influence of various environmental factors, the information collection model of the traditional manual measurement cannot meet the requirements of the modern agricultural production and management1,2. In addition, most agricultural production bases are vast in area, and the cost of deploying wired networks is high. ZigBee has the characteristics of low power consumption, low cost and easy deployment, so many large farms use the monitoring equipment built with ZigBee and sensors. WSN (Wireless Sensor Network) plays an important role in the agricultural environment monitoring, but the traditional WSN is susceptible to energy limitations. So many farms build the rechargeable wireless sensor network (RCWSN) to collect the environmental data by installing the energy devices3. Because the RCWSN is a rechargeable wireless sensor network, in which each sensor node is equipped with an energy harvesting module (such as a solar panel) and a charging device, the service life of the network can be extended by self-charging. However, the monitoring equipment is installed directly in the outdoor environmental monitoring. Although there are protection measures such as waterproofing and high-temperature protection for the monitoring equipment, various natural factors (such as heavy rain, strong wind, etc.) and other factors (such as wild Animal damage, etc.) may still cause problems such as communication link breakage, network fragmentation and functional failure. Therefore, the route repair capability of WSN has become a key factor to ensure the smooth network communication4.

In order to solve the above problems, some scholars have made some achievements. Yang et al.5 established a risk aversion model and path optimization scheme based on the concept of CVAR (conditional value at risk) to optimize the energy usage and node charging efficiency, which solves the network robustness problem. Khan et al.6 proposed an efficient multilevel probabilistic model for the abnormal traffic detection of wireless sensor networks. It used the new mechanism of Bayesian model to detect abnormal data traffic and distinguish FC and DDoS attacks, which solves the problem of difficulty in repairing abnormal nodes. Moussa et al.7 proposed an energy-aware cluster-based routing protocol (ECRP). It introduced a multi-hop routing algorithm and fault-tolerance mechanism to balance the energy consumption and maximize the battery life of sensor nodes. Li et al.8 proposed an adaptive seasonal autoregressive integrated moving average model (ASARIMA) for the solar energy harvesting forecasting to extend the node lifetime of solar-powered wireless sensor networks. Surenther et al.9 introduced a deep learning-based grouping model approach (DL-GMA). It uses a recurrent neural network (RNN) with long short-term memory (LSTM) to improve the energy and transmission efficiency of WSNs.

Gao et al.10 proposed a multi-layer iterative decoupling optimization algorithm based on the Dinkelbach method to optimize the solar energy collection cooperative SWIPT energy scheduling (SS-CoES), which achieves the energy self-sustainability of rechargeable wireless sensor networks (RWSNs) in remote areas. Li et al.11 proposed an immune clustering and coverage optimization algorithm (ICCO), which includes the immune clustering and coverage optimization algorithm. Zhang et al.12 proposed a link fault repair algorithm of wearable wireless sensor networks based on the polygon fermat point. It can repaired faulty links by inserting relay nodes after a fault occurs, which eliminates nodes that do not meet energy requirements. Zhao et al.13 used the deep reinforcement learning (DRL) to formulate our resource allocation strategy to manage power and time, which maximizes the throughput under a Markov decision process model constructed for this optimization problem in the subnetwork. Roberts et al.14 proposed an optimized ticket manager based energy-efficient multipath routing protocol (TMERP) to extend the lifetime of WSNs and minimize energy consumption.

Although the above studies have made some achievements in renewable energy driven wireless sensor networks (RCWSN), they mainly focus on node anomaly detection and repair, but lack the ability to tolerate routing failures. Some algorithms are highly complex, which brings problems such as excessive computational and communication overhead in practical applications, affecting the working life of nodes. Although Liu et al.15 proposed the AR-TORA strategy, it is unsuitable for ZigBee-based RCWSN, because the node energy of AR-TORA is limited. In addition, the AR-TORA algorithm does not consider the quality preservation of n disjoint routes. In order to solve these problems, this study proposes a ZIRRA algorithm for the rechargeable agricultural wireless sensor networks. It uses routing abnormalities caused by the node energy depletion as antigens and corresponding repair nodes as antibodies. It also designs the detection, response, learning and memory modules to simulate the working principle of the immune system. The improved clone tracking algorithm is adopted in the learning module, which improves the clone and mutation mechanism. Through the simulated immune system modules work together, the ZIRRA algorithm provides repair strategies for abnormal nodes, evaluate the quality of backup nodes and determine whether backup nodes need to be replaced, so the RSWSN can be repaired in time to ensure the stable monitoring of agricultural environmental data.

In RCWSN network, \(\:\text{H}=\left(\text{F},\:\text{G}\left\{\text{m}\right\},\:\text{L}\right)\) represents that RCWSN consists of the static nodes F and the middle relay node M, in which each device node is equipped with an energy collection module, such as the solar power generation panel. Among them, H represents the number of communication hops from the source node to the relay node, and the unit is “hop”. F represents the set of all nodes in the RCWSN network, where each node has a fixed communication range and data collection radius. G{m} represents the set of all available paths from the source node to the relay node m. L represents the total physical length of the path, which is the sum of all hop distances from the source node to the relay node. M represents a specific relay node; m represents the general symbol of the relay node involved in the current processing. Unlike M, m is a variable that represents any one of multiple relay nodes. In the path selection process of the ZIRRA algorithm, this study considers the remaining energy, communication distance, communication energy consumption, communication delay and number of hops, which calculates the affinity function of nodes and paths to identify and select the optimal path.

In this way, nodes can collect natural energy and exhausted nodes can be resurrected to work again. Each node has the fixed transmission and data collection radius. If the distance between nodes and middle relay nodes is less than the transmission distance of nodes, there is an edge between these nodes \(\:\text{l}\in\:\text{L}\). When the distance between the source node and middle relay nodes is greater than the transmission distance of the node, the data collected by the source node will be sent to the middle relay node through multi-hop transmission. The middle relay node collects the data of each source node and provides it to users15.RCWSN is similar to the fault-tolerant topology control strategy in traditional WSN, RCWSN also achieves the fault tolerance by establishing n disjoint paths to operate normally under conditions such as energy exhaustion, hardware abnormalities and communication link errors. In the worst case, RCWSN can tolerate up to n-1 routing anomalies. Figure 1 shows the disjoint routes in RCWSN.

The disjoint routes in RCWSN.

The RCWSN is shown in Fig. 1 which assumes n =4. It passes the phase from the source node 2 to the middle relay node (M = 19). The four disjoint routes are established by the Guan algorithm16, such as 2-4-9-13-17-19, 2-5-10-14-19, 2-6-11-15-19 and 2-7-12- 16-18-19. But it does not involve the closed paths and loops. Its goal is to find multiple non-intersecting paths from the source node to the relay node, thereby optimizing the reliability and efficiency of data transmission. Assume that during use, the node 9/11/15 runs out of energy, and the first and third paths become abnormal. In this case, the ZIRRA algorithm will repair n disjoint routes and update other available nodes. The original abnormal 1st and 3rd paths will be updated as 2-4-8-13-17-19, 2- 6-12-16-18-19, the abnormal path resumes its normal operation. The specific update process is as follows:

Detect abnormal nodes

In the identification module, the algorithm first detects and identifies abnormal nodes in the path. Assume that node 9 of path 2-4-9-13-17-19 and node 11 of path 2-6-11-15-19 are disconnected due to energy exhaustion. The abnormal node reports to the relay node through neighbor nodes (parent node and child node).

