Fusion of Big Data Analytics, Machine Learning and Optimization Algorithms for Internet of ThingsView this Special Issue
Elite Oppositional Farmland Fertility Optimization Based Node Localization Technique for Wireless Networks
Wireless networks include a set of nodes which are connected to one another via wireless links for communication purposes. Wireless sensor networks (WSN) are a type of wireless network, which utilizes sensor nodes to collect and communicate data. Node localization is a challenging problem in WSN which intends to determine the geographical coordinates of the sensors in WSN. It can be considered an optimization problem and can be addressed via metaheuristic algorithms. This study introduces an elite oppositional farmland fertility optimization-based node localization method for radio communication networks, called EOFFO-NLWN technique. It is the goal of the proposed EOFFO-NLWN technique to locate unknown nodes in the network by using anchor nodes as a starting point. As a result of merging the principles of elite oppositional-based learning (EOBL) and the agricultural fertility optimization algorithm (FFO), we have developed the EOFFO-NLWN approach, which is described in detail below. The EOBL concept makes it easier to populate the FFO algorithm’s population initialization, which results in an increase in the exploration rate. Various BNs and CRs were tested, and the findings revealed that the EOFFO-NLWN technique outperformed all other known techniques in all cases. A comprehensive experimental result analysis of the EOFFO-NLWN technique is performed under several measures, and the results described the sovereignty of the EOFFO-NLWN method associated to existing techniques.
As an emergent model of computing and networking, wireless sensor network (WSN) has been applicable and relevant in different domains like military, medicine, climate forecasting, surveillance , environmental control, and so on. Reliable advances and development in networks have considerably enabled and extended wide-ranging applications of WSN. In recent times, WSN has been incorporated with another concept includes internet of things (IoT) . Wireless communications and electronics have advanced significantly in recent years, enabling the expansion of multifunctional devices that are low in cost and power ingesting, and that can communicate over relatively short distances. Sensors that are inexpensive, intelligent, wirelessly interacted, and widely scattered open the door to possibilities for monitoring and regulating homes, communities, and the natural environment that were previously unimaginable. Furthermore, networked sensors provide a plethora of defence applications, enabling for the development of new capabilities in reconnaissance, surveillance, and a variety of other tactical applications. A feature that is highly desirable in wireless sensor networks is their ability to self-locate. Wireless sensor networks are utilized in a variety of environmental applications to perform a variety of activities, including environmental monitoring, disaster response, target tracking, and defences. In military applications, such as battlefield surveillance, wireless sensor networks were developed and are now used in a wide range of industrial and civilian applications, including industrial process monitoring and control, machine health monitoring, environment and habitat monitoring, healthcare applications, home automation, and traffic control. Wireless sensor networks are also used in military applications, such as battlefield surveillance. It is critical to build efficient localization solutions since the vast majority of applications rely on proper localization, i.e., computing their positions in a particular coordinate system. A WSN is a network substructure that contains massive amount of diminutive, minuscule, low-cost autonomous devices represented as sensors that detect and monitor the environments for compiling information . The information gathered from the framework is later transmitted to the sink nodes, a destination whereby information is redirected or treated locally to another network for dissimilar usages . Because of the node communication, accessible deployment, self-organization, and data transfer, WSN has several usage and advances, but they confront few problems . Figure 1 depicts the structure of WSN.
