An In-Depth Examination of Localization Strategies Employed in Wireless Sensor Networks (WSNs): Methods, Challenges and Prospective Developments

Abstract

Wireless Sensor Networks (WSNs) rely heavily on localization to provide location aware services for applications including military surveillance, smart agriculture, environmental monitoring and healthcare. Morden methods that combine range-based and range-free techniques including Time of Arrival (ToA), Received Signal Strength Indicator (RSSI) and hybrid approaches have greatly increased the localization accuracy. Furthermore, machine learning based models with improved adaptability in dynamic situations incorporate: Convolutional Neural Networks (CNNs) and Artificial Neural Networks (ANN). Despite such developments, several challenges and obstacles still exist. In case of complicated terrains, environmental obstacles conjointly with multipath fading and signal interference curtail the localization accuracy. Further, the advanced techniques like Ultra-Wide Band (UWB) and directional antennas in networks with limited resources get hampered by high energy consumption and escalated hardware costs. Additionally, the lack of standard models for real time localization makes the system design even more difficult since sensor node mobility and dynamic topologies compromise the accuracy of the conventional methods. In this context, localization strategy is also seriously threatened by security issues such as spoofing and data manipulation. The current paper provides a thorough analysis of various current localization strategies employed in WSNs, thereafter classified them as machine learning based, range based, range free and hybrid approaches. The objective is to highlight the serious issues associated with the existing systems and to provide possible design suggestions for developing precise, safe and energy efficient localization frameworks. The findings of the current work are meant to expedite the future investigations more towards scalable, reliable, and contextually aware localization technologies appropriate for novel applications in the Internet of Things (IoT) and smart environments while considering both cost and security constraints.

Introduction

In recent days, sensor nodes are extensively used in Wireless Sensor Networks (WSNs) to attain information about their concerned environment, however, such information is useless unless the precise location from where it was collected is known. Many applications including: environmental monitoring, vehicle tracking with mapping, emergency response depends on WSNs as one of their main enablers. In this context, sensors are sprinkled in realistic environments for applications such as habitat monitoring, target tracking, battlefield surveillance, etc., where position of each sensor becomes crucial. However, in WSNs, localization still poses as a difficult challenge [1]. Each sensor node receives a location reference at the start of distribution either manually or mechanically. Further, depending on the attached GPS devices of sensor nodes; the respective distance is calculated. Because of the high cost and labor requirements in the setting of a vast network, it might not be possible to manually calculate the position or to install a possible GPS device [2, 3, 4–5].

In this context, distributed localization entails nodes that exchange information and work together to determine their locations, while on the contrary, centralized localization depends on the mobile nodes that send data to a central node, which subsequently determines their respective positions. Likewise, decentralized localization algorithms are more accurate and efficient than centralized ones, especially when there are several mobile nodes involved [3, 6, 7, 8, 9–10]. Through direct communication with anchor nodes, distributed systems let each sensor to detect and estimate its own position, while centralized algorithms demand cheaper operational expenses and waste less amount of energy while running on a central system as opposed to individual nodes [11, 12, 13, 14–15]. In order to get around this problem, sensor nodes are designed to use nearby nodes to pinpoint their exact positions [4].

The Locality Reference of the sensor nodes is utilized by the Location Aided Routing (LAR) protocol to obtain the quickest path. Robots are employed in the military to find out land mines, and certain enterprises that utilize sensor nodes to detect minute changes like pressure, temperature, and gas leaks. In both of these instances, location information is currently becoming crucial [16, 17, 18, 19–20]. Sensors with GPS receivers or those who are aware of their positions beforehand are considered as anchor nodes [5, 21, 22, 23, 24–25]. Localization has become an important problem in WSNs since sensor nodes’ locations are essential for both the network operation and application-specific uses [26, 27, 28, 29–30]. In this context, anchor locations are used for multilateration in converting relative coordinates to absolute coordinates and for adding constraints to methods, built on mathematical programming. To maintain the accuracy of the solution, distributed algorithms built on multilateration need a substantial number of anchors. Anchors, on the other hand are much more expensive than a typical gauge [6]. The performance of techniques based on multilateration is also impacted by the placement of anchors. Getting the longitude, latitude, and altitude of the sensor node’s location involves proper data collection [31, 32, 33, 34, 35, 36–37].

There are various types of localization, including topological, geometric, relative, and local. These localization classifications are predicated on the viewpoint of robotic or mobile agents. These ideas are essential to how any autonomous system, including robots, locates itself in its surroundings. Localization basically establishes the location of a point with respect to a local reference point. The position of an object in respect to other objects is referred to as relative localization. There are two legitimate approaches for localization: topological and geometric [38, 39]. While topological localization relies on linking to landmarks, geometric localization combines both distance and angle to pinpoint an object’s location [7]. In this connection, Geometric and topological location descriptions are too valid methods [38, 39]. Geometric method allows for the use of both angle and distance to pinpoint an object’s placement [7] and through linking to the landmarks; the topological location gets determined.

WSNs are essential for many applications such as healthcare, industrial automation and environmental monitoring etc. WSNs consist of large number of sensor nodes dispersed over a region for gathering and transmitting data to monitor environmental phenomena such as temperature, humidity, pressure, mobility and so forth. Despite extensive usages and numerous advantages, WSNs involve effective localization of sensor nodes [8]. In this connection, localization in a WSN refers to the process of identifying the spatial position of individual nodes within the network [40, 41–42]. This is crucial as the physical locations of nodes directly influence the accuracy and relevance of data that they collect and transmit [9]. Localization is more significant than simply making sure that every node is aware of its location. It serves as the foundation for many other vital functions within the network such as data routing, energy management, fault detection and for the performance of many advanced applications including tracking, monitoring and surveillance [10, 43].

Without accurate localization, a WSN would be inefficient, prone to errors and unable to perform in real-time which may lead to reduced quality of service. Therefore, maximizing the performance of WSNs requires a thorough grasp of the techniques used to localize nodes in these networks [44, 45–46]. In WSNs, nodes are typically deployed in large numbers over potentially vast and sometimes inaccessible regions [11, 47, 48, 49–50]. They are often of low cost, less power and smaller in size which poses as a unique challenge for localization [51, 52–53]. Due to these constraints, the traditional Global Positioning System (GPS) is not a viable solution for localizing sensor nodes, especially in large-scale or dense networks [54, 55, 56, 57–58]. Therefore, several alternative localization techniques have been developed to meet the unique demands of WSNs. These methods vary in terms of accuracy, complexity, scalability, and resource consumption [12]. Hybrid localization approaches have been more popular in recent years because they combine several measuring methods or algorithms to increase accuracy and robustness [59, 60, 61, 62, 63, 64–65]. For example, by utilizing time-based precision, RSSI combined with TDoA can overcome the constraints of signal intensity change [1]. Similarly, improved spatial resolution is provided by combining fingerprinting techniques with AoA, particularly in cluttered or multipath situations. In dynamic WSNs, where environmental considerations like noise, fading, and mobility need adaptive and fault-tolerant systems; hybrid techniques are especially advantageous.

Concurrently, machine learning-based translation has become a game-changing method, particularly in unpredictable and changing environments. Artificial Neural Networks (ANNs), Random Forests, and Support Vector Machines (SVMs) are some of the methods used to learn from past signal patterns and environmental data to provide highly accurate position predictions. For instance, labelled RSSI data can be used to train a supervised learning model that can locate a mobile node in real-time and modify predictions in response to changing environmental variables [2]. To extract spatial information from complex signal distributions, deep learning techniques like Convolutional Neural Networks (CNNs) have been applied in indoor WSN localization settings [3]. Hybrid and machine learning-based approaches help WSN localization systems to become more scalable, context-aware, and able to function well in both dynamic and real-world settings.

Literature Review

Significant progress has been achieved in localization methods for WSNs in recent years with the goal of enhancing precision, energy economy, and environmental adaptability. From the conventional techniques like ToA, AoA and RSSI to more complex models, researchers have investigated and discovered a variety of ways by combining hybrid techniques and machine learning algorithms, Deep learning, Ultra-Wide Band (UWB) integration, and federated learning have been specifically highlighted in the studies conducted between 2023 and 2025 for improved performance in demanding scenarios. This section examines and contrasts these recent contributions, while pointing out their advantages and disadvantages as well as the way they help to create localization frameworks that are more reliable and scalable. The Table 1 represents the comparisons of various recent localization techniques.

Table 1. Comparison of recent localization strategies

Author and Year	Technique Used	Key Contribution	Limitation	How to Address it?
Zhang et al. 2023 [51]	UWB, AoA, Hybrid Localization	Achieved sub-meter accuracy in dense Urban environments	High hardware cost, unsuitable for low-power WSNs	Emphasizing on cost-effective hybrid methods for energy-constrained, scalable sensor networks
Chen et al. 2024 [52]	Federated Learning-based Localization	Enabled privacy-preserving and positioning across distributed nodes	High coordination overhead and complex training required	Focusing on lightweight, secure localization frameworks suitable for decentralized WSN deployments
Ahmed et al. 2024 [53]	RSSI and Machine Learning method (like SVM)	Improved localization accuracy in indoor environments	Accuracy degrades under dynamic environmental changes	Highlight adaptive ML models and hybrid RSSI fusion for improved stability in dynamic settings
Li et al. 2025 [54]	TDoA and Kalman Filtering	Enhanced mobility-aware localization with real-time updates	Sensitive to synchronization errors and requires precise clocks	Recommending synchronization-free or tolerant alternatives with mobility support
Liu et al. (2021) [55]	CNN-based RSSI localization	High indoor accuracy using deep learning	High computational cost and requires large training data	Our work reviews ML methods while emphasizing lightweight, scalable models for resource-constrained WSNs

The novelty of this study is found in its thorough and current assessment of localization strategies in WSNs with a particular emphasis on fusing traditional methods with cutting-edge technologies like edge computing, machine learning, and hybrid models [51, 52, 53, 54–55]. This paper presents a detailed comparison with recent studies from 2023 to 2025 and systematically compares a wide range of strategies that get divided into centralized, distributed, and machine learning-based methods, in contrast to previous works that narrowly focus on either range-based or range-free methods. Furthermore, these hybrid works highlight important issues including security, mobility, energy efficiency, and environmental interference and offers useful design guidance for creating reliable, scalable, and context-aware localization systems. This research offers a thorough analysis of current localization methods employed in WSNs and afterwards classified them as machine learning based, range based, range free and hybrid approaches. The main motto is to highlight the serious issues with existing systems and to provide appropriate design suggestions for developing precise, safe and energy efficient localization frameworks.