Start the path repair mechanism

The relay node starts the repair module of the ZIRRA algorithm and updates the path according to the following rules:

Rule 1: Select the node with the highest remaining energy from the backup node as the new path node.

Rule 2: Give priority to the node with the shortest communication delay to reduce the path delay.

Rule 3: The number of path hops is kept to a minimum to reduce the energy consumption.

Selection and replacement of spare nodes

For path 2-4-9-13-17-19: the algorithm searches for a set of available spare nodes (such as node 8) between nodes 4 and 13, evaluates its remaining energy and communication quality, and replaces abnormal node 9 according to priority. The updated path is 2-4-8-13-17-19.

For path 2-6-11-15-19: search for a set of available spare nodes (such as nodes 12 and 16) between nodes 6 and 15, and prioritize nodes with the best energy and communication performance to replace abnormal nodes 11 and 15, which builds the new path 2-6-12-16-18-19.

Storage of repair information

After the repair is completed, the updated path information (such as the replacement node and its status) is stored in the storage module for quick call in subsequent repairs to reduce the repeated calculations. Table 1 shows the main symbols and their meanings to facilitate the algorithm description.

In all the routing, it sets \(\:\text{R}\left({\text{p}}_{\text{x}},\text{m}\right)\:\:\)between the node \(\:{\text{p}}_{\text{x}}\)and m, the factors affects to select the abnormal repair node \(\:{\text{R}}_{\text{y}}\left({\text{p}}_{\text{x}},\text{m}\right)\:\)mainly include: (1) \(\:\text{n}\text{r}\text{e}\text{n}\left({\text{p}}_{\text{x}}\right)\) represents the remaining energy of all nodes in \(\:{\text{R}}_{\text{y}}\left({\text{p}}_{\text{x}},\text{m}\right)\:\); (2) \(\:\text{n}\text{d}\text{s}\text{t}\left(\text{l}\right)\) represents the adjacent sections distance between points; (3) \(\:\text{n}\text{n}\text{e}\text{n}\left(\text{l}\right)\:\) represents the energy consumption of communication between adjacent nodes ; (4)\(\:\:\text{n}\text{d}\text{e}\text{y}\left({\text{p}}_{\text{x}}\right)\) represents the communication delay between nodes ; (5)\(\:\:\text{m}\text{h}\text{o}\text{p}\left({\text{R}}_{\text{y}}\left({\text{p}}_{\text{x}},\text{m}\right)\right)\) represents the number of communication hops from the source node \(\:{\text{p}}_{\text{x}}\) to the middle relay nodes17. In ZIRRA, the above factors determine the routing \(\:{\text{R}}_{\text{y}}\left({\text{p}}_{\text{x}},\text{m}\right)\) as the affinity function of \(\:\text{a}\text{n}\text{f}\left({\text{R}}_{\text{y}}\left({\text{p}}_{\text{x}},\text{m}\right)\right)\)is defined as follows:

In Formula 1–5, \(\:{\text{T}}_{1}\) represents the ratio of the route length of the routing node \(\:{\text{R}}_{\text{y}}\left({\text{p}}_{\text{x}},\text{m}\right)\) to is the length of all route nodes in \(\:\text{R}\left({\text{p}}_{\text{x}},\text{m}\right)\); \(\:{\text{T}}_{2}\) represents \(\:\text{t}\text{h}\text{a}\text{t}\) the ratio of the energy consumed by routing node \(\:{\text{R}}_{\text{y}}\left({\text{p}}_{\text{x}},\text{m}\right)\) to the sum of the energy consumed by all routing nodes in \(\:\text{R}\left({\text{p}}_{\text{x}},\text{m}\right)\) to complete a transmission; \(\:{\text{T}}_{3}\) represents the ratio of the communication delay of a routing node \(\:{\text{R}}_{\text{y}}\left({\text{p}}_{\text{x}},\text{m}\right)\:\) to the sum of the network communication delays of all routes \(\:\text{R}\left({\text{p}}_{\text{x}},\text{m}\right)\) after complete a transmission; \(\:{\text{T}}_{4}\) represents the ratio of the hops number for routing node \(\:{\text{R}}_{\text{y}}\left({\text{p}}_{\text{x}},\text{m}\right)\) to the sum of hops number for all routes in \(\:\text{R}\left({\text{p}}_{\text{x}},\text{m}\right)\:\)to complete a transmission. \(v_{1} ,v_{2} ,v_{3}\) and\(\:\:\:{\text{v}}_{4}\) represents the weights corresponding to \(\:{\text{T}}_{1}\), \(~T_{2} ,T_{3}\) and \(\:{\text{T}}_{4}\) and \(\:{\text{v}}_{1}+{\text{v}}_{2}+{\text{v}}_{3}+{\text{v}}_{4}\)=100%, of which \(\:{\text{v}}_{1}\)=35% ,\(\:{\text{v}}_{2}\) =30% ,\(\:{\text{v}}_{3}\) =20% and\(\:{\:\text{v}}_{4}\) =15%. Because the selection of weights is based on the degree of influence of each factor on the path optimization. Therefore, \(\:{\text{v}}_{1}\) corresponds to the weight of the remaining energy of the node, which has the highest priority and accounts for 35%. The remaining energy directly determines the survival time of the node, which is crucial to the stability and long-term operation of the network. \(\:{\text{v}}_{2}\) corresponds to the weight of communication delay, accounting for 30%. The lower latency can significantly improve the efficiency and real-time performance of data transmission. Especially the agricultural monitoring, the transmission of real-time data is particularly important. \(\:{\text{v}}_{3}\) corresponds to the weight of the path hop count, which accounts for 20%. The excessive hop count will increase the energy consumption and transmission delay of data packets, so the proper optimization of the hop count is necessary, but its priority is lower than the remaining energy and delay. \(\:{\:\text{v}}_{4}\) corresponds to the weight of communication quality (such as signal strength), which accounts for 15%. Although the communication quality also has an impact on path selection, which considers the adaptive capabilities of ZigBee in the network topology, so this factor has a relatively low priority. The higher the similarity of the route is, the better the quality of the route is. The energy model is also adopted as the energy consumption model of nodes in RCWSN17. The energy consumed by a single node to collect, send and receive w-bit data are as follows:

In Formula 6–8, j represents the distance between the node and its transmission target node,\(\:\:{\text{n}\text{n}\text{e}\text{n}}_{\text{c}\text{o}\text{l}\text{l}}\left(\text{w}\right)\) represents the cost in collecting w bits of data. \(\:{\text{n}\text{n}\text{e}\text{n}}_{\text{s}\text{e}\text{n}\text{d}}(\text{w},\:\text{j})\)and \(\:{\text{n}\text{n}\text{e}\text{n}}_{\text{r}\text{e}\text{c}\text{e}}\left(\text{w}\right)\:\)represents the energy consumed by sending and receiving w-bit data, e represents the channel attenuation coefficient, \(\:{\text{a}}_{1}\),\(\:{\text{b}}_{1},{\text{b}}_{2\:\:}\)and\(\:{\:\:\text{c}}_{1}\) represents the energy consumption parameters of the acquisition circuits, transmitting circuit, the transmitting amplifier and the receiving circuit. For ZIRRA, it is used for the antibody cloning, mutation and reselection in each iteration of the optimization process. The energy efficiency consumption of selection and memory cell storage are defined as \(\:{\text{n}\text{n}\text{e}\text{n}}_{\text{c}\text{l}\text{o}\text{n}}\), \(\:{\text{n}\text{n}\text{e}\text{n}}_{\text{m}\text{u}\text{t}\text{a}}\),\(\:{\text{n}\text{n}\text{e}\text{n}}_{\text{r}\text{s}\text{e}\text{l}}\) and \(\:{\text{n}\text{n}\text{e}\text{n}}_{\text{s}\text{t}\text{o}\text{r}}\). Therefore, the total energy consumption of RCWSN can be calculated by estimating the energy, which is consumed in each iteration of ZIRRA18.

The immune system is an important biological system that protects the body from disease. It can detect and remove various antigens, which have viruses and parasites to maintain good health19,20. Figure 2 shows the working process of the immune system. Its main functions are as follows:

Recognition module: During the antigen recognition, the immune system can detect and recognize non-self-antigens21.

Processing module: It produces many antibodies to clear antigens and infected cells.

Cloning module: During the antibody generation, it produces antibodies highly related to antigens.

Storage module: Through the immune system’s memory function, each invading antigen’s information is stored in memory cells. When the same antigen invades again, the memory cells will be activated rapidly to produce the corresponding antibodies to clear the antigen.

The working process of the immune system.

The above analysis shows that the problem of repairing n disjoint routes in RCWSN is very similar to the immune system that protects human health. Antigens are similar to abnormal nodes in RCWSN due to node energy depletion, and antibodies are similar to repair nodes in RCWSN for routing abnormalities22. Therefore, this study simulates the working mechanism of the immune system and designs the ZIRRA algorithm to deal with the routing repair problem in RCWSN.

In ZIRRA, the abnormal nodes caused by node failures are regarded as antigens, and the function of the detection module is to detect and identify antigens. Each node can communicate bidirectionally with its neighbor nodes within 1 hop. \(\:{\text{R}}_{\text{y}}\left({\text{p}}_{\text{x}},\text{m}\right)\) represents the node in the routing \(\:{\text{p}}_{\text{x}}\) (Node 11 is taken as an example in Fig. 3), if \(\:{\text{p}}_{\text{x}}\) works normally and send data, its parent node (node 6) can receive \(\:{\text{p}}_{\text{x}}\) as the sent data. If node 6 does not receive the message within the specified time interval, node 11 will resend the data, which is considered as the abnormal state. If the node 11 is called fail v, the parent node (node 6) and child node (node 15) of node \(\:{\text{p}}_{\text{l}\text{o}\text{s}\text{e}}\)11 are called \(\:{\text{p}}_{\text{l}\text{o}\text{s}\text{e}\_\text{k}}\) and \(\:{\text{p}}_{\text{l}\text{o}\text{s}\text{e}\_\text{r}}\). Then, the m value and abnormal status information \(\:{\text{p}}_{\text{l}\text{o}\text{s}\text{e}}\left({\text{m}}_{\text{l}\text{o}\text{s}\text{e}}\right)\) of node 11 is reported to the middle relay node (node 19) by \(\:{\text{p}}_{\text{l}\text{o}\text{s}\text{e}\_\text{r}}\), and \(\:{\text{p}}_{\text{l}\text{o}\text{s}\text{e}\_\text{r}}\) is reported to the source node (Node 2). When the source node 2 receives the node abnormality information, it will enable other backup routes for the data transmission. So the middle relay node starts the ZIRRA processing module to repair the abnormal node.

The ZIRRA immune mechanism workflow.

When the recognition module is completed, it will first search in the memory unit of ZIRRA to determine whether the memory unit has the information about the detected antigen in the meta. In the immune system, the antigen-presenting cells will provide the different information to immune cells, and the corresponding antibodies will recognize this information and stimulate the production of many antibodies23. Finally, the effect of clearing the antigen is achieved. In ZIRRA, the disjoint routes are established, so each relay node will only appear in one route. Therefore, the m value (\(\:{\text{m}}_{\text{l}\text{o}\text{s}\text{e}}\)) of the abnormal node is used as the specific antigen information to search the repository.

During the search process, the response module will search m node value of the abnormal node stored in the repository. Assuming that there are NRM pieces of abnormal node information stored in the repository, the abnormal node information of the repository satisfies Formula 9, the repair node corresponding to the abnormal node is regarded as an antibody and used to repair the abnormal node:

In Formula 9,\(\:{\:\text{Q}}_{\text{i}}\) represents the number stored in the repository i piece of abnormal node information,\(\:\:\text{N}\text{R}\text{M}\_\text{M}\text{a}\text{x}\) represents the maximum storage capacity of the repository.

The cloning module of the ZIRRA algorithm is a crucial part of the path repair process, which ensures that the optimal path can be selected in the node anomaly repair by generating multiple alternative repair nodes (antibodies). By introducing the diversity mechanism, it avoids falling into the local optimal solution. If the corresponding memory information cannot be found in the repository to match the abnormal node, the middle relay node repairs the request through the broadcast route (BDRR) information to start the cloning mechanism. The BDRR information is propagated to the source node by the backup node\(\:\:{\text{p}}_{\text{x}}\). When \(\:{\text{p}}_{\text{x}}\:\)receives the ZRR information, it opens the ZigBee route repair to response the ZRR information, which contains the remaining available energy and the node location24. This process continues until the middle relay node receives the ZRR information, and the middle relay node is extracted from BDRR, so the improved clone tracking algorithm is used to select routes with high similarity \(\:{\text{R}}^{\text{s}}\left({\text{p}}_{\text{x}},\text{m}\right)\) as the repair node.

The above ZRR information relay process continues until the middle relay node receives the ZRR information. Then the corresponding information of middle relay node is extracted from ZRR and an improved clone tracking algorithm is used to select routing nodes with the high similarity as repair nodes. In this process, the clone tracking algorithm generates the corresponding antibodies to repair the abnormal nodes, which is important in the routing repair process25. However, the traditional clone tracking algorithm is resistant to the defects of poor population diversity and the inaccurate cloning hinders the production of optimal antibodies. In order to overcome these defects, this study improves the clone tracking algorithm26.