There are several problems in WSN execution processes, like coverage, node localization (NL), data routing issues, energy consumption of sensor nodes, and so on . Data routing issues: even though a directing procedure should be energy-efficient, load-balancing, and fault-tolerant in addition to being scalable and providing a high level of safety, this is, to put it mildly, a difficult challenge to accomplish. Sensor nodes make use of energy in the following ways: the fundamental role of a sensor node as a microelectronic device is to detect events, do data processing on the fly and locally, and transmit and receive data. Sensor nodes are made up of four parts: a power supply unit, a sensing unit, a computing/processing unit, and an interactive unit. The sensing node is typically powered by a limited-capacity, nonrechargeable battery in the vast majority of application scenarios. All other components, with the exception of the power unit, will require energy in order to perform their functions. Node localization (NL): nodes are located in the following locations: because of the use of localization techniques, the deployment of WSNs is quite inexpensive. An anchor or beacon node that is aware of its location is used by the vast majority of localization algorithms. Using the position data provided by the anchor node or beacon node, the other nodes can identify their own location and that of their neighbours. In spite of each challenge and issue, the more important one is defining the position of sensors. The technique of NL could track and locate nodes; thus, the data monitoring is very beneficial that is information collected at sink node would be valueless to the client with no localizing information of the node in the sensor region . The localization can be described as location of the unknown sensors named as target node (TN) with the known location of the sensors named as anchor node (AN) depending on the measurements including period of influx, period change of influx, maximal likelihood, and angle of arrival, triangulation . The localization problem of WSN can be addressed by employing global positioning system (GPS) with sensors; however, this is not favourable as a consequence of cost, size, and energy problems. It even does not function appropriately underwater and indoor . WSNs use localization to locate sensor nodes. Installation GPS on each WSN node is pricey, because GPS does not perform well inside. In a dense network, manually referencing each sensor node’s position is impractical. In this situation, the sensor nodes must self-locate without GPS or manual configuration. Implementing WSN with localisation saves money. An anchor or beacon node understands its present location. Consequently, better and efficient alternative is needed for localizing the sensors. Different non-GPS-based localizing approaches are utilized that are divided into range-based and range-free models . To control the present position of instrument nodes in wireless sensor networks (WSNs), localization is a technique that is widely utilized. An indoor WSN can include hundreds or even thousands of nodes, making the connection of GPS on each device node excessively expensive. Additionally, GPS will not offer precise location findings in an indoor environment. Node localization is extremely important in order to locate and determine the location of sensor nodes with the use of a particular procedure. As previously stated, localization is the process of decisive the geographic position of nodes , because data and information are rendered meaningless if the nodes do not know where they are in relation to one another. Several metaheuristic approaches have been working for solving the localization problematic in WSN that dramatically minimizes the localization fault.
This study introduces an elite oppositional farmland fertility optimization-based node localization technique for radio systems, called EOFFO-NLWN technique. The proposed EOFFO-NLWN technique mainly intends for classifying the position of unidentified bulges from the network using ANs. Besides, the EOFFO-NLWN technique is derived by the combination of the concepts of elite oppositional based learning (EOBL) and farmland fertility optimization (FFO) algorithm. The EOBL concept assists in the population initialization of the FFO algorithm and thereby improves the exploration rate. To promote population diversity, the first step is to use elite opposite-based learning (EOBL). The second enhancement is the integration of three innovative local search procedures to avoid becoming caught in local optima. This is the fundamental goal of EOBL: to turn existing search results into more relevant ones. Through the investigation of solutions in both the current search space and the altered search space, EOBL can boost the possibility of identifying explanations that are closer to the global optimum than they would otherwise be. The procedure delivers chaotic disruption into the population at the same time as it works to increase population variety. A comprehensive experimental result analysis of the EOFFO-NLWN technique is performed under several measures.
1.1. Related Works
According to Phoemphon and colleagues , leveraging AN from a nearby group to approximation the locations of unidentified nodes improves localization precision by a factor of ten. It is also possible to execute PSO with increased FF in order to predict the positions of unknown nodes. The effectiveness of localisation was evaluated in depth in environments that were prone to obstacles. An algorithm for node placement that relies on the Voronoi diagram and SVM was described in  for use in this scenario. The fundamental goal of the technique, which makes use of a Voronoi diagram and an AN retrieved from the localization region, is to initially segment the district into a great number of parts. Once all regions had been utilized to place the TN in, the SVM was used to determine the most optimal precise location for the TN’s primary location.  proposes a network-level strategy for WSN that is based on virtual partition and distance correction (VP-DC) techniques.
The Monte Carlo node placement technique developed by Song et al.  is based on the enhanced QUASI-Affine Transformation Evolutionary (QUATRE) algorithm and is described in detail below. After selecting the best general nodes within one hop of unidentified nodes as provisional ANs, and using the temporary AN and AN as orientation nodes to construct a more accurate sampling region, an improved QUATRE-optimized technique was used to obtain the evaluated position of unidentified nodes from the sampling area. QUATRE-optimized techniques were also used to obtain the evaluated position of unknown nodes from the sampling region. By employing the GSO combined localization method, Yu et al.  develop novel localization algorithms for wireless sensor networks. The primary function was derived analytically using a 3D localization approach and the Pareto distance as a starting point. The glow-worm sets were created by dividing the GSO population into their upgrade locations in order to improve the precision with which the swarms might be located.