Hybrid-based localization methods in WSNs represent a strategic fusion of multiple localization techniques designed to enhance the overall accuracy, reliability, and energy efficiency. These methods typically combine range-based techniques such as ToA, TDoA, RSSI with range free techniques like centroid method, DV-Hop or APIT [66, 67]. Range based techniques tend to offer high precision but require additional hardware and are sensitive to environmental conditions like noise and interference [68]. On the other hand, range-free methods are more scalable and cost-effective but typically offer lower accuracy. By integrating both the approaches, hybrid methods aim to compensate for the weakness of one method with the strengths of another.

For example, in a real-world WSN deployment, nodes might initially use a range-free method to obtain a rough estimate of their positions. This estimate is then refined using a range-based method which provides higher resolution positioning [69]. This two-phase process enhances the reliability of the system especially in environments where certain signal properties might be inconsistent or unreliable. Some hybrid techniques also incorporate mobility models, Kalman filters or probabilistic methods to further refine position estimates in dynamic scenarios [70, 71]. The adaptability of hybrid localization makes it particularly suitable for applications such as environmental monitoring, industrial automation and disaster recovery where varying conditions and constraints are becoming common. Although hybrid methods can be more computationally complex and may consume more power than single-method systems, their balanced performance often justifies the trade-offs in many practical WSN scenarios. ML-based localization methods in WSNs utilize the power of data-driven models to estimate the positions of sensor nodes with high accuracy and adaptability. Unlike traditional localization methods that rely heavily on geometric calculations or signal propagation models, ML techniques learn patterns from historical or real-time data which enables them to generalize across complex and dynamic environments [72, 73]. These methods are particularly effective in Non-Line-of-Sight (NLOS) conditions and environments with significant signal distortion or multipath interference conditions where approaches often fail [74]. In ML-based localization, various algorithms such as Support Vector Machine (SVM), k-NN, Decision Tree, Random Forest and Deep Learning models like Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNNS) can be trained on signal features such as RSSSI, Time of Flight (ToF) or Channel State Information (CSI) [75].

Once trained, these models can predict node positions with minimal computation during the deployment. Furthermore, ML models can be continually updated or retrained using new data, enabling the system to adapt overtime to changes in the environment or network topology [76]. This adaptability is especially useful in applications such as indoor localization, smart buildings, and healthcare monitoring [77]. However, ML-based localization also comes with challenges. Training accurate models requires large data sets which may be difficult or costly to obtain. Additionally, some algorithms may demand significant processing power and memory, thereby posing various difficulties for resource constrained sensor nodes. Despite these challenges, ongoing advancements in lightweighting ML models and edge computing are making ML-based localization increasingly viable and attractive for next generation WSN applications.

Scope and Significance

The localization techniques used in WSNs are thoroughly reviewed in this research. Various methods, such as range-based and range-free approaches are examined along with their effects on network accuracy, efficiency, and energy consumption. By examining localization’s obstacles such as hardware restrictions, financial limitations, and environmental interference, the study provides insights into new developments like hybrid localization [51] and machine learning-based methods [13]. Future advancements in WSN localization could be guided by this research, which would improve applications in smart cities, healthcare, military surveillance, and environmental monitoring. In order to create more reliable and effective localization solutions for next-generation WSNs, researchers and practitioners can use this paper as a useful reference by examining current approaches and pointing out their advantages and disadvantages [14].

Key Contribution

The primary contributions of this study are thorough examination and in-depth analysis of localization strategies employed in WSNs. It divides localization techniques into two categories: Centralized and Distributed. It also categorizes distributed techniques into range based and range free methods. In this context, Trilateration, Triangulation and Multilateration are different types of localization methods those are discussed in this research along with their advantages, drawbacks and potential applications [15]. It also covers uses of GPS-enabled sensors and hybrid localization techniques to enhance precision and effectiveness. This study offers important insights into optimizing localization performance in WSNs by looking at the most recent developments and potential research areas while considering the cost and security constraints. It acts as a fundamental resource for scientists working to create localization algorithms those are more scalable, precise and energy efficient for real world applications [16].

Organization of the Paper

This paper is structured into multiple sections, each dedicated to a thorough examination of localization strategies in WSNs. It begins with an introduction that highlights the significance of localization, by outlining its associated challenges and presenting its various classifications. The threads of literary works are presented that offer insights into recent developments and comparative analysis of recent works. Section 2 incorporates the objective of the study. Subsequently, Sect. 3 discusses the purpose behind localization, while emphasizing its role in enhancing network performance. Section 4 elaborates on practical applications and key benefits which shows the way localization enables diverse real-world WSN deployments. Section 5 explores the role of modern technologies such as machine Learning, hybrid models and IoT integration in providing localization. Section 6 deals with Target/Source Localization in dynamic environments by highlighting both single and multi-target strategies. Section 7 is dedicated to Node-Self Localization which is further elaborated again by diving it into range based and range free localization techniques with subcategories such as ToA, TDoA, AoA, RSSI, DV-HoP, Centroid, APIT and Gradient methods. Section 8 discusses challenges and future approaches by proposing advanced solutions for existing limitations. Further, Sect. 9 provides the limitations of the existing works and Sect. 10 involves Result and Discussion section where various methods are compared and observations are made regarding their ability and trade-offs. Subsequently, Sect. 11 concludes the study by summarizing the key findings and contribution while outlining the future Scope while emphasizing on adaptive, scalable and energy efficient approaches. Finally, we list the references used throughout the paper.

Objective of the Study

This paper aims to provide an in-depth examination of the various localization methods employed in WSNs. The methods discussed range from basic techniques to more sophisticated approaches by highlighting their advantages, limitations and specific use cases and conjointly considering both the cost and security constraints. The goal of this work is to present a thorough analysis of different localization strategies applied in WSNs [18]. It will discuss the principles behind these techniques while evaluating their performance in different scenarios and exploring their advantages and disadvantages. Through this examination, this paper will also highlight the emerging trends and potential future directions in the field of WSN localization such as integration of machine learning techniques, hybrid localization approaches and the use of Internet of Things (IoT) based technologies to improve the localization performance [19]. As the demand for WSNs continues to grow, particularly in emerging areas like smart cities, vehicles automation and industrial IoT; this paper aims to contribute to the ongoing research and development efforts in such vital area of WSN topologies [20].

Localization refers to the general positions of the sensor nodes used in the network. For the network’s data transfer, this procedure is crucial. The location network of a WSN is made up of anchor nodes, displaced nodes, and a central server, according to [21]. Between the sensors, radio frequency impulses are used for communication which can be GPS-equipped. The two types of localization used in practice include centralized and distributed approaches. Again, it is further divided into two types of localization systems: range-based and range-free. Moreover, Coarse-grained (range-free) and fine-grained (range-based) localization techniques are the two main categories [22] in this context. For providing a thorough overview of state-of-art on localization techniques, the information used for localization also varies widely across them.

The range-free strategies solely employ the content of the messages, whereas the range-based methods use range measurements. Range-based localization computes a node’s location in relation to nearby nodes. Range-based systems employ a variety of methods to first calculate the distances (range) between nodes and (a few) of their neighbors before utilizing geometrical principles to compute location. To monitor range metrics like TOA, TDOA, AOA, and RSSI, they need more sophisticated gear. However, the factors that affect an estimation’s accuracy are the transmission route and the surrounding environment. Using the received communication signal’s time, angle, and other properties, as well as the signal intensity, range-free localization does not try to determine the precise point-to-point distance [23].

Purpose Behind Usage of Localization Process

In WSNs, localization is widely used to examine the sensor nodes’ precise position at any given time. Considering that a WSN comprises of hundreds of nodes, it is expensive to install GPS on each sensor node, and GPS also doesn’t provide good localization results in enclosed environments. Since WSNs are frequently deployed in hostile environments with little supervision, where individual sensors are vulnerable to security breaches. Further, the sensor readings are useless without a coordinate tag and the capacity of a node to self-localize is a highly desired feature of the sensor networks. In this context, flooding is a technique used by the majority of routing technologies to route packets [24]. A selective forward routing strategy could be utilized to route packets if the exact locations of all the sensor nodes are known. By conserving power on individual nodes, selective forwarding lowers the network load and lengthens the network’s longevity. For the use of energy-efficient routing systems and source localization algorithms, sensor data must be associated with its precise position. Figure 1 shows different categories of localization strategies used in WSN and basically these strategies are of two types namely target/source localization and node self-localization which are again segregated individually into different types. The effectiveness of WSNs heavily depends upon accurate spatial awareness which can be achieved through the localization process [25]. To identify the actual locations of sensor nodes inside a network is known as localization. It is important for many WSN applications such as tracking, surveillance and environmental monitoring [26].