First, the middle relay node uses the extracted information to generate the antibody set ABCT randomly; Formula 1 is used to calculate the similarity between each antibody and the antigen; It selects O antibodies with the best similarity from ABCT to build a temporary antibody set ABCT ‘, which calculates the similarity between antibodies. Then, \(\:{x}_{a}\) is one of ABCT ' antibody, q’ is the number of ABCT ' antibodies, \(\:{dist}_{ab}\) is the euclidean distance between the antibody \(\:{x}_{a}\:{and\:x}_{b}\). The larger \(\:{anf}_{ijij}\left({x}_{a}\right)\) is, the greater the antibodies’ similarity between \(\:{x}_{a}\) and ABCT ' is. It is shown in Formula 10 :

Then each ABCT ' antibody is cloned.The clone size and the proximity between antibody and antigen, and the correlates with the proximity between antibodies are calcuated. In order to maintain the diversity of good antibodies and antibodies, the clone size should be directly proportional to the similarity between the antibody and antigen, and inversely proportional to the similarity between the antibodies27. If The antibodies and other antibodies of ABCT ' have higher / lower similarity, the clone size of that antibody will be suppressed. Therefore, CS is the clone size of \(\:{\text{x}}_{\text{a}}\) antibody in ABCT ‘, which is shown in Formula 11 :

In Formula 11, \(\:\left|\:.\right|\)represents the upper limit of non-integer, \(\:{\upomega\:}\) represents the pre-defined cloning coefficient, \(\:\left|\:\text{q}\right|\) represents the number of antibodies in ABCT, \(\:\text{a}\text{n}\text{f}\left({\text{x}}_{\text{a}}\right)\) represents the similarity between the antibody \(\:{\text{x}}_{\text{a}}\) and the antigen. Some cloning antibodies are selected for mutation. In order to prevent the antibody from falling into the local optimum, the mutation rate is modified from the original fixed probability to the dynamically changing mutation rate according to the regulation mechanism of the immune system. \(\:{\text{x}}_{\text{a}}^{\text{K}}\) represents the mutation rate of the cloned antibody, which is shown in Formula 12 :

In Formula 12, \(\:{\text{G}}_{{\upgamma\:}}^{{\upsigma\:}}\) represents the initial mutation rate, \(\:{\upalpha\:}\) represents the fixed mutation coefficient. In this way, the resistance antibodies with high original similarity have lower mutation rates, and the original antibodies are similarity to other antibodies with mutation rates, which avoids the distribution of many similar antibodies in local areas. The antibodies of ABCT’ are reselected. In the reselection rule, the original antigen and selected activity of the cloned variant antibody are againsted the antigen. The euclidean distance between antibody \(\:{{\upbeta\:}}_{{\uptau\:}}\:\)and antigen \(\:{{\uprho\:}}_{{\uptau\:}}\) is shown in Formula 13 :

The activity of the antibody against the antigen is shown in Formula 14 :

In Formula 13 and 14,\(\:\:\text{E}\text{L}\text{D}\text{T}\left(\text{a},\text{b}\right)\) represents the Euclidean distance between the antibody \(\:{{\upbeta\:}}_{\text{a}{\uptau\:}}\) and antigen \(\:{{\uprho\:}}_{\text{b}{\uptau\:}}\); \(\:\left|\text{q}{\prime\:}\right|\) represents the number of antibodies in ABCT’, which is the collection of antibodies after the selection and cloning. \(\:{{\upbeta\:}}_{\text{a}{\uptau\:}}\) represents the expression value of the a-th antibody on the specific dimension \(\:{\uptau\:}\). \(\:{{\uprho\:}}_{\text{b}{\uptau\:}}\) represents the expression value of the antigen on the same dimension \(\:{\uptau\:}\). \(\:\text{I}\text{R}\text{T}\text{A}\left(\text{a},\text{b}\right)\) represents the activity of the antibody against the antigen.

Then it compares the activity of each antibody to the antigen with the set threshold. If the activity is higher than the threshold of antibody retention, the lower one will be replaced by the new antibody. If the end condition of the immune mechanism is met, it stops the optimization process and return the best similarity antibody, which becomes the optimal repair node corresponding to the abnormal node \(\:{\text{R}}^{\text{s}}\left({\text{p}}_{\text{x}},\text{m}\right)\). Finally, compare the similarity between the backup node and other antibodies, if there are some antibodies whose similarity is better than the backup node, the antibody can build the backup replacement node. Through the above-improved clone tracking algorithm, the ZIRRA algorithm designs the corresponding repair nodes for abnormal nodes strategy, compares the quality of backup nodes and other available routes and provides the corresponding replacement node. The middle coordinator broadcasts the ZRRSIG signal, which has all the information of the repair node and the replacement node. The ZIRRA algorithm will also send abnormal nodes and the corresponding repair node information is stored in the repository.

In the ZIRRA algorithm, the storage module is the core component for optimizing the efficiency of path repair, which includes the abnormal information storage, historical repair record managementand fast data retrieval. The information of cloned modules is stored in the repository, including the abnormal node and repair node information. The storage mechanism comprises different storage databases, and each database records the information of corresponding abnormal nodes and repair nodes28. When the ZIRRA algorithm is used for some abnormal nodes in the first time, the corresponding repair nodes are designed and stored in the repository according to certain rules. In subsequent use, the repository rules can be updated online.

In addition, the repository has a certain capacity limit, and its maximum capacity is NRM_Max. The nodes of the repository are sorted by the useful frequency .When the cloning mechanism provides the new information that needs to be stored, if the information stored in the repository has not reached NRM_Max, the new information will be stored in the repository; if the repository has reached NRM_Max, the new information will be stored to replace the least used storage information. When the same routing exception occurs again, the ZIRRA algorithm can provide the corresponding repair node, so the frequency of the routing failure will be much higher than the traditional WSNs.

The ZIRRA algorithm contains four main modules: identification, processing, cloning and storage. In the initialization module, the initial parameters and energy levels of all nodes are set, and n disjoint paths from the source node to the relay node are established, which detects and identifies abnormal nodes in the network. Each node can communicate directionally with its neighboring nodes within one hop range. If the parent node fails to receive the data of the child node within a predetermined time, the node is marked as the abnormal node and reported to the relay node. The processing module first searches for the information of the abnormal node in the repository. If the corresponding information is found, the stored repair strategy is used; if not found, the cloning module is entered. The cloning module collects the responses of potential repair nodes by broadcasting route repair requests (BDRR). The affinity function of each potential repair node is calculated, the best candidate node is selected for cloning and mutation, and the optimal repair node is selected. The storage module stores the new repair node information in the repository, if the repository is full, the least used information is replaced with the new information. The route using the repair node is re-established, the continue monitoring and repair when necessary. The pseudocode of the ZIRRA algorithm is as follows:

ThThe pseudocode of the ZIRRA algorithm

Input: Network topology, node energy levels

Output: Optimized routing paths, repaired nodes

1: Initialization:

2: Set initial parameters and energy levels for all nodes

3: Establish n disjoint routes from source node to middle relay node

4: Identification Module:

5: for each node do

6: if node cannot communicate with its parent node then

7: Mark the node as failed

8: Report to the middle relay node

9: end if

10: end for

11: Processing Module:

12: Search for the failed node information in the repository

13: if found then

14: Use the stored repair strategy

15: else

16: Proceed to the cloning module

17: end if

18: Cloning Module:

19: Broadcast the repair request from the middle relay node

20: Collect responses from potential repair nodes

21: Calculate affinity functions for each potential repair node

22: Select top candidates based on affinity and clone them

23: Mutate the cloned candidates to maintain diversity

24: Select the best candidate as the repair node

25: Storage Module:

26: Store the new repair node information in the repository

27: Update the repository by replacing the least used information if necessary

28: Return to normal operation:

29: Use the repair node to reestablish the route

30: Continue monitoring and repairing as needed

To ensure the security of the ZIRRA algorithm in wireless sensor networks, this study incorporates the multi-level security measures to prevent or mitigate the common network threats. It ensures that the ZIRRA algorithm can resist various network threats, which ensures the integrity, confidentiality and availability. These measures are as follows:

Preventing man-in-the-middle attacks

The ZIRRA algorithm uses the symmetric encryption algorithms to protect data transmission in communications between nodes. Each node encrypts data before sending it, and the target node can decrypt it, thereby effectively preventing the man-in-the-middle attacks. In addition, the algorithm uses a two-way authentication mechanism to ensure that the identities of both parties in communication are reliable to prevent malicious nodes from joining.

Preventing node tampering.

The ZIRRA algorithm performs integrity checks regularly and verifies the integrity of nodes and data through verification and technology. If tampering of node data is detected, the algorithm isolates the infected node and notifies the network administrator. In terms of design, the algorithm takes into account physical protection measures, such as the tamper-proof design of the node shell to prevent physical damage and tampering.

Preventing other common threats.

The ZIRRA algorithm prevents denial of service attacks by limiting the request frequency of a single node and using the distributed resource management, thereby ensuring that the network can remain operational in the face of large-scale attacks. The algorithm integrates an intrusion detection system (IDS) to monitor the network traffic in real time and identify and respond to abnormal activities. The IDS system can detect potential attacks and take timely measures to prevent the spread of attacks.

This study uses the NS2 (Network Simulator Version 2.0), which is a software simulation tool29. In order to make the simulation results closer to practical applications, this study simulated \(\:2000\times\:2000{\text{m}}^{2}\:\)as the farmland environment, and 200–500 ZigBee sensor monitoring nodes are deployed in this area (Fig. 4 shows the simulation experiment node distribution),

So the middle relay node is deployed in the two center position of the dimension area. Each node has a 20 mm × 20 mm solar power panel. When the node power is exhausted or falls below a certain threshold, the nodeuses the solar power module to replenish power. This laboratory uses NREL’s open-source solar energy database (the Sun Community Solar Project Data). Each ZigBee node’s data collection radius and transmission radius are 25 and 100 m. In each round of data transmission, the size of the transmitted data packet is 500 bits, and the transmission bandwidth of the node is 250 kb/s. The number of disjoint routes from each source node to the middle relay node is n =4. All parameters of the ZIRRA algorithm are initialized. Table 2 shows the relevant initial parameter values.

Simulation experiment node distribution.

In order to evaluate the performance of the ZIRRA algorithm, this study introduces another three algorithms, which are the LFRA (Link Fault Repair Algorithm Based on Polygon Fermat Point, which is a local fault repair algorithm designed to cope with node or path failures through a local repair mechanism)30, AR-TORA (Adaptive Repair Tempo-rally Ordered Routing Algorithm, which is an on-demand distance vector routing protocol with adaptive recovery)31 and ICCO (Immune Clonal Clustering Optimization, which is an improved clone correction optimization algorithm that aims to improve the repair efficiency by optimizing the cloning process) algorithm32. The results are as follows:

Routing energy consumption

Figure 5 shows the energy consumption of abnormal node repair process in the case of 200 and 500 nodes. Compared with the LFRA, AR-TORA and ICCO algorithm, the average routing energy consumption of the ZIRRA algorithm is 6.08\(\:\text{J}\), which is 35.33%, 58.37% and 45.15% lower than another three algorithms in the case of 200 nodes. In the case of 500 nodes, its average routing energy consumption is 7.18\(\:\text{J}\), which is 79.18%, 112.88% and 96.15% lower than another three algorithms. Because another three algorithms use the centralized repair, so the number of inserted relay nodes is higher than the ZIRRA algorithm. In order to complete the repair work, unnecessary relay nodes are inserted around non-abnormal nodes, and the values of relay points and abnormal nodes are recalculated to repair the abnormal nodes, which consumes a lot of energy. In addition, the ZIRRA’s 45th, 72th, 124th and 240th repairs of 200 nodes, the 22th, 41th, 80th, 115th, 138th and 275th repairs of 500 nodes consume far less energy than another three algorithms. Because the ZIRRA algorithm has placed the corresponding repair node information in the repository, and the clone cells are activated to produce corresponding antibodies by repairing the abnormal nodes, which reduces the energy consumption in the repair process. But in another three algorithms, even if the same routing exception occurs, the same repair process must be performed, which can not save energy consumption during the repair process.

The energy consumption of abnormal node repair process in the case of 200 and 500 nodes.

In order to verify the energy loss performance of the ZIRRA algorithm in a large-scale node network, this study introduced 1000 and 2000 nodes respectively, as shown in Fig. 6. Compared with the LFRA, AR-TORA and ICCO algorithm, the average routing energy consumption of the ZIRRA algorithm is 10.22\(\:\text{J}\), which is 32.53%, 42.58% and 30.23% lower than another three algorithms in the case of 200 nodes. In the case of 2000 nodes, its average routing energy consumption is 11.45\(\:\text{J}\), which is 44.49%, 57.78% and 45.20% lower than another three algorithms. In addition, the ZIRRA’s 80th, 126th, 124th and 266th repairs of 1000 nodes, the 48th, 95th and 196th repairs of 2000 nodes consume far less energy than another three algorithms. The results show that as the number of nodes increases, the ZIRRA algorithm can still maintain high energy efficiency and reliability. The ZIRRA algorithm optimizes the path selection and energy management to ensure efficient operation in the case of a large number of nodes. It adopts a distributed processing method to reduce the computing and communication burden of the central node, improve the scalability of the systemand enable each node to autonomously participate in fault detection and repair, which reduces the communication delays and energy consumption in large-scale networks.

The energy consumption of abnormal node repair process in the case of 1000 and 2000 nodes.