In device systems, Shayanfar et al.  suggesed that the localization can be labelled as the procedure of decisive the location of a sensor node. Any wireless sensor network’s localization mechanism must be highly accurate. Localization is the procedure of determining the sensor node’s geometrical location inside the network. The localization challenge entails decisive the position and coordination of wireless sensor nodes. Localization is a problem that has been explored for many years when it comes to wireless sensor nodes. There are numerous alternatives, which are evaluated based on their cost, size, and energy usage. Localization is critical when the precise location of some permanent or mobile equipment is unknown. One example is the monitoring of humidity and temperature in woods and/or fields, where hundreds of sensors are dropped from a plane, with the operator having little or no control over the precise location of each node.
The remaining sections of this proposed work are structured as follows. In Section 2, fundamental measurement techniques for localization in WSNs are briefly presented along with their usual problems and obstacles. Section 3 discusses various localization algorithms and their comparative analysis. Section discusses numerous localization evaluation criteria. Then, in Section 4, we discuss perspectives and issues in range-free localization methods in conclusion.
2. The Proposed Model
This study has developed an effective EOFFO-NLWN technique to recognize the position of unidentified bulges in the network using ANs. The EOFFO-NLWN technique is majorly derived by the combination of the concepts of EOBL and FFO algorithm. The EOBL concept assists in the population initialization of the FFO algorithm and thereby improves the exploration rate.
2.1. Design of EOFFO Algorithm
Metaheuristics are a type of model-free technique to resolve different kinds of optimized problems which are existing employed in a wide range of applications . Elite Oppositional Farmland Fertility Optimization (EOFFO), the optimization strategy for MOIFF is based on a novel bioinspired meta experiential technique called Farmland Fertility algorithm (FF), which was proposed in 2018. Since FF has been shown to outperform a variety of well-known metaheuristic approaches (including GA, DE, PSO, and ABC) in relations of meeting accurateness, constancy, and speed. Metaheuristics are a kind of modeless technique used to solve various types of optimal problems in a wide variety of applications. The FFO technique contains 6 important parts that are described here. During initiation procedure, the quantity of sections and the possible solution to them () under the farmland are determined. For that purpose, the population () of technique was modelled as follows: where refers the positive digit from the range of 1 and , and defines the integer number. The rate to during this learning was chosen 2 that is attained by trial and error. For making the primary individuals from the possible range, the subsequent formula was adapted: where and represent the lower and upper limits of dimension and represent an arbitrary rate between zero and one. The farmland in the technique was divided into three portions of local memory (A, B, and C) and global memory, with section A containing the lowest grade soil.
The estimation stage directs the farmland decision variable from the section. To evaluate the rate purpose worth to the choice variable. Also, the excellence of soil was attained by the subsequent:
During memory update stage, the resident as well as international memories were upgraded. An optimum solution of farmlands is kept from the limited memory, and explanations among them are regarded as total memory. For determining the amount of optimum local as well as total recollections, the subsequent formulas were utilized : where , and and define the amount of saving solutions from local as well as global memory correspondingly. To define the quality of sections and save an optimum one from the local memory. Also, optimum solutions are kept from the global memory. For increasing the worse-case outcomes, it can be upgraded by relating them with optimum-case solution of global memory. At last, the variable of novel solution is upgraded as where portrays an arbitrary value with global solutions, refers the worse case that is chosen to upgrade, and defines the decimal number as follows: where represents the constant value from the range of zero and one, and implies the arbitrary value from the range of -1 and 1. For updating another solution, where implies the arbitrary value from the range of zero and one, and defines the constants from the range of zero and one, that is assumed as the start of farmland fertilities.