[See PDF for image]

Fig. 1

Localization methods

Enabling Spatial Awareness

Providing spatial awareness in WSNs is the primary purpose of localization. At the time of data collection, in many WSN applications the location of each sensor node is critical to interpret the data. For example, in environmental monitoring applications, nodes may be used to measure various parameters such as temperature, humidity or air quality at various points in a region. Knowing the specific location of each node allows the network to associate the collected data with precise geographic co-ordinates, which is essential for analyzing spatial trends and patterns. The data would be meaningless in terms of spatial context which leads to inaccurate analysis without localization [27]. Similarly, in applications such as military surveillance or wildlife monitoring, the ability to track the location of targets or animals is the key in providing real-time insights [28].

Efficient Data Collection and Routing

The localization process supports routings and efficient data collection in WSNs. WSNs often operate in a distributed manner with nodes gathering and transmitting data to a central sink node or gateway. However, without the knowledge of their locations, sensor nodes would be unable to perform efficient routing. Localization ensures that each node knows its position and based on this information nodes can make intelligent routing decisions [29]. For example, nodes can use the localization data to forward messages to the sink node in the most energy-efficient manner, thereby minimizing the communication costs while conjointly optimizing the use of network resources. In the absence of localization, sensor nodes might blindly forward data lading to redundant communication paths, with increased energy consumption and network congestion [30]. Localization enables the design of energy-efficient routing protocols such as geographical routing where the nodes forward messages based on the geographic position of the neighbouring nodes. This reduces the overall energy usage of the network, while extending its lifetime and making it more sustainable. Additionally, knowing the position of each node can enhance the accuracy of data aggregation [31].

Practical Applications of Localization and its Key Benefits in WSNs

Location forms the backbone of a wide range of applications that require precise positioning of sensor nodes. By lowering the communications overhead and promoting energy-efficient routing algorithms, it prolongs network lifetime. Moreover, effective monitoring, control and event tracking in real-time systems are made possible by precise placement. Economics of scale are made possible by cost-effective range-free techniques that do not require costly hardware. Furthermore, advanced applications like asset tracking, mobile node management and IoT integration are supported by localization techniques that are flexible and extensible to changing surroundings. Some of these applications include the following:

Environmental Monitoring

Applications like air quality monitoring, temperature sensing and pollution detection rely on the accurate location of nodes to map out the environmental conditions across different geographical regions. With location awareness, the network can correlate data with specific areas, thereby allowing for better analysis of patterns and trends.

Health Monitoring

In healthcare applications, particularly in the context of body sensor networks, the exact placement of sensors on a patient’s body is crucial. Localization can assist in determining the precise positioning of sensors and improving the effectiveness of patient monitoring systems.

Smart Cities

In smart city applications, the localization of sensors enables features like intelligent traffic management, smart enabling parking systems and urban safety aspects. Localized sensors can detect real-time events in specific city regions, thereby allowing authorities to respond quickly and efficiently.

Agricultural Monitoring

In agriculture, localized sensor networks are used to monitor soil conditions, moisture levels and crop growth. The ability to co-relate sensor data with exact locations enables more precise farming techniques, thereby helping farmers to optimize the crop yield and reduce the waste.

Optimization of Network Resources

Localization plays a critical role in optimizing the use of network resources such as energy, bandwidth, and computation. WSNs are often composed of nodes with limited battery life and computational resources. Therefore, efficient resource usage is essential to extend the lifetime of the network. Localization data helps the nodes to make informed decisions about communication and resource allocation [33]. For example, in terms of energy consumption, localized nodes can adjust their transmission power based on their distance from the sink node. This prevents the nodes from unnecessarily expending energy by communicating at higher power levels when lower levels would suffice. Additionally, nodes can use their locations to identify nearby nodes, thereby allowing them to form energy-efficient communication clusters. Such optimizations significantly extend the network’s operational lifetime. Moreover, by knowing their positions, nodes can prioritize their communication and sensing activities. This can help to minimize the unnecessary data transmissions and reduce congestion in the network, thereby conjointly optimizing resource usage and network performance [34].

Enhancing Security and Fault Tolerance

Localization also contributes to the security and fault tolerance of a WSN. In security-critical applications, for knowing the location of nodes, it allows the network to detect and respond to potential attacks such as node capture or malicious interference. The ability to verify the positions of nodes can prevent the security breaches and ensure the integrity of data being transmitted. Additionally, the localization process helps to identify and isolate faulty or malfunctioning nodes [35]. If a node is transmitting incorrect data due to hardware issues or external interference then its location can be pinpointed and corrective measures can be taken to maintain the reliability of the network [36]. By allowing nodes to be aware of their locations, localization ensures that sensor networks can operate more effectively, efficiently and accurately leading to more successful deployments and better outcomes in real world applications. Table 1 presents a comparative analysis of the main methods, benefits, drawbacks, accuracy levels and complexity of many localization mechanisms employed in WSNs [37].

Role of cutting-Edge Technologies in WSN Localizations

In this context, the term cutting edge technology refers to the most innovative and advanced tools and systems employed currently in this field of study i.e. localization. They assist in significant enhancements like performance, efficiency, and functionality of the whole system. Modern technology has greatly improved the precision, effectiveness, accuracy, and versatility of localization employed in WSNs. Intelligent prediction models are made possible by methods like Machine Learning (ML) and Deep Learning (DL), which improve positioning accuracy by adapting to noisy data and changing situations. Real time decision making is made possible by edge computing which lowers latency by processing location data closer to the source. Furthermore, hybrid localization strategies that combine range-based and range-free procedures provide a fair trade-off between the accuracy and cost.

The integration of Internet of Things (IoT) frameworks facilitates centralized management of sensor data and smooth interconnection. These contemporary technologies not only enhance the performance but also broaden the range of scenarios in which WSNs can be used, including industrial automation, autonomous systems, and smart cities.

Target/Source Localization in Dynamic Environments and Mobility of Sensor Nodes

Target/Source Localization in WSN plays a critical role in tracking and monitoring applications. To enhance clarity and specificity, this section is organized into two sub sections: Single-Target Localization and Multi-Target Localization.

Single-Target Localization

Determining the location of a single source or target inside the monitored environment is known as single-target localization, which is a localization technique as shown in Fig. 1. For accurate estimation, methods like the AoA, ToA and RSSI are frequently used. These techniques work quite accurately in static contexts. However, problems like signal fluctuation, delay, and localization mistake are brought about by dynamic settings and the presence of mobile sensor nodes. To overcome this, mobility-aware models and adaptive algorithms are used to preserve the accuracy and resilience in real-time localization. Single-source localization can be categorized into two types like localization algorithm based on energy decay model and model independent localization algorithm.

1. Energy Decay Model based Localization: The decay model is represented by Eq. (1). The following is an expression for the signal strength that the jth sensor received over time t:

   1

   If we assume that , then represents the gain factor of the j^th sensor. is the symbol for the signal energy at a just 1 meter distance. To find the Euclidean distance between the source and the j^th sensor, we use . The measurement noise, also has a zero-mean white Gaussian distribution with variance , which may be written as . Despite appearing straightforward, this energy decay model is frequently used in the literature. In [26], the authors proposed a distributed variant of the projection onto convex sets technique and characterized the problem as a convex feasibility problem.

   The normalized incremental sub gradient approach was put up by the authors of [28] to solve the source localization problem of energy-based sensor networks, The decay factor of the energy decay method is unknown. To reduce computational complexity and to create a more thorough statistical model for energy efficiency, the authors investigated a weighted direct or 1-step least-squares-based approach, in contrast to the signal models used in [29, 30–31]. Furthermore, these approaches perform better than the quadratic elimination method since they were impacted by a correction strategy that took the reliance of unknown elements into consideration.

   The computational complexity and localization performance were well-balanced by this approach. Another method that is not dependent on the source energy (t) is the formulation of energy ratios. This was achieved by calculating the energy reading ratios in the noise-free environment. To minimize the energy usage in localization, the authors of [35] suggested an energy-aware source localization technique.

2. Model Independent Methods: An innovative model-independent localization technique was put out by Liu, et al. [45]. A distributed sorting approach is used since the nodes that measured the strongest received signal were located closer to the source. The required distance estimations are calculated using the estimated values of the associated probability density functions, assuming the sensor nodes to know their ranks. The origin is finally approximated using the Projection onto Convex Sets (POCS) method.

Multiple Target Localization

The goal of multi-target localization is to concurrently detect and track numerous sources, which is much more difficult because of signal interference, overlapping detection zones, and higher processing demands. It represents a type of target localization (as illustrated in Fig. 1). To deal with this complexity, methods like machine learning techniques, probabilistic models and clustering-based algorithms are frequently used. Localization systems must constantly update position estimates and adjust to shifting network topologies in dynamic environments with mobile sensors. To guarantee scalability and dependability, this necessitates intelligent data fusion, energy-efficient communication protocols and strong node co-ordination.

Node Self-Localization

Range-Based Localization

These localization methods (as portrayed in Fig. 1) rely on the measurement of distances or angles between nodes to estimate the positions of a target in a given space. Range based localization techniques utilize metrics like RSSI, TOA, TDOA, AOA to estimate distances or angles. These measurements are subsequently processed using mathematical approaches such as trilateration and triangulation to accurately pinpoint the target’s location [38]. While these applications may increase the overall cost and complexity of the system, the range-based localization system remains to be widely used. Despite challenges such as multipath propagation, signal attenuation and noise, particularly in indoor or densely populated environments; they offer high accuracy and adaptability across a wide range of applications. Angle as well as distance estimation are the foundations of these types of localization methods. We set the location of beacon node as and the position of unknown node as shown below in Eq. (2):

2

is the approximated range between i^th beacon node and unknown node. The unknown node’s coordinate matrix can be found as illustrated in Eq. (3):

3

1. Time of Arrival (TOA)

This localization technique (indicated in Fig. 2) estimates the position of a target node by calculating the travel time of a signal from the target node to several reference nodes. In this context, accurate time of synchronization between the transmitter and receiver is crucial as these time measurements are converted into distances using the known speed of signal propagation. The target’s location is determined through trilateration with the distance data from multiple reference points. TOA is particularly effective in applications requiring high accuracy such as asset tracking, environmental monitoring and navigation. However, its performance can be affected by factors like signal noise, multipath propagation and clock synchronization errors which necessitates advanced algorithms or filtering techniques to enhance robustness and precision. The communication signal’s time of flight is utilized for finding out the distance between a reference point and reception node. In the ToA technique, each sensor sends a signal to each of its neighbours with a predetermined fixed velocity V.