Data transmission delay

Figure 7 show the data transmission delays of repair nodes. The transmission delay time of ZIRRA algorithm is 4.245ms, which reduces 23.72%, 36.74% and 16.28% compared with the AR-TORA, LFRA and ICCO algorithm. It shows that the transmission delay of ZIRRA algorithm repair nodes is much smaller than another three algorithms in the same bandwidth. Because when the ZIRRA algorithm selects the repair node, the transmission delay and route length are considered. However, The cluster head node of LFRA algorithm uses a single hop to reach the aggregation directly, which causes those cluster head nodes far from the sink node need more transmission time to maintain the communication with the sink node, it also increases the data transmission delay. The AR-TORA algorithm considers the node’s remaining energy and does not consider the transmission delay or route length. Moreover, it is a centralized large-scale repair. The ICCO algorithm optimizes network performance through the inter-cluster communication, but this method leads to excessive load on cluster head nodes. When cluster head nodes need to handle a large number of routing requests and data transmission, their processing capacity and energy consumption become bottlenecks, which increases the transmission delays. After the repair is completed, other nodes around the node that have just been repaired will die, which causes the link to fail again and increases the data transmission delay time.

The data transmission delays of repair nodes.

Node survival time

Figure 8 show the average survival time of repaired nodes. The average survival time of ZIRRA algorithm is 206 rounds, which increases 25.08%,33.55% and 13.88% compared with the LFRA, AR-TORA and ICCO algorithm. Because when the repair node is selected, the ZIRRA algorithm considers factors such as the node’s available remaining energy, the distance between nodes, the communication delay and the number of relay hops. It also compares the quality of backup nodes with other available routing nodes, so some poor-quality routes are replaced. Through this process, the ZIRRA algorithm provides n optimal disjoint routes that are always maintained. But the influence quantity factor of routing quality in another three algorithms is not as much as the ZIRRA algorithm, so only abnormal nodes are repaired. The backup nodes are not considered and replaced accordingly. Therefore, the average survival time of ZIRRA routing nodes is higher than another three algorithms.

The average survival time of repaired nodes.

Maximum communication time between nodes

Figure 9 shows the maximum inter-node communication time of repaired nodes. The average maximum communication time of ZIRRA algorithm is 8306.7s, which increases 25.60%, 45.61% and 6.10% compared with the LFRA, AR-TORA and ICCO algorithm. Because it compares the available remaining energy and distance between nodes when the repair nodes are selected, which can quickly find n disjoint shortest target routing node repair paths and put the corresponding repair node information in the repository. The cloned cells are directly activated to produce the corresponding antibodies for repairing abnormal nodes and avoid the repeated connections between nodes. The LFRA algorithm build the discrete distribution of polygonal nodes in the repair process, which is convenient to determine the position of relay nodes.

But when the number of nodes is relatively small, it must traverse to find the abnormal nodes first, which consumes a lot of energy. After the relay node and the abnormal node are connected, the available energy is less, and the continuous connection time between nodes is shorter. The AR-TORA algorithm lead to a large number of cut points in the network (the cut point refers to the key node that will cause the network to be divided when the node fails). When the number of the network nodes is small and the entire network is failed, the maximum communication time by using the AR-TORA algorithm to repair cannot be guaranteed. The ICCO algorithm focuses on fast path discovery and data transmission instead of energy optimization, which leads to faster energy exhaustion and shortens the continuous communication time between nodes.

The maximum inter-node communication time of repaired nodes.

Recovery time for outages

Figure 10 shows the recovery time for outages at different numbers of network routing nodes. The ZIRRA, LFRA, AR-TORA and ICCO algorithms have the smallest increase in the network routing recovery time, which are 18s, 40s, 46s and 35s in the range of 0-300 nodes; its route recovery time increase in the range of 300–800 nodes is 50s, 103s, 122s and 100s; its route recovery time increase in the range of 1200–2000 nodes is 81s, 104s, 115s and 94s. The ZIRRA algorithm has the most stable change in all node number ranges, and the fluctuation is smaller than other algorithms, which can maintain relatively consistent performance under different node densities, and its optimal node density is between 0 and 300.

In addition, the average recovery time of ZIRRA algorithm is 81.98s, which reduces 35.24%, 44.99% and 31.25% compared with the LFRA, AR-TORA and ICCO algorithm. This is because the ZIRRA algorithm uses a routing repair mechanism based on the immune principle, which can identify and repair faulty nodes in the network. It enables the network to quickly resume normal communication and reduces the time required for routing recovery. It has powerful error detection and correction capabilities, which can detect and correct errors when the network is interrupted, thereby reducing the recovery time and enabling the network to return to normal in the shortest time.

The recovery time for outages at different numbers of network routing nodes.

Network throughput

Figure 11 shows the changing trend of network throughput as the number of nodes changes. The average network throughput of ZIRRA algorithm is 26.96 Mbps, which increases 13.03%, 8.86% and 11.95% compared with the LFRA, AR-TORA and ICCO algorithm. Because the ZIRRA algorithm uses a simulated immune system to ensure fast and accurate repair of abnormal nodes,which reduces the downtime and maintains the best routing path. By continuously evaluating and replacing poor-quality backup nodes, it is able to maintain the best path between source and relay nodes,which minimizes transmission interruptions and ensures the efficient of data transmission. It also minimizes the need for unnecessary relay nodes and reduces the energy consumption associated with the repair process, so that lower energy consumption can be converted into longer node life and more stable network operation, which improves the total throughput.

The changing trend of network throughput as the number of nodes changes.

In terms of computational complexity, the complexity of the ZIRRA algorithm is mainly reflected in the clone tracking algorithm and path repair mechanism. The clone tracking algorithm needs to calculate the similarity between nodes and dynamically adjust the clone size, and its time complexity grows linearly with the number of candidate repair nodes. In contrast, the traditional algorithms require a full network traversal during the repair process, which is more complex, while ZIRRA reduces repeated calculations by introducing storage modules. In terms of communication and computing overhead, the distributed processing mode of the ZIRRA algorithm effectively reduces the computing burden of the central node. During the repair node selection process, the ZIRRA algorithm stores historical repair information through the storage module, enabling rapid repair of known abnormal nodes and avoiding repeated path optimization processes. Although the computational complexity is slightly higher than some traditional methods, its significantly reduced the communication overhead and energy consumption,which makes the overall overhead at a better level in large-scale networks and performs more stably.

In order to test the performance of the algorithm in sparsely connected networks, this study designed additional sparse network experimental scenarios. By reducing the node density, the average spacing was adjusted to 50–100 m, and the number of cluster head nodes was reduced. The result of sparse deployment is that some regional nodes cannot connect to the cluster head nodes through the single-hop communication, forming a sparsely connected network. In the sparsely connected network, the ZIRRA algorithm successfully connected the isolated nodes through a multi-hop routing repair mechanism, and the repair success rate reached 87.5%. However, the average transmission delay of the sparse network increased 18.2% compared with the dense network, and the node energy consumption increased 12.4%. It shows that the performance of the ZIRRA algorithm in sparse networks is better than other algorithms (LFRA, AR-TORA and ICCO). The ZIRRA algorithm also considers factors such as node residual energy, communication distance, communication energy consumption, and transmission delay. By assigning the highest weight (35%) to the remaining energy, it ensures that high-energy nodes are prioritized to participate in the path construction and prolongs the survival of network nodes. time. Experimental results show that the routing energy consumption of the ZIRRA algorithm in 500-node scenario is 35.33% and 58.37% lower than the LFRA and AR-TORA algorithm. The ZIRRA’s storage module records the repaired node information. When the similar exception occurs again, the storage information can be quickly called to repair it without re-searching the entire network. This mechanism reduces the computational overhead and energy consumption in large-scale networks. Finally, The ZIRRA algorithm introduces a dynamically adjusted cloning and mutation mechanism (Sect. 3.3.3), which avoids the unnecessary node insertion by selecting optimal candidate nodes and generating high-quality antibodies. Compared with traditional algorithms (such as ICCO), the ZIRRA algorithm reduces the energy waste when repairing nodes, which shows that its energy consumption is 30.23% lower than ICCO in the 1000-node scenario.