After determining optimum local solutions , the farmland optimum soil combination was chosen by agriculturalist. But, an optimum global solution is obtained for combining by the farmland for developing the quality of soils. This stage was mathematically processed by the subsequent: where demonstrates the optimum worldwide mixture to the solution and is a continuous from the choice of zero and one defines the random standards from the range of zero and one, and implies the limit of country fecundities which is determined at starting time and is expressed as follows:
Estimate the feasible solution to search spaces. In this procedure, when the end condition is attained, this technique ends, else, it can be continuous still attaining optimum solutions. EOBL is a novel approach used to enhance the performance of metaheuristics . Let elite individual in present population is , for an individual , the elite opposition solution of can be represented using the following equation: where , denotes population size, and implies generalized coefficient, and is an adaptive limit of dimension searching area and can be attained as follows.
The static margin is nonconducive in storing the searching experience, and therefore, adaptive bound is used for replacing the fixed bounds in preserving the searching experience for making narrower opposition solutions. Besides, when operator of dynamic bound creates jumps out of , Eq. (15) is applied for resetting :
The EOBL produces opposition population based on elite individuals and assesses the present as well as elite population concurrently. Also, it completely utilizes the features of elite individuals to comprise meaningful searching data compared to normal individuals. Besides, the EOBL helps to boost the global exploration abilities of the FFO algorithm. The recent application of EOBL is OBL is mostly used in bioinformatics and medicine, specifically for illness detection and forecast in particles and proteins, as well as for drug discovery and development. The core OBL concept and the quasi-inverse are the two OBL schemes that are most frequently used in a wide range of request industries. It begins with a thorough examination of the basic OBL idea before going on to describe the many OBL schemes and precise propositions that are employed in machine learning methods. It then goes into detail on the adjustments to OBL’s usage in reinforcement learning, artificial neural networks, fuzzy systems, and variant optimization techniques, among other requests. It provides a succinct overview of the majority of OBL’s diverse range of applications. A comprehensive overview of OBL investigates from a variety of angles. In some cases, exact proofs and theoretical definitions for investigating and utilising the benefits of OBL are presented, while other emphasis on special growths for various schemes of incorporating OBL into machine learning methods is presented, and still other attention on the numerous applications of OBL in several science and engineering grounds, such as power schemes, pattern credit and image processing (including facial recognition and image processing), documentation problems (including bioinformatics), and drug, among other field.
2.2. Steps Involved in EOFFO-NLWN Technique
The EOFFO-NLWN approach includes the subsequent phases for localizing the sensor from WSN.
Place AN and TN arbitrarily in the device part. Each AN was place aware and assist to recognize the residence of additional node . Each target and ANs include communication range . (1)Distance among the AN and TN are evaluated and altered with preservative Gaussian noise. The TN controls the distance as whereas represent to the actual distance, i.e., computed among the place of TN and place of beacon :
The variable controls the noise that touches the assessed distance in whereas denotes the sound relation from the predictable distance. (2)The desirable node is named a localization node when it has three ANs within the transmission radius of TNs [20–24](3)In case of localized nodes, the EOFFO-NLWN technique was implemented separately to recognize the place of TNs. The EOFFO-NLWN method is performed by applying the centroid of AN inside a transmission radius:
In which indicates the complete quantity of ANs within the transmission series of restricting TNs [21, 25–28]. (4)The EOFFO-NLWN approach is suitable to identify the coordinate as TN which reduces the localizing fault. The primitives implemented in localization issue are a mean four-sided detachment between the anchor and TNs:
Whereas represents the quantity of ANs confidential a broadcasting radius of TNs. (5)Once the highest quantity of repetitions is accomplished, afterward, the optimal location coordination is determined by EOFFO-NLWN model [29–31]
A whole localizing error can be described after that computing the localizing TN . It is estimated as a mean four-sided of coldness in node coordinate from the real node coordinate: (6)Steps 2 to 6 are iterated till the TN is localized. The localized method is depending on the high mistake contained , and quantity of unlocalized bulges is strongminded by . The least scores of and show an effective restricted method [32–34]
3. Performance Validation
The localization result analysis of the EOFFO-NLWN model takes place in this section. Table 1 and Figure 2 demonstrate the comparative ALE examination of the EOFFO-NLWN model with other approaches below distinct beacon bulges (BNs). The results portrayed that the EOFFO-NLWN model has attained reduced ALE values under all BNs. For instance, with 5 BNs, the EOFFO-NLWN model has obtained lower ALE of 6.95% whereas the DV-Hop, WND-DV-Hop, MGDV-Hop, and VP-DC techniques have attained higher ALE of 61.39%, 42.72%, 65.15%, and 14.49%, respectively. Simultaneously, with 20 BNs, the EOFFO-NLWN model has provided minimal ALE of 5.98% whereas the DV-Hop, WND-DV-Hop, MGDV-Hop, and VP-DC techniques have resulted to maximum ALE of 60.32%, 34.83%, 21.83%, and 10.71%, respectively. Concurrently, with 35 BNs, the EOFFO-NLWN model has gained least ALE of 3% whereas the DV-Hop, WND-DV-Hop, MGDV-Hop, and VP-DC techniques have accomplished increased ALE of 45.47%, 24.36%, 12.95%, and 6.70%, respectively.