4

Where,

 = Distance between nodes X and Y

 = Power received by node X

 = Transmission capacity of node X

 = Power received by node Y

 = Transmission capacity of node Y

[See PDF for image]

Fig. 2

ToA-based range estimation

After the distance as illustrated in Eq. (4) has been established, multilateration is utilized for finding the node’s position reference. Since RF signals propagate at the speed of light, indoor settings experience different RF propagation patterns. This resulted in significant translation costs. In prior research articles, the issue of RF propagation indoors was addressed by the coupling of RF waves with ultrasound.

When compared to the rate of light, ultrasound travels at a slower rate. The two impulses’ TDoAs are utilized to calculate the distance. Another way to locate a node using TDoA is to measure how long it takes a signal to travel between two or more receivers [40]. Fig. 2 shows the TOA based range estimation technique. The receiver nodes are all designed to be time synchronized. The TDoA is determined in the manner described analytically as Eq. (5) below:

5

Where, is the TDoA,

are the two receivers’ respective distances from the transmitter.

is a measure of propagation speed.

1. Angle of Arrival (AoA)

AoA localization (indicated in Fig. 3) is a technique used in WSNs to estimate the position of a target by measuring the angles at which signals(from the target) arrive at multiple sensor nodes. This method leverages directional antennas or arrays of antennas to determine the angle relative to a reference direction, such as the true north. By using the angles of arrival from multiple sensors, the target’s position can be triangulated.

[See PDF for image]

Fig. 3

AoA in localization

AOA localization is advantageous because it provides high accuracy in dense networks and can operate effectively without requiring time synchronization across nodes unlike time-based methods. However, its performance may degrade in the presence of multipath propagation, where signals reflect off obstacles, while causing errors in angle estimation. Despite these challenges, AOA localization is widely used in applications such as environmental monitoring, asset tracking, and security systems in WSNs. The Direction of Arrival (DoA) methodology is another name for the AoA method. It calculates the absolute or relative angles between neighbours [42].

AOA is defined as the angle between a wave’s incident propagation direction and a reference direction, also referred to as orientation. We can define orientation as the constant direction measured in degrees anticlockwise from the North, against which the AoAs are measured. In this context, utilizing an antenna array on each sensor node is a typical method for obtaining AOA measurements. In [41], a brief discussion on methods for determining angles between nodes has been mentioned. This concept involves measuring the direction of neighbors using an antenna array. Special landmarks transmit location and bearing, and they also have a compass along with GPS and they follow flood beacons to receive bearing updates. It is possible to compute position once the bearing of three landmarks is known. In big sensor networks, this technique costs a lot to implement and call for more technology. Figure 3 explains the AoA method used in WSN.

1. Received Signal Strength Indicator (RSSI)

RSSI localization (indicated in Fig. 1) is a widely used technique in WSNs for determining the position of a target or node based on the strength of the received signal. This technique makes use of the well-established path-loss concept, which states that the signal intensity decreases with increasing distance between the transmitter and the receiver.

RSSI localization is popular due to its simplicity, low cost, and minimal hardware requirements, as most of the wireless devices inherently support RSSI measurements. RSSI localization usually makes use of machine learning algorithms, calibration methods, or hybrid approaches that combine RSSI with other metrics like ToA or AoA in order to improve accuracy. It finds applications in asset tracking, environmental monitoring, and location-based services within WSNs. When a radio transmission is considered as an electromagnetic wave, its energy drops as it travels over space. This indicates that with the advancement of the wave, the signal strength weakens with the weakening being inversely equal to the square of the distance traversed which is given in below Eq. (6).

6

The characteristics of RSSI measurement include its lower configuration cost compared to other methods. Estimates are imprecise because of multi-path fading, background interference, and uneven signal transmission [43]. Range-based localization, or RSSI, is predicated on the idea that one or more aspects of the communication signal between a sender and a receiver could be utilized to find the absolute distance between them. Since the RF signal is impacted by the surroundings, RSSI measurement is not more important because it cannot determine the precise distance between the nodes. The following are some measuring terms that are crucial to the measurement of RSSI:

1. Model for Path Loss

Path loss models include the Hata model [9], the logarithmic distance route loss model, and the free space propagation model. The following formula of Eq. (7) illustrates the logarithmic distance path loss model:

7

Where, s is the distance between the transmitter and the receiver in kilometres, n is the path loss exponent, which measures the rate at which the RSSI declines with distance and the value of n depends on the particular propagation environment, is a zero-mean Gaussian distributed random variable with a mean value of 0 and it represents the change in the received signal power over a specific distance, s0, which is typically equal to 1metre, and is a known reference power value.

1. Received Signal Power at Reference Distance

The following formula in Eq. (8) can be constructed if T is the received signal power at the transmitter and receiver’s distance, denoted by d0.

8

Where, P_m is power of transmitter and is a known reference power value in dBmilliwatts at a reference distance s0 from the transmitter.

1. Distance Calculated by RSSI Measurement

The formula given in Eq. (9) is used to compute the RSSI Value at a specific distance.

9

Where,

T: Strength of the signal that was received at a distance of one metre.

n: Path loss exponent and is related to the environment.

The maximum RSSI value is chosen after that and by using the provided formula, it is converted to distance. We can figure out RSSI max after computing the RSSI values as seen in Eq. (10).

10

Where, RSSI max is the highest received signal power chosen among all RSSI values. T represents the received signal power at 1 meter while n is the path loss exponent which depends upon surrounding environment.

1. Time Difference of Arrival (TDOA)

This localization technique (indicated in Fig. 1) commonly employed in WSNs aims to determine the position of a target or source node. It works by measuring the time difference of signal arrival at multiple spatially distributed sensor nodes, thereby enabling estimation of the source’s relative distance from each sensor. This information is then used to compute the node’s location using geometric approaches like hyperbolic positioning. One of the key advantages of TDOA is its capability to offer high localization accuracy without requiring global time synchronization, provided that each sensor pair is locally synchronized [44]. In order to calculate the distance difference, the TDOA algorithm first multiplies the speed at which electromagnetic waves propagate by the time difference at which the moving target is there, also referred to as the target signal (label signal), arrived at each known base station. A hyperbolic equation is then created using this distance difference.

Suppose is the arrival time between the label signal and the base station i, and is the equivalent clock error from the base station to the label signal, then the pseudo-range can be calculated from the base station to the label signal. If is the exact location coordinates of base station i, and is the position coordinate of the label, then the TOA equation can be computed as shown in Eq. (11):

11

Where, C- Speed of Light

Range-Free Localization

Pattern Matching Method

In recent years, one of the most practical options for range-free localization methods has been pattern matching localization, often known as map-based or fingerprint algorithm. Here the node’s moderate transmission range gets multiplied by the hop count and yields the distance estimate [20]. These methods have the advantages of being inexpensive because no specialized hardware is required and they make use of network topology information. There are two stages to fingerprint localization. Radio map is an offline database that stores the signals that were received at particular places during the first phase. Second phase operation occurs while it is online. By comparing the recently observed signal features to the values previously recorded on the map, the pattern matching algorithms are utilized to deduce the location of the unknown node.

A novel technique was put forth by Fang et al. [31] to extract the robust signal’s feature for effectively reducing the multipath effect. Apart from that there are multiple issues with indoor location, including the way to compensate the receiver gain differences between various mobile devices and the way to capture the characteristics of propagation signal in a dynamic setting with a complex environment. Wang et al. [24] devised a localization technique depending upon a unique differential radio map to overcome these issues by modelling them as common mode noise. With the utilization of a neural network, Gogolak et al. [25] suggested a method for fingerprint localization that could be used in actual experimental indoor settings. For the purposes of fingerprint localization, it delivered the appropriate measurement data. Pattern matching localization methods in WSNs are efficient and versatile techniques that are used to find out the position of sensor nodes by comparing the observed patterns with pre-established references that are too taken into account. The process involves three main steps: i) creating a reference database by mapping environmental patterns at known locations ii) observing real-time patterns from unknown nodes iii) matching these patterns with the database to estimate the positions. Techniques like K-Nearest Neighbor (KNN), Probabilistic methods, Co-relation analysis and Machine Learning approaches (e.g., Support Vector Machine and Neural Network) are employed to perform the accurate matching.

In this context, pattern matching is scalable, cost-effective and resilient to Non-Line-of-Sight (NLoS) conditions, thereby making it suitable for large-scale and dynamic environments. However, its effectiveness depends on various factors like the quality of the reference database, environmental stability and computational efficiency. This localization method is used in various fields including smart homes, industrial automation, health care and environmental monitoring etc. Challenges such as environmental dynamics, database construction and localization errors due to overlapping patterns or sparse data can impact accuracy, however hybrid methods like dynamic updates and advanced deep learning models can mitigate these issues. Afterall, this localization offers a balance of simplicity and adaptability to ensure its relevance in modern WSN applications.