The ZIRRA algorithm takes into account the dependence of the solar power supply system on the renewable energy, and gives the highest weight to the remaining energy through the priority energy management mechanism, which allows the high-energy nodes to participate in routing repair and ensures the efficient allocation of energy resources during routing repair; it also combines energy perception to adjust the mutation rate in the clone module, and introduces a lightweight backup routing mechanism to maintain the basic communication capability of the network at the lowest possible energy consumption. Finally, it allocates routing tasks among multiple nodes through the distributed strategy, which avoids the single node running out of energy due to excessive participation in routing repair. The above four strategies achieve network reliability in a low-energy state. The ZIRRA algorithm considers the remaining energy, connection reliability and network connectivity when selecting backup nodes. Itmonitors the remaining energy of the node in real time through the energy perception module, and allows the nodes with higher remaining energy as backup nodes; then it uses the signal strength and packet loss rate of neighboring nodes to evaluate the connection reliability. It also evaluates the number and connection strength of its neighboring nodes to ensure that thedata transmission can be maintained in the scenario of multi-node failure. Finally, it dynamically adjusts the weight parameters during operation to adapt to different network environments and repair scenarios.

When a node needs maintenance or replacement, the ZIRRA algorithm transfers the nodes data to the adjacent standby node through a data transfer module. It introduces a redundant data storage strategy to avoid data loss caused by a single node being offline or hardware repair by storing the critical data (such as environmental monitoring data) between multiple nodes. During the repair operation, the ZIRRA algorithm uses a real-time verification algorithm to verify the integrity of the data packet (for example, using a hash value or a checksum). When a node needs service or replacement, the ZIRRA algorithm dynamically adjusts task allocation through the distributed strategy. The standby node will take over the core tasks of the node that needs maintenance while ensuring that the normal data processing process is not interrupted. Finally, the ZIRRA algorithm has designed the data priority mechanism, where the critical data will be transmitted to the gateway node first, while non-critical data is cached locally on the node and resumed after the repair is completed, thereby ensuring that the overall data quality of the network is not significantly affected during the repair period.

Although the ZIRRA algorithm has made significant progress in path repair and network optimization, there are still some limitations in practical applications. The ZIRRA algorithm needs to calculate the affinities of multiple candidate repair nodes for each path repair and performs the clone generation and mutation operations. As the size of the network increases, especially when the number of nodes reaches hundreds or even thousands, the amount of calculation will increase significantly, resulting in the longer execution time of the algorithm. Future research will introduce heuristic methods or parallel computing frameworks to reduce the amount of calculation and optimize the time complexity. In some cases, the energy consumption calculation model between nodes of the ZIRRA algorithm may not completely match the actual application scenario. Especially for complex sensor networks, the simplification of energy consumption models may lead to incompletely accurate energy efficiency optimization. Follow-up research will introduce more actual environmental parameters (such as transmission media, weather effects, etc.) into the energy consumption model, and combine machine learning methods to accurately model energy consumption prediction, thereby improving the accuracy of energy efficiency optimization.

This study proposes the ZIRRA algorithm for rechargeable agricultural wireless sensor networks. It simulates the working mechanism of the immune system and designs modules such as identification, processing, cloning and storage, which can provide a better repair strategy for abnormal nodes. Then it compares the quality of the backup nodes and replaces the backup nodes with poor quality, so that the optimal paths are maintained between source nodes and middle relay nodes, which increases the optimization ability of the algorithm. The experimental result shows that the ZIRRA algorithm has better performance than LFRA, AR-TORA and ICCO algorithm, which ensures the stability and reliability of data transmission and the stability of outdoor farmland environment monitoring. It plays an important role in the field of agricultural monitoring and has potential applications in multiple interdisciplinary fields such as smart cities, environmental monitoring, industrial automation, health monitoring and military security. For example, the ZIRRA algorithm can provide the stable communication support in the Internet of Things (IoT) environment of smart cities. Therefore, the ZIRRA algorithm has wide applicability and development potential.

The author will provide it upon request, please contact the corresponding author of this article, thank you! (Email: [email protected])

Huerta, M. K., García-Cedeño, A., Guillermo, J. C. & Clotet, R. Wireless sensor networks applied to precision agriculture: A worldwide literature review with emphasis on latin America. IEEE Geoscience Remote Sens. Magazine. 9 (4), 209–222 (2021).

Article Google Scholar

Nurlan, Z., Zhukabayeva, T., Othman, M., Adamova, A. & Zhakiyev, N. Wireless sensor network as a mesh: Vision and challenges. IEEE Access. 10, 46–67 (2021).

Article Google Scholar

kumar, C. S. & Anand, R. V. A review of energy-efficient secured routing algorithm for IoT-Enabled smart agricultural systems. J. Biosyst. Eng. 48(3), 339–354 (2023).

Article MATH Google Scholar

Gupta, N. & Kumar, A. Study on the wireless sensor networks routing for low-power FPGA hardware in field applications. Comput. Electron. Agric. 212, 108145 (2023).

Article MATH Google Scholar

Yang, W. et al. Robust wireless rechargeable sensor networks. IEEE/ACM Trans. Netw. 31(3), 949–964 (2022).

Article MATH Google Scholar

Khan, M. A., Nasralla, M. M., Umar, M. M., Khan, S. & Choudhury, N. An efficient multilevel probabilistic model for abnormal traffic detection in wireless sensor networks. Sensors 22(2), 410 (2022).

Article ADS PubMed PubMed Central MATH Google Scholar

Moussa, N., Hamidi-Alaoui, Z., Alaoui, E. B. E. & A ECRP: an energy-aware cluster-based routing protocol for wireless sensor networks. Wireless Netw. 26, 2915–2928 (2020).

Article MATH Google Scholar

Li, L. & Han, C. ASARIMA: an adaptive harvested power prediction model for solar energy harvesting sensor networks. Electronics 11(18), 2934 (2022).

Article MATH Google Scholar

Surenther, I., Sridhar, K. P. & Roberts, M. K. Maximizing energy efficiency in wireless sensor networks for data transmission: A deep learning-based grouping model approach. Alexandria Eng. J. 83, 53–65 (2023).