Table 2 and Figure 3 depict the proportional LNT examination of the EOFFO-NLWN model with other methods under different BNs. The results portrayed that the EOFFO-NLWN technique has reached lower LNT values under all BNs. For instance, with 5 BNs, the EOFFO-NLWN technique has obtained lesser LNT of 0.169 min whereas the DV-Hop, WND-DV-Hop, MGDV-Hop, and VP-DC techniques have attained superior LNT of 0.344 min, 0.755 min, 2.703 min, and 0.207 min correspondingly. Concurrently, with 20 BNs, the EOFFO-NLWN methodology has provided minimal LNT of 0.169 min whereas the DV-Hop, WND-DV-Hop, MGDV-Hop, and VP-DC techniques have resulted in increased LNT of 0.336 min, 0.656 min, 2.163 min, and 0.306 min correspondingly. Concurrently, with 35 BNs, the EOFFO-NLWN system has gained least LNT of 0.192 min but the DV-Hop, WND-DV-Hop, MGDV-Hop, and VP-DC algorithms have accomplished maximum LNT of 0.367 min, 0.648 min, 2.087 min, and 0.344 min correspondingly.
Table 3 and Figure 4 demonstrate the comparative ALE examination of the EOFFO-NLWN model with other approaches below distinct communication radius (CR). The results portrayed that the EOFFO-NLWN model has reached lesser ALE values under all CR. For instance, with 5 m CR, the EOFFO-NLWN methodology has obtained lesser ALE of 5.18 min while the DV-Hop, WND-DV-Hop, MGDV-Hop, and VP-DC methodologies have attained superior ALE of 64.39 min, 46.96 min, 23.38 min, and 13.89 min correspondingly. Concurrently, with 20 m CR, the EOFFO-NLWN system has provided minimum ALE of 5.95 min whereas the DV-Hop, WND-DV-Hop, MGDV-Hop, and VP-DC techniques have resulted in increased ALE of 53.88 min, 29.27 min, 24.15 min, and 10.56 min, respectively. Concurrently, with 35 m CR, the EOFFO-NLWN technique has reached least ALE of 2.87 min whereas the DV-Hop, WND-DV-Hop, MGDV-Hop, and VP-DC methodologies have accomplished higher ALE of 46.96 min, 25.68 min, 17.225 min, and 8 min correspondingly.
Table 4 and Figure 5 demonstrate the comparative LNT examination of the EOFFO-NLWN method with other methods below separate CR. The consequences portrayed that the EOFFO-NLWN method has attained reduced LNT values under all BNs. For instance, with 5 m CR, the EOFFO-NLWN methodology has reached lesser LNT of 0.245 min whereas the DV-Hop, WND-DV-Hop, MGDV-Hop, and VP-DC techniques have reached increased LNT of 0.598 min, 1.0835 min, 2.260 min, and 0.337 min correspondingly. Simultaneously, with 20 m CR, the EOFFO-NLWN technique has offered lesser LNT of 0.206 min whereas the DV-Hop, WND-DV-Hop, MGDV-Hop, and VP-DC systems have resulted in maximal LNT of 0.391 min, 0.668 min, 2.175 min, and 0.329 min correspondingly. At last, with 35 m CR, the EOFFO-NLWN system has gained least LNT of 0.183 min whereas the DV-Hop, WND-DV-Hop, MGDV-Hop, and VP-DC algorithms have accomplished increased LNT of 0.329 min, 0.197 min, 2.191 min, and 0.260 min correspondingly.
As previously stated, the EOFFO-NLWN method outdone all other currently available methods for a variety of BNs and CRs, as proven by the findings presented here.
3.1. Limitations of Proposed Systems
Elite oppositional farmland fertility optimization-based node localization method for radio communication networks (EOFFO-NLWN) measuring the sending time of the transmitting signal and the receiving time of the signals are delayed. Increasing sensor density presents several difficulties for localisation. One such difficulty is information loss due to wireless signal collision.
This study has developed an effective EOFFO-NLWN approach to recognize the position of unidentified bulges in the network using ANs. The EOFFO-NLWN technique is majorly derived by the combination of the concepts of EOBL and FFO algorithm. The EOBL concept assists in the population initialization of the FFO algorithm and thereby improves the exploration rate. A comprehensive experimental results examination of the EOFFO-NLWN method is achieved under several measures, and the consequences described the supremacy of the EOFFO-NLWN procedure associated to existing methods. It is proposed in this study that the EOFFO-NLWN technique, which is an elite oppositional farmland fertility optimization-based bulge localization method for wireless networks, be used in future research. It was proved through the experimental results that the EOFFO-NLWN strategy outperforms all other known methods when exposed to various BN and CR conditions. Therefore, the EOFFO-NLWN method can be practical as an actual instrument for restricting nodes in WSN. As a portion of upcoming possibility, the network presentation can be increased by the usage of hybrid metaheuristics-based clustering schemes. A new hybrid optimization technique for finding the ideal CH while taking into account all aspects such as latency, distance, and energy in order to enhance the network’s lifetime. Due to a variety of complex circumstances, effective data transfer among nodes is nearly impossible. Clustering is a well-known technique for improving the efficiency of data transmission. The clustering model separates the sensor nodes into several clusters.
The manuscript contains all the data.
Conflicts of Interest
The corresponding author declares that there is no conflict of interest on the part of any of the other authors, including themselves.
T. T. Nguyen, T. W. Sung, D. T. Pham, T. G. Ngo, and V. D. Vu, “A hybrid swarm evolution optimization for solving sensornet node localization,” Intelligent Systems and Networks. ICISN 2021. Lecture Notes in Networks and Systems, Springer, Singapore, 2021.View at: Google Scholar
Z. Jiang, W. Hu, and H. Qin, “WSN node localization based on improved sparrow search algorithm optimization,” In international conference on sensors and instruments (ICSI 2021), International Society for Optics and Photonics, 2021.View at: Google Scholar
J. Yu, T. T. T. Nguyen, T. K. Dao, and T. G. Ngo, “A solution to sensor node localization using glow-worm swarm optimization hybridizing positioning model,” Advances in Intelligent Information Hiding and Multimedia Signal Processing, Springer, Singapore, pp. 260–268, 2021.View at: Publisher Site | Google Scholar
C. P. D. Cyril, J. R. Beulah, P. Mohan, A. Harshavardhan, and D. Sivabalaselvamani, “An automated learning model for sentiment analysis and data classification of Twitter data using balanced CA-SVM,” Concurrent Engineering Research and Applications, vol. 29, no. 4, pp. 386–395, 2021.View at: Publisher Site | Google Scholar
A. Priyanka, “BIG data based on healthcare analysis using IOT devices,” IOP Conference Series: Material and Science Engineering, vol. 263, no. 4, 2017.View at: Google Scholar
M. E. Kumar and G. T. Reddy, “Vehicle theft identification and intimation using GSM & IOT,” IOP Conference Series: Material and Science Engineering, vol. 263, no. 4, 2017.View at: Google Scholar
R. Kaluri and S. Karnam, “A comparative study on selecting and ranking the test cases in software testing,” ARPN Journal of Engineering and Applied Sciences, vol. 11, no. 1, pp. 754–757, 2016.View at: Google Scholar