Range-free localization techniques don’t rely on direct distance or angle measurements. They estimate a target’s location using connectivity and proximity information between nodes. These methods are cost-effective and require simpler hardware, thereby making them suitable for large scale deployments in WSNs where energy efficiency and cost are critical considerations [45]. A common range-free approach is the centroid-based method, where the location of a target node is approximated as the centroid of its neighbouring anchor nodes (nodes with known locations). Additionally, Approximate Point-In-Triangulation (APIT) method uses the relative positions of nodes to confine the target within certain triangular regions, thereby progressively narrowing down its location. Range-free methods are particularly robust in environments with irregular signal propagation as thy do not depend on precise measurements. They are widely applied in applications like disaster monitoring, habitat tracking and agricultural management. However, these methods generally offer lower accuracy compared to range-based localization, since they rely on approximations and assumptions about network connectivity. To enhance performance, hybrid models combining range-free and range-based techniques are often adopted while conjointly leveraging the simplicity of range-free methods with the precision of range-based localization. Both range-based and range-free localization methods cater to distinct application needs, with trade-offs between accuracy, cost, and complexity.

Hop-Count-Based Localization

1. DV-Hop: The Distance Vector-Hop (DV-Hop) algorithm is one of the most widely used hop count-based localization techniques. In this method, anchor nodes broadcast hop count information throughout the network, while allowing the unknown nodes to calculate their minimum hop count to multiple anchors. Anchor nodes also calculate the average hop distance using their Euclidean distances to other anchors. Unknown nodes use this average hop distance to convert their hop counts to the estimated distances and subsequently apply trilateration to determine their locations. The DV-Hop algorithm is scalable and does not require hardware for distance measurement, but its accuracy can degrade network performance with irregular topology or obstacles that affect the uniformity of hop distances [46].

   DV-Hop is a classic example of a range-free positioning system. It is not necessary to calculate the exact separation among the beacon node and the unidentified node. It lowers the amount of hardware needed by approximating the actual distances using the average hop distance. It may be applied to huge networks and is simple to build. But this also causes a commensurate rise in positional error. Three stages that make up the DV-Hop positioning process include: information broadcast, distance estimation, and location estimation. The hop count is included in the position information package where the beacon nodes send during the information broadcast stage. Initially, their nearby nodes’ hop count is set to zero. Each beacon node’s smallest hop is recorded by the receiver, which ignores its largest hop. The receiver then sends it to the nearby nodes after increasing the hop count by 1. Each beacon node’s minimum hop counts can be recorded by every other node in a network [47]. Using the formula shown in Eq. (12), each beacon node calculates the true distance of each hop by taking into account its position and the number of hops, during the distance computation stage:

   12

   Where, beacon node i and j’s coordinates are (x_i, y_i) and (x_j, y_j), respectively. This is the number of hops between the beacon nodes. Beacon nodes will then compute the average distance and send the data out to the network. Only the initial average distance is recorded by the unknown nodes, after which it is transmitted to neighboring nodes [48]. The created method primarily focuses on the following areas to increase the localization accuracy based on the corresponding node details and the average hop distance between the corresponding beacon nodes and beacon node deployment.

2. Hop Terrain: Hop terrain enhances traditional hop count-based methods by integrating signal strength information to refine hop distance calculations. Instead of assuming uniform transmission ranges, Hop-TERRAIN accounts for the variations in signal propagation caused by environmental factors, obstacles, or non-uniform node placement. By combining hop counts with signal strength measurements, it achieves more accurate localization in complex environments. This method is particularly useful in scenarios where network irregularities significantly impact hop-based distance estimations [49].

   However, it requires additional computational resources for processing the signal strength data, which may increase energy consumption in resource-constrained networks. When determining the interval between an anchor node and an unlocalized node, hop terrain is comparable to the DV hop method. Basically, the procedure is divided into two phases. The unlocalized node in the first section determines its position from the anchor node. This is an initial estimate of position. The second component of the process, which follows initial location estimation broadcasts the initial estimated position to nearby nodes. This information is sent to nearby nodes together with the distance information. Using the least square method, a node adjusts its position until the desired outcome is achieved [50].

3. Multi-Hop: Multi-hop localization is a widely used approach in hop count-based localization, particularly suited for large-scale WSNs where direct communication between nodes is not always feasible. In this method, unknown nodes calculate their positions by propagating the hop count information from anchor nodes through intermediate nodes. Algorithms DV-Hop utilizes this principle by combining multi-hop communication with trilateration to localize unknown nodes, multi-hop localization is scalable and effective in sparse networks. However, it can suffer from inaccuracies due to irregular node placement, non-uniform transmission ranges, or environmental interference.

Advanced variants address these issues by incorporating techniques like adaptive hop distance calculation or probabilistic modeling to improve the accuracy in complex network scenarios [51]. A connectivity graph can be computed using multi hop methods. For determining whether the nodes are within the communication range, the multidimensional scaling (MDS) leverages connection information. This plan consists of the following three steps.

• The first stage involves estimating the distance between every feasible pair of nodes.

• To determine the places that fit the predicate distance, MDS is utilized in the second stage.

• Lastly the known locations are taken into consideration during the final optimization stage.

Classical, Metric, Non-metric and weighted MDS algorithms are among the types of MDS utilized in large-scale sensor networks. Through the multi-hop based multilateration method, multirow nodes can work together to find more accurate position estimates.

1. Centroid System: Centroid Localization is one of the simplest hop count-based localization methods. It works by assuming that unknown nodes can estimate their positions as the geometric centroid of anchor nodes within a specific hop range. The idea is that nodes closer to multiple anchors can determine their approximate location based on their connectivity to these anchors. This method is computationally efficient and requires minimal communication overhead, thereby making it suitable for resource-constrained WSNs [52]. However, its accuracy heavily depends on the density and distribution of anchor nodes. In sparse networks or uneven deployments, the centroid approximation may lead to significant localization errors. Multiple anchor nodes that broadcast their locations using (X_i, Y_i) coordinates are used in the proximity-based approach and fine-grained localization technique is used by the centroid system. Unlocalized nodes calculate their positions after getting the received information. Anchor nodes use a GPS receiver to determine their location. The following formula in Eq. (13) is used by the node [16] to localize itself after getting the receiving anchor node beacon signals:

   13

   Where, X_est and Y_est are the locations of the unlocalized nodes that are estimated.

2. APIT: Anchor nodes in Approximate Point Triangulation scheme obtain their location using transmitters or GPS. Overlapping triangles provide position information to the unlocalized nodes. There are overlapping triangles in this area [17]. APIT is a prominent range-free localization technique used in WSNs to estimate the positions of sensor nodes. APIT is designed to strike a balance between accuracy and efficiency without relying on specialized hardware or precise distance measurements, while conjointly making it a cost-effective solution for large-scale and resource-constrained networks. The method operates by exploiting the geometric relationships between unknown nodes and anchor nodes (nodes with known positions) through triangulation.

In APIT, the network is divided into triangular regions formed by anchor nodes, and an unknown node determines whether it lies inside or outside these triangles based on the signal strength and connectivity information. The core process involves three key steps: (1) Anchor Beaconing, where anchor nodes broadcast their locations to the network, enabling nearby unknown nodes to receive signals and identify anchor nodes within their communication range; (2) Point-in-Triangulation Test, where the unknown node evaluates its relative position to a triangle formed by three anchor nodes, using qualitative RSS comparisons to decide if it is inside the triangle; and (3) Intersection Computation, where the results of multiple point-in-triangle tests are combined, and the estimated location of the unknown node is derived as the centroid of the overlapping triangular regions.

APIT localization method stands out or its robustness stands against Non-Line-of-Sight (NLOS) conditions and environmental noise by using qualitative rather precise distance data. Being range-free, it avoids costly hardware thereby making it suitable for low cost and scalable WSN deployments. However, its accuracy can suffer due to irregular node placement, sparse anchor nodes and environmental changes. Coarse localization may also occur in networks with few overlapping triangular regions. To overcome these issues, enhancements like adapting beaconing, mobility models and machine learning have been introduced.

Additionally, hybrid approaches that integrate APIT with range based methods such as RSSI or TOA can improve accuracy. APIT finds applications in disaster management, environmental monitoring, health care and industrial automation. Recent advances in computational capabilities and algorithms have enabled its integration with IoT and AI technologies, thereby enhancing its adaptability and precision. Overall, APIT offers a cost-effective and balanced solution for localization in WSNs. The following four steps are included in this scheme: Once anchor nodes send beacon signals to unlocalized nodes, the latter preserve the table. Here, location, signal strength, and anchor ID are all included in the table.

(ii) Randomly selecting three anchor nodes from the surrounding area, unlocalized nodes check to see whether they form a triangle and Point in Triangulation (PIT) testing or not.

(iii) PIT testing keeps going on until any combination of the three anchor nodes can be used to determine an exact location for the unlocalized nodes. In the end, the estimated position of an unlocalized node is determined by computing its Centre of Gravity (COG), which is the intersection of all triangles.

1. Gradient Algorithm: The gradient algorithm is another type of hop count-based localization that exploits the gradient of hop counts from anchor nodes to determine the positions of unknown nodes. In this method, anchor nodes broadcast the hop count information, and each node calculates its minimum hop count to one or more anchors. Unknown nodes then estimate their locations by interpolating the gradient field created by the hop count values. Gradient based localization is highly efficient in terms of computational and communication overhead as it avoids the need for complex mathematical operations like trilateration. This method performs well in networks with uniform node distribution but is sensitive to irregular topologies, environmental noise and errors in hop count propagation.

Variants of the gradient algorithm: Variant of the gradient algorithm such as that incorporate signal strength or refine hop counts with machine learning aims to enhance its robustness and adaptability in real world scenarios. Unlocalized nodes in gradient algorithms employ multilateration to determine their locations. The second stage involves an unlocalized node figuring out the quickest route among anchor nodes and itself from which it receives beacon signals. The mathematical Eq. (14) below is used to get the approximate distance between the anchor node and the unlocalized node:

14

Where, d_hop is the estimated distance covered by one hop.

1. In the last phase, multilateration is employed for finding out the node’s coordinates to get the least error as illustrated in Eq. (15).

   Additionally, it makes use of the hop count value which starts out at 0 and is increased as it spreads to neighboring nodes [18]. From anchor nodes, each sensor node obtains information about the shortest path. The steps of the gradient algorithm are as described in [19]. Here, the anchor node transmits a beacon message in the initial stage that contains its hop count and co-ordinate information.

   15

Where, d_ji is the estimated distance using gradient propagation. The Table 2 shows a comparative study of various localization strategies.

Table 2. Comparative analysis of localization techniques based on performance and complexity

Localiza- tion Strategy	Methods Used	Advantages	Drawbacks	Accuracy	Complexity
Range −Based	TOA	High accuracy in LOS (Line of Sight)	Requires precise time synchroni zation	High	High
	TDOA	Reduces clock synchronization issues	Requires multiple receivers, affected by multipath propagation	High	High
	AoA	Effective in directional antennas	Needs special hardware for angle estimation	High	High
	RSSI	Simple, no extra hardware required	Affected by environmental interference and multipath fading	Low to Medium	Low
Range Free	DV-Hop	No need for additional hardware, suitable for large networks	Prone to hop-distance estimation errors	Medium	Low
	Centroid Method	Low complexity, simple implementation	Low accuracy in sparse networks	Low	Low
	APIT	Does not require distance measurements	Accuracy depends on node density	Medium	Medium
Hybrid Methods	Combination of Range-Based and Range-Free Techniques	Improved accuracy and efficiency	Higher computational cost and complexity	High	High
Machine Learning-Based	Deep Learning, Neural network	Adaptive and can improve accuracy in dynamic environments	Requires training datasets and high processing power	High	Very High

Challenges and Future Approaches

To improve their accuracy, effectiveness, and suitability for a variety of settings, localization techniques must find creative answers to a number of problems. By using AI-driven error correction, hybrid systems like GPS-Wi-Fi fusion and GPS-based localization are employed that can overcome its challenges with signal blocking in indoor and urban environments with high power consumption in case of decreased accuracy in crowded regions. Since Wi-Fi and Bluetooth-based localization techniques are impacted by signal interference and infrastructure reliance; adaptive beacon placement and machine learning models are required for location refinement. Advancements in active RFID and affordable hardware can help to offset the high implementation costs associated with RFID technology, notwithstanding its effectiveness for short-range tracking.

Although Ultra-Wide Band (UWB) localization is very accurate, however, it has problems with cost and power consumption, which has led to development of chip designs that use less energy. Since vision-based techniques are sensitive to changes in the environment and demand a lot of processing resources, real-time edge. AI and multi-modal data fusion are posing as the essential developments.

Similar to this, localization based on Inertial Measurement Units (IMUs) is susceptible to drift errors [40, 41–42]. These can be reduced by employing sensor fusion and AI-based drift correction approaches. Hybrid integration with RF and vision-based approaches and AI-driven noise filtering could be beneficial for acoustic-based methods, which are frequently impacted by ambient noise and have a restricted range. In the future, more robust and dependable localization systems will result from addressing these issues with innovative technology developments. Table 3 below provides a summary of the issues that exist now and potential solutions for various localization techniques in the future

Table 3. Research problems and possible future approaches for different localization methods

Localization Methods	Research Problems	Possible Future Approaches
GPS-based Localization	Signal blockage in indoor and urban environments	Integration with other localization methods (e.g., Wi-Fi, LiDAR, UWB) in a cost effective and secured manner
	High power consumption	Development of energy-efficient GPS modules while curtailing the cost
	Low accuracy in dense urban areas	AI-based error correction techniques
Wi-Fi-based Localization	Dependence on infrastructure (AP density, signal interference)	Deployment of hybrid Wi-Fi and BLE-based solutions considering cost constraint.
	Signal fluctuations due to environmental changes	Machine learning models for adaptive positioning
RFID-based Localization	Short-range and tag dependency	Development of active RFID with enhanced range
	High deployment cost for large-scale environments	Cost-effective hardware and open-source localization frameworks
Bluetooth (BLE) Localization	Limited accuracy due to signal attenuation and interference	AI-driven signal processing and data fusion with other techniques
	Requires high density of beacons for better accuracy	Use of adaptive beacon placement algorithms
UWB (Ultra-Wideband) Localization	High deployment cost	Development of cost-effective UWB chips
	Power consumption concerns in mobile devices	Energy-efficient signal processing algorithms
Vision-based Localization	High computational requirements	Edge AI processing for real-time object detection
	Susceptibility to lighting and environmental variations	Multi-modal data fusion with LiDAR and IMU
IMU-based Localization	Error accumulation (drift) over time	Sensor fusion with GPS, Wi-Fi, and vision-based methods
	Requires frequent recalibration	AI-based drift compensation techniques
Acoustic-based Localization	Environmental noise interference	AI-based noise filtering and adaptive signal processing
	Limited range in large open areas	Hybrid integration with RF and vision-based methods
	Environmental noise interference	AI-based noise filtering and adaptive signal processing

Limitations of Existing Work

Numerous constraints still exist in the body of existing work, despite notable progress in localization approaches for WSNs now-a-days. The majority of conventional methods concentrate only on range-based or range-free techniques, each of which has inherent trade-offs. For example, range-based techniques generally require costly hardware and are susceptible to interference from the environment, while range-free techniques are less accurate and heavily rely on the node density. Furthermore, in dynamic environments with mobile nodes or obstructions, many of the algorithms are now in use that perform less well because they assume perfect network circumstances and static node placement. Lightweight sensor nodes may not always be able to handle the computational resources and in this context, huge labelled datasets need for machine learning-based techniques, despite their potential usage. Furthermore, there is still much more to learn about security issues like spoofing and data manipulation in localization procedures. These drawbacks emphasize the need for more resource-efficient, safe, and adaptable localization methods that can function dependably in diverse and real-world WSN installations.

Result and Discussion of Existing Approaches

This study examined the effect of variety of localization strategies in WSNs, including range-free strategies like DV-Hop, Centroid, and APIT, as well as more conventional range-based strategies like RSSI, ToA, and AoA existing in literature. It also covered recent advanced strategies such as Machine Learning driven algorithms and hybrid localization methods. A detailed comparison shows that range based techniques offer high accuracy in controlled environments, however they are susceptible to noise and require additional hardware support. On-the other hand, range-free methods are the most cost-effective ones and easier to implement but often suffers from lower localization precision, particularly in sparse network scenarios. Hybrid techniques such as combining RSSI with TDOA or AOA have shown promising results in achieving a balanced trade-ff among accuracy, scalability, and cost. Furthermore, it has been seen that machine learning-based models, including SVMs, Random Forests, and Neural Networks can learn intricate signal patterns and adjust to changing contexts, which improves localization accuracy in practical settings. Nevertheless, these models can be computationally demanding for low-power sensor nodes and frequently rely on sizable training datasets. Such findings are further supported by a comparison of contemporary methods from 2023 to 2025 (as indicated in Table 1) and Table 2 illustrating comparative analysis of existing localization techniques based on performance and complexity. In order to address issues like real-time processing, mobility, and data security; recent research has been increasingly incorporated in case of edge computing and privacy-preserving models. Our research asserts that there is no one approach that is always the best; rather, the choice should be based on the situation, taking into account the requirements of the application, network size, node mobility, and environmental factors. Considering all things, this conversation highlights the necessity of flexible, energy-conscious, and safe localization solutions, particularly for upcoming implementations in IoT-based systems, smart cities, and extensive sensor applications.

The advancement of more precise, scalable and energy efficient localization methods that can adjust to changing and diverse setting is the key to the future of localization in WSNs. New developments like the fusion of edge computing, machine learning and IoT frameworks present encouraging avenues for improving localization performance in real-time application. More-over, multimodal data fusion and hybrid localization models that blend range-based and range-free techniques can aid in overcoming restrictions associated with node mobility and environmental interference. To satisfy the changing needs of autonomous systems, smart cities and remote sensing applications; future research should also concentrate on low-cost hardware solutions, self-calibrating algorithms and security-aware localization.

Conclusion and Future Scope

This paper offers a thorough analysis of localization strategies used in WSNs, while emphasizing on how crucial and precise the node placement is with respect to effective network operation and application success. By methodically dividing localization strategies into range-based and range-free categories and further subdividing them into centralized, decentralized and distributed models. The paper offers a thorough grasp of their fundamental ideas, benefits, drawbacks and applicability towards a variety of real-world situations. Algorithms including RSSI, ToA, TDoA, AoA, DV-Hop, Centroid, and APIT, as well as hybrid and machine learning-based techniques are compared to justify that no one method performs consistently well in every setting. Rather, the selection of a localization technique depends on the context and necessitates in balancing trade-offs between hardware cost, energy usage, accuracy, and complexity. AI-driven models and hybrid techniques have demonstrated potential in enhancing localization accuracy and responding to changing environments, especially in large-scale, mobile, or obstacle-rich networks. Along with highlighting potential future possibilities including IoT integration, edge computing, AI-enhanced models, and cross-technology fusion; the article also examines the difficulties presented by signal interference, node mobility, and environmental variability. In order to close the remaining gaps and provide scalable, economic, and energy-efficient localization solutions, these developments will be becoming very much essential. Afterall, this research is a fundamental resource for researchers and professionals who want to create reliable and flexible localization systems for new WSN applications in fields like industrial automation, smart cities, environmental monitoring, and healthcare, etc.

In this context, the advancement of more precise, scalable and energy efficient localization methods that can adjust to changing and diverse setting is the key to the future of localization in WSNs. In this context, new developments like the fusion of edge computing, machine learning and IoT frameworks present encouraging avenues for improving localization performance in real-time applications. More-over, multimodal data fusion and hybrid localization models blend range-based and range-free techniques that can aid in overcoming restrictions associated with node mobility and environmental interference. To satisfy the changing needs of autonomous systems, smart cities and remote sensing applications, future research should also concentrate on low-cost hardware solutions, self-calibrating algorithms and security-aware localization.

Acknowledgements

The authors would like to express their gratitude to Siksha ‘O’ Anusandhan University for providing an encouraging academic environment that facilitated this research.

Author Contributions

First author S. Panda contributed to the conceptualization, literature review, writing and overall structuring of the manuscript. Corresponding Author S.B.B. Priyadarshini assisted in reviewing, editing and refining the concept to enhance its quality and clarity. All authors have read and approved the final version of the manuscript.

Funding

Open access funding provided by Siksha 'O' Anusandhan (Deemed To Be University). This research was conducted without any specific financial support from government agencies, private organizations or non-profit institutions.

Data Availability

Since this study is a review article, no new datasets were generated or analyzed. All relevant information has been sourced from publicly available literature, which is appropriately cited in the manuscript.

Declarations

Ethics Approval and Consent to Participate

This research is based on a review of existing literature and does not involve human participants, animals or sensitive data that require ethical approval. Therefore ethics approval and consent to participate are not applicable to this study.

Consent for Publication

All authors have carefully reviewed the final manuscript and have provided their full consent for its publication.

Competing Interests

The authors declare that they have no competing financial or non-financial interests that could have influenced the work presented in this paper.

Abbreviations

WSN

Wireless Sensor Network

AoA

Angle of Arrival

DV-Hop

Distance Vector-Hop

ToA

Time of Arrival

TDoA

Time Difference of Arrival

RSSI

Received Signal Strength Indicator

GPS

Global Positioning System

LAR

Location Aided Routing

RF

Radio Frequency

MDS

Multidimensional Scaling

APIT

Approximate Point-In-Triangulation

COG

Centre of Gravity

EM

Expectation Maximization

MR

Multi-Resolution

Publisher’s Note

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

References

1. Guha, S; Khuller, S. Approximation algorithms for connected dominating sets. Algorithmica; 1998; 20, 4 pp. 374-387.1600830 [DOI: https://dx.doi.org/10.1007/PL00009201]

2. Guoqiang, M; Fidan, B; Brian, DOA. Wireless sensor network localization techniques. Comput Netw; 2007; 51, 10 pp. 2529-2553. [DOI: https://dx.doi.org/10.1016/j.comnet.2006.11.018]

3. Wang J, Gao Q, Wang H et al. (2011) Differential radio map-based robust indoor localization. EURASIP J Wireless Commun Netw 2011 (article 17)

4. Whitehouse K, Culler D (2002) Calibration as parameter estimation in sensor networks. In Proceedings of the First ACM International Workshop on Wireless Sensor Networks and Applications

5. Alrajeh, NA; Bashir, M; Shams, B. Localization techniques in wireless sensor networks; 2013; Hindawi Publishing Corporation: pp. 1-9.

6. Meng, W; Xiao, W; Xie, L. An efficient EM algorithm for energy-based multisource localization in wireless sensor networks. IEEE Trans Instrum Meas; 2011; 60, 3 pp. 1017-1027. [DOI: https://dx.doi.org/10.1109/TIM.2010.2047035]

7. Liu, H; Darabi, H; Banerjee, P; Liu, J. Survey of wireless indoor positioning techniques and systems. IEEE Trans Syst Man Cybern C Appl Rev; 2007; 37, 6 pp. 1067-1080. [DOI: https://dx.doi.org/10.1109/TSMCC.2007.905750]

8. Want, R; Hopper, A; Falcão, V; Gibbons, J. The active badge location system. ACM Trans Inf Syst (TOIS); 1992; 10, 1 pp. 91-102. [DOI: https://dx.doi.org/10.1145/128756.128759]

9. Wang, Y; Wang, X; Wang, D; Aggarwal, DP. Range-free localization using expected hop progress in wireless sensor networks. IEEE Trans Parallel Distrib Syst; 2009; 20, 10 pp. 1540-1552. [DOI: https://dx.doi.org/10.1109/TPDS.2008.239]

10. Savvides A, Han CC, Srivastava MB (2001) Dynamic fine-grained localization in ad-hoc networks of sensors. In Proceedings of the 7th Annual International Conference on Mobile Computing and Networking

11. Hofmann-Wellenhof A, Lichtenegger H, Collins J (2001) Global positioning system: theory and practice, 5th edn. Springer

12. Bachrach J, Taylor C (2005) Localization in sensor networks. In: Handbook of sensor networks: algorithms and architectures, 1st edn, 1

13. Priyadarshini, SBB; Panigrahi, S; Bagjadab, AB. A distributed triangular scalar cluster premier selection scheme for enhanced event coverage and redundant data minimization in wireless multimedia sensor networks. Indian J Sci Res; 2017; 14, 2 pp. 96-102.

14. Ampeliotis,; Berberidis, K. Low complexity multiple acoustic source localization in sensor networks based on energy measurements. Signal process; 2010; 90, 4 pp. 1300-1312. [DOI: https://dx.doi.org/10.1016/j.sigpro.2009.10.015]

15. Luo, X; O’Brien, WJ; Julien, CL. Comparative evaluation of Received Signal-Strength Index (RSSI) based indoor localization techniques for construction jobsites. Adv Eng Inf; 2011; 25, 2 pp. 355-363. [DOI: https://dx.doi.org/10.1016/j.aei.2010.09.003]

16. IEEE. (2011 Dec) Proceedings of the International Symposium on Intelligent Signal Processing and Communications Systems, Chiang Mai, Thailand, 1–5

17. Patwari, N; Ash, J; Kyperountas, S; Hero, AO; Moses, RL; Correal, NS. Locating the nodes: cooperative localization in wireless sensor networks. IEEE Signal Process Mag; 2005; 22, 4 pp. 54-69. [DOI: https://dx.doi.org/10.1109/MSP.2005.1458287]

18. Flocchini, P; Prencipe, G; Santoro, N. Self-deployment of mobile sensors on a ring. Theor Comput Sci; 2008; 402, 1 pp. 67-80.2434238 [DOI: https://dx.doi.org/10.1016/j.tcs.2008.03.006]

19. Am, RY; Youssef, M. A taxonomy of localization schemes for wireless sensor networks. ICWN; 2007; pp. 444-450.

20. Mao, G; Fidan, B. Localization algorithms and strategies for wireless sensor networks; 2009; Hershey, IGI Global: [DOI: https://dx.doi.org/10.4018/978-1-60566-396-8]

21. Li W, Cassandras CG (2005) A minimum-power wireless sensor network self-deployment scheme. In IEEE Wireless Communications and Networking Conference, vol 3. pp 1897–1902

22. Garg V, Jhamb M (2013 Apr) A review of wireless sensor network on localization techniques. Int J Eng Trend Technol (IJETT) 4(4)

23. Kulaib AR, Shubair RM, Al-Qutayri MA, Ng JWP (2011 Apr) An overview of localization techniques for wireless sensor networks. In International Conference on Innovations in Information Technology, Abu Dhabi, 167–172

24. Wilson, J; Patwari, N. Radio tomographic imaging with wireless networks. IEEE Trans Mob Comput; 2010; 9, 5 pp. 621-632. [DOI: https://dx.doi.org/10.1109/TMC.2009.174]

25. McGuire M, Sheehan JP, Davis JP (2008) Network-assisted GPS positioning for mobile phones. In Proceedings of the 13th IEEE International Conference on Personal Wireless Communications

26. Priyadarshini, SBB; Al, E. A distributed approach based on maximal far-flung scalar Premier selection for camera actuation. ICDCIT 2016, lecture notes in computer science(); 2016; Cham, Springer: 9581

27. Teng J, Bolbrock T, Cao G, Porta T (2007) Sensor relocation with mobile sensors: design, implementation, and evaluation. In Mobile Adhoc and Sensor Systems, 2007. MASS 2007. IEEE International Conference on, Pisa, 1–9

28. Wang G, Cao G, LaPorta T (2004 Jun) Movement - assisted sensor deployment. In IEEE INFOCOM

29. Fang, S-H; Lin, T-N; Lee, K-C. A novel algorithm for multipath fingerprinting in indoor WLAN environments. IEEE Trans Wireless Commun; 2008; 7, pp. 3579-3588. [DOI: https://dx.doi.org/10.1109/TWC.2008.070373]

30. Shang Y, Ruml W, Zhang Y, Fromherz M (2003) Localization from mere connectivity. In Proceedings of the 4th ACM International Symposium on Mobile Ad Hoc Networking and Computing

31. He T, Huang C, Blum B, Stankovic J, Abdelzaher T (2003) Range-free localization schemes in large scale sensor networks. In Proceedings of the 9th Annual International Conference on Mobile Computing and Networking

32. Savarese C, Rabaey JM, Beutel J (2001) Locationing in distributed ad-hoc wireless sensor networks. In Proceedings of the International Conference on Acoustics, Speech, and Signal Processing

33. Niculescu, D; Nath, B. DV based positioning in ad hoc networks. J Telecomm Syst; 2003; 22, 1–4 pp. 267-280. [DOI: https://dx.doi.org/10.1023/A:1023403323460]

34. Doherty L, Ghaoui LE, Pister KSJ (2001) Convex position estimation in wireless sensor networks. In Proceedings of the 20th Annual Joint Conference of the IEEE Computer and Communications Societies (INFOCOM), vol 3. pp 1655–1663

35. Xu Y, Winter J, Lee W-C (2004) Prediction-based strategies for energy saving in object tracking sensor networks. In Proceedings of the 5th International Symposium on Mobile Ad Hoc Networking and Computing (MobiHoc)

36. Cheng X, Thaeler A, Xue G, Chen D (2004) TPS: a time-based positioning scheme for outdoor sensor networks. In Proceedings of IEEE INFOCOM

37. Langendoen, K; Reijers, N. Distributed localization in wireless sensor networks: a quantitative comparison. Comput Netw; 2003; 43, 4 pp. 499-518. [DOI: https://dx.doi.org/10.1016/S1389-1286(03)00356-6]

38. Capkun, S; Hamdi, M; Hubaux, J-P. GPS-free positioning in mobile ad-hoc networks. Cluster Comput; 2002; 5, 2 pp. 157-167. [DOI: https://dx.doi.org/10.1023/A:1013933626682]

39. Nagpal R, Shrobe H, Bachrach J (2003) Organizing a global coordinate system from local information on an ad hoc sensor network. In Proceedings of the 2nd International Conference on Information Processing in Sensor Networks (IPSN)

40. Girod L, Estrin D (2001) Robust range estimation using acoustic and multimodal sensing. In Proceedings of IEEE/RSJ International Conference on Intelligent Robots and Systems

41. Boukerche, A; Oliveira, H; Nakamura, E; Loureiro, A. Localization systems for wireless sensor networks. IEEE Wireless Commun; 2007; 14, 6 pp. 6-12. [DOI: https://dx.doi.org/10.1109/MWC.2007.4407221]

42. Kusy, B; Lédeczi, A; Maróti, M. Time synchronization for robust wireless sensor networks. IEEE Trans Parallel Distrib Syst; 2006; 17, 4 pp. 556-567.

43. C.-T, C; T.-Y, H; T.-Y, L; C.-C, C. Application of a deep learning algorithm for detection and visualization of hip fractures on plain pelvic radiographs. Eur Radiol; 2019; 29, 10 pp. 5469-5477. [DOI: https://dx.doi.org/10.1007/s00330-019-06167-y]

44. Tian, G; Xia, Y; Zhang, Y; Feng, DDF. Hybrid genetic and variational expectation-maximization algorithm for gaussian-mixture-model-based brain MR image segmentation. IEEE Trans Inf Technol Biomed; 2011; 15, pp. 373-380. [DOI: https://dx.doi.org/10.1109/TITB.2011.2106135]

45. Stoleru R, He T, Stankovic JA, Luebke D (2005) A high-accuracy, low-cost localization system for wireless sensor networks. In Proceedings of the 3rd International Conference on Embedded Networked Sensor Systems (SenSys)

46. Hill, JL; Horton, M; Kling, R; Krishnamurthy, L. The platforms enabling wireless sensor networks. Commun ACM; 2004; 47, 6 pp. 41-46. [DOI: https://dx.doi.org/10.1145/990680.990705]

47. Bahl P, Padmanabhan VN (2000) RADAR: an in-building RF-based user location and tracking system. In Proceedings of IEEE INFOCOM

48. Goldenberg D, Bihler P, Cao M, Fang J, Anderson B, Morse A (2006) Localization in sparse networks using sweeps. In Proceedings of the 12th Annual International Conference on Mobile Computing and Networking (MobiCom)

49. Behura A, Priyadarshini SBB (2021) Assessment of load in cloud computing environment using C-means clustering algorithm. In Intelligent and Cloud Computing: Proceedings of ICICC 2019, vol 1. Springer Singapore (pp. 207–215)

50. Priyadarshini, SBB et al. A quadrigeminal scheme based on event reporting scalar premier selection for camera actuation in wireless multimedia sensor networks. J King Saud Univ Eng Sci Elsevier; 2019; 31, 1 pp. 52-60. [DOI: https://dx.doi.org/10.1016/j.jksues.2017.04.002]

51. Zhang, Y; Wang, L; Zhao, H. Hybrid localization using UWB and AoA in dense urban WSN environments. IEEE Sensors J; 2023; 23, 4 pp. 3345-3354.

52. Chen, L; Xu, M; Li, Y. Federated learning-based privacy-preserving localization for wireless sensor networks. IEEE Internet Things J; 2024; 11, 2 pp. 1678-1689.4719210

53. Ahmed, T; Kumar, R; Mehta, S. Indoor localization using RSSI and machine learning: a support vector machine approach. Ad Hoc Netw; 2024; 150, 102632.

54. Li, X; Han, J; Zhou, W. Mobility-aware node localization using TDoA and Kalman filtering in WSNs. IEEE Access.; 2025; 13, pp. 9820-9831.

55. Liu, Y; Zhang, B; He, Q. RSSI-based indoor localization using CNNs for dynamic environments. Sensors; 2021; 21, 17 5889.

56. Sharma, K; Kumar, P; Alam, MS. Hybrid RSSI-ToA based adaptive localization for WSNs using neural networks. IEEE Syst J; 2023; 17, 1 pp. 456-467.

57. Gupta, S; Kaur, A. A lightweight federated learning scheme for location prediction in IoT sensor networks. IEEE Internet Things J; 2023; 10, 5 pp. 4150-4161.

58. Wu, J; Zhang, T; Wang, L. Edge-assisted adaptive localization for dynamic WSN environments. IEEE Trans Mob Comput; 2023; 22, 4 pp. 1823-1835.

59. Roy, A; Dey, S; Saha, P. Multi-hop RSSI-based localization enhanced by fuzzy logic in WSNs. Wireless Networks; 2024; 30, pp. 1745-1760.

60. Zhao, C; Huang, Y. Deep reinforcement learning for real-time node localization in harsh WSN environments. IEEE Sensors J; 2024; 24, 2 pp. 2200-2212.

61. Ali, M; Singh, R. CNN-LSTM hybrid architecture for indoor localization in WSNs. Ad Hoc Netw; 2024; 143, 102630.

62. Nguyen, H; Le, B; Vo, T. A survey on localization techniques for wireless sensor networks: trends and future directions. IEEE Access; 2023; 11, pp. 56512-56535.

63. Wang, R; Li, J. Secure anchor-based localization for wireless sensor networks using blockchain. Comput Netw; 2023; 228, 109400.

64. Ahmed, T; Khan, H; Patel, N. Lightweight hybrid trilateration for large-scale WSNs in smart agriculture. IEEE Sens Lett; 2023; 7, 5 pp. 1-4.

65. Fernandes, L; Rebouças, M. Indoor positioning with improved accuracy using UWB and fingerprinting fusion. Sensors; 2023; 23, 4 2145.

66. Sun, Z; Wu, H; Liu, Y; Zhou, S; Guan, X. A hybrid localization algorithm for an adaptive strategy-based distance Vector-Hop and improved sparrow search for wireless sensor networks. Sensors; 2023; 23, 20 Art. 8426. [DOI: https://dx.doi.org/10.3390/s23208426]

67. Agrawal, U; Srivastava, A. A novel hybrid approach for localization in wireless sensor networks. Communications in computer and information science; 2023; Cham, Switzerland, Springer: pp. 22-33. [DOI: https://dx.doi.org/10.1007/978-3-031-25380-5_2] 1668

68. Hada, RPS; Aggarwal, U; Srivastava, A. A study and analysis of a new hybrid approach for localization in wireless sensor networks. J Web Eng; 2023; 22, 2 pp. 279-302. [DOI: https://dx.doi.org/10.13052/jwe1540-9589.2224]

69. Panwar K, Babu P (2022. Available: May) Majorization-minimization based hybrid localization method for high precision localization in wireless sensor networks,” arXiv preprint, https://arxiv.org/abs/2205.03881

70. Yadav, et al. Hybrid localization scheme using K-fold optimization with machine learning in WSN. Int J Commun Syst; 2022; 35, 12 e5206. [DOI: https://dx.doi.org/10.1002/dac.5206]

71. Kumar, S; Batra, N; Kumar, S. Optimized localization in large-scale heterogeneous WSN. J supercomput; 2023; 79, pp. 6705-6729. [DOI: https://dx.doi.org/10.1007/s11227-022-04922-5]

72. Yadav, P; Sharma, SC. A systematic review of localization in WSN: machine learning and optimization-based approaches. Int J Commun Syst; 2022; 36, e5397. [DOI: https://dx.doi.org/10.1002/dac.5397]

73. Abdullah OA, Al-Hraishawi H, Chatzinotas S (2023) Deep learning-based device-free localization in wireless sensor networks. In Proc. IEEE WCNC, pp 1–6). https://doi.org/10.1109/WCNC55385.2023.10118744

74. Gouda A et al. (2022 Dec) A grid-based sensor floor platform for robot localization using machine learning. arXiv preprint, https://arxiv.org/abs/2212.04721

75. Alhmiedat, T. An efficient optimal localization technique for WSN using hybrid machine learning algorithms. Wireless Pers Commun; 2023; 130, pp. 651-672. [DOI: https://dx.doi.org/10.1007/s11277-024-10892-z]

76. Abdelhakim, A. Machine learning for localization of radioactive sources via a distributed sensor network. Soft Comput; 2023; [DOI: https://dx.doi.org/10.1007/s00500-023-08447-8]

77. Abdullah OA et al. (2023) Fingerprint-based localization approach for WSN using machine learning models. Appl Sci 13(5):Art. 3037. https://doi.org/10.3390/app13053037