Article MATH Google Scholar

Zhang, L., Zhang, C., Wei, K. & Zhang, Q. Link fault repair algorithm of wearable wireless sensor networks based on polygon fermat point. In 2020 IEEE International Conference on E-health Networking, Application & Services (HEALTHCOM) (pp. 1–5). IEEE. (2021).

Gao, J., Wu, R., Hao, J., Xu, C. & Guo, H. SWIPT-based energy scheduling for solar-powered WSN in full-duplex mode. IEEE Sens. J. 22(13), 13668–13681 (2022).

Article ADS Google Scholar

Li, H., Liu, X., Huang, M., Chen, Q. & Chen, L. ICCO: immune clustering-based coverage optimisation algorithm in wireless sensor networks. Int. J. Internet Protoc. Technol. 15(2), 76–93 (2022).

Article MATH Google Scholar

Zhao, B. & Zhao, X. Deep reinforcement learning resource allocation in wireless sensor networks with energy harvesting and relay. IEEE Internet Things J. 9(3), 2330–2345 (2021).

Article MATH Google Scholar

Roberts, M. K. & Thangavel, J. An optimized ticket manager based energy-aware multipath routing protocol design for IoT based wireless sensor networks. Concurrency Computation: Pract. Experience, 34(28), e7398. (2022).

Menaria, V. K., Nayyar, A., Kumar, S. & Kotecha, K. Multiway Relay based Framework for Network Coding in Multi-hop WSNs. Computers Mater. Continua 74(1). (2023).

Mohankumar, B. & Karuppasamy, K. Network lifetime improved optimal routing in wireless sensor network environment. Wireless Pers. Commun. 117(4), 3449–3468 (2021).

Article MATH Google Scholar

Tang, Z., Chen, X., Fang, D. & Chen, F. Research on java software protection with the obfuscation in identifier renaming. In 2009 Fourth International Conference on Innovative Computing, Information and Control (ICICIC) (pp. 1067–1071). IEEE. (2009), December.

Cao, L., Cai, Y. & Yue, Y. Swarm intelligence-based performance optimization for mobile wireless sensor networks: Survey, challenges, and future directions. IEEE Access. 7, 161524–161553 (2019).

Article MATH Google Scholar

Shyama, M., Pillai, A. S. & Anpalagan, A. Self-healing and optimal fault tolerant routing in wireless sensor networks using genetical swarm optimization. Comput. Netw. 217, 109359 (2022).

Article Google Scholar

Ghawy, M. Z., Amran, G. A., AlSalman, H., Ghaleb, E., Khan, J., Al-Bakhrani, A. A.,… Ullah, S. S. (2022). An effective wireless sensor network routing protocol based on particle swarm optimization algorithm. Wirel. Commun. Mobile Comput., (2022).

Li, J. Q., Liu, Z. M., Li, C. & Zheng, Z. X. Improved artificial immune system algorithm for type-2 fuzzy flexible job shop scheduling problem. IEEE Trans. Fuzzy Syst. 29(11), 3234–3248 (2020).

Article MATH Google Scholar

Pinto, R. & Gonçalves, G. Application of artificial immune systems in advanced manufacturing. Array, 100238. (2022).

Tsoumanis, G., Oikonomou, K., Aïssa, S. & Stavrakakis, I. Energy and distance optimization in rechargeable wireless sensor networks. IEEE Trans. Green. Commun. Netw. 5(1), 378–391 (2020).

Article MATH Google Scholar

Eiz-Vesper, B. & Schmetzer, H. M. Antigen-presenting cells: Potential of proven und new players in immune therapies. Transfus. Med. Hemother. 47(6), 429–431 (2020).

Article Google Scholar

Zhang, Y., Wang, K. & Zhang, J. Adaptive and lightweight abnormal node detection via biological immune game in mobile multimedia networks. Algorithms 14(12), 368 (2021).

Article MATH Google Scholar

Banerjee, S. Lymph node inspired computing: towards immune system inspired human-engineered complex systems. Interdisciplinary Description Complex. Systems: INDECS 19(4), 502–510 (2021).

Article MATH Google Scholar

Lindsay, A. D., Gaston, D. R., Permann, C. J., Miller, J. M., Andrš, D., Slaughter, A. E., Wong, C. (2022). 2.0-MOOSE: Enabling massively parallel multiphysics simulation. SoftwareX, 20, 101202.

Velmurugan, P. G. S. et al. Performance analysis of full duplex bidirectional machine type communication system using IRS with discrete phase shifter. Appl. Sci. 13(12), 7128 (2023).

Article CAS MATH Google Scholar

Chiwariro, R. & N, T Quality of service aware routing protocols in wireless multimedia sensor networks: Survey. Int. J. Inform. Technol. 14(2), 789–800 (2022).

MATH Google Scholar

Zhao, H. et al. Intelligent Computing collaboration for the security of the fog internet of things. Symmetry 15(5), 974 (2023).

Article ADS MATH Google Scholar

Liu, S., Zhang, D., Liu, X., Zhang, T. & Wu, H. Adaptive repair algorithm for TORA routing protocol based on flood control strategy. Comput. Commun. 151, 437–448 (2020).

Article MATH Google Scholar

Zhang, W., Zhang, W., Yen, G. G. & Jing, H. A cluster-based clonal selection algorithm for optimization in dynamic environment. Swarm Evol. Comput. 50, 100454 (2019).

Article MATH Google Scholar

Download references

Faculty of Engineering, University of Malaya, 50603, Kuala Lumpur, Malaysia

Ruipeng Tang, Narendra Kumar Aridas & Mohamad Sofian Abu Talip

Faculty of Education, Nanchang Normal University, Nanchang, 330032, China

Yinhe Wu

Faculty of Business, Law, Communication and AC, INTI International University, 71800, Nilai, Negeri Sembilan, Malaysia

Jun Tan

Faculty of Business and Economics, University of Malaya, 50603, Kuala Lumpur, Malaysia

Binghong Guan

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

Author contribution statement: Ruipeng Tang: Conceptualization, Methodology, Software, Validation, Formal Analysis, Investigation, Data Curation, Writing—Original Draft, Writing – Review & Editing, Visualization. Yinhe Wu: Conceptualization, Methodology, Validation, Formal Analysis, Writing—Review & Editing. Jun Tan: Software, Investigation, Data Curation, Visualization, Writing—Original Draft, Supervision,. Binghong Guan: Writing—Review & Editing, Investigation. Narendra Kumar Aridas : Writing—Review & Editing, Project Administration. Mohamad Sofian Abu Talip: Project Administration, Resources.Note: All the above authors agree to be responsible for the content and conclusions of the article.

Correspondence to Ruipeng Tang.

The authors declare no competing interests.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

Tang, R., Wu, Y., Tan, J. et al. Research on rechargeable agricultural wireless sensor network based on ZigBee immune routing repair algorithm. Sci Rep 15, 5756 (2025). https://doi.org/10.1038/s41598-025-89710-w

Download citation

Received: 01 May 2024

Accepted: 07 February 2025

Published: 17 February 2025

DOI: https://doi.org/10.1038/s41598-025-89710-w

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative