Design Issues of Wireless Sensor Network and Open Research Challenges

WSN is becoming more popular day by day among researchers due to its vast real-time applications. Various design parameters directly affect the overall performance and lifetime of the network. The design of WSN system is depends upon application. Many parameters need to consider to design WSN system such as number of nodes, types of nodes, routing strategies, way of communication among the nodes and in the network, etc. The network structure and its operations categorize the WSN into flat-based and cluster-based systems. At present, the major challenge is designing and deploying the WSN system. The objective of the paper is to study and analyze all the essential parameters to overcome the challenge. The paper also critically reviews the literature. Finally, the paper is summarized and concluded with open research issues to design robust, low cost, energy-aware WSN systems.


Introduction
Over the last few years, the research community and industry significantly developed an interest in Wireless Sensor Network (WSN) due to its wide range of real-time applications. After the invention of radio communication in WSN, the scope of applications such as rescue & disaster systems, wildlife monitoring, health monitoring, medical diagnosis, agriculture, security applications, etc. have tremendously increased [1] [2]. WSN usually comprises a large number of sensor nodes placed uniformly or randomly in the target area.
A WSN is a communication between sensor nodes and the base station (BS) for data sensing and gathering a wireless communication channel. The sensor node sends sensed data to BS for processing. The processed data by BS is forwarded to the gateway to send it to the internet or satellite [3] [4]. Through the internet or satellite, the data is communicated to the end-users.
The sensor nodes may either be homogeneous or heterogeneous depending on the type of application [5]. Sensor nodes are used to monitor physical and environmental conditions like temperature, humidity, pressure, sound, heat, light, etc. In multimedia WSN applications, sensor nodes are equipped with cameras. As the way of communication in the WSN is merely the exchange of messages among sensor nodes and BS, it leads it to be more applicable in complex and intelligent applications.

Cluster-based WSN architecture
In the cluster-based WSN system, the target field is divided into several groups, and each group is termed a cluster as shown in figure 2. Each of these clusters works as a flat based WSN system and select CH from among the sensor nodes, instead of BS. Sensor nodes send their sensed data to CH. CH collects the data and aggregates it and transfers it to the BS through other CH if it is not in the communication range of BS [7]. For transmission static or dynamic route is selected. All sensor nodes should be in the communication range of BS is not a must.
To design and implement cluster-based WSN system, many design issues such as cluster formation, cluster head selection, data collection, and dissemination, data aggregation, intra-cluster communication, inter-cluster communication, routing protocols, data load management, relate nodes, latency reduction, node failures, security, and network energy, etc. are detected. A robust system with techniques that are more Innovative and optimized, needs to be designed.

Global Positioning System (GPS)
The use of location devices is necessary for WSN systems, as they are very useful in detecting sensor nodes, faulty diagnostic mechanisms, and dynamic routing. The GPS systems are found to function efficiently in an open area, but they cannot be used in a closed area. It is a requisite to design a support system for GPS, to give more accuracy and precision and, that is able to operate in a covered area [8]. Some GPS techniques are mentioned below.

Traditional GPS
Satellite-based GPS is a radio navigation system. It provides locations in the 2D space of the GPS receiver if the receiver is placed in the obstacle-free field, and clearly in the line of sight of at least four or more GPS satellites which is shown in figure 3. Its accuracy lies in between 5 meters to 100 meters and precision, 5 meters to more than 20 meters, hence GPS is not advisable to use in WSN because WSN needs more accuracy and precision [8].  [8] [9] [10] The GPS base station is placed in a location where there are at least, four or more GPS satellites that are in sight. The distance is calculated by a satellite and communicated with the rovers. At the same time, the rover's receiver gets the measurements from the satellites and processes them with the BS information. It calculates their location relative to the base station. The accuracy of GPS based RTK is up to 2 centimetres. This technique can be used in the localization of sensor nodes and routing in WSN [11].
More supportive techniques like these need to be designed with the existing system to give good accuracy and precision.

Structure of the paper
The paper is mainly focused on the design issues of Flat based and Cluster-based WSN systems and finally, open research challenges are discussed. All the literature related to WSN is studied in-depth and explained all possible and required parameters in the abstract. The structure of the paper is shown in figure 5.

Contribution of the Paper
The prime aim of the paper is to give a brief overview to the beginners or the inexperienced researchers in the field of WSN, regarding the design issues of traditional and clusterbased WSN. It gives brief guidelines to study and implement WSN systems in reality. The author had critically reviewed all essential parameters and extracted all the pros and cons which affect the performance of WSN. The paper highlights the current challenges of the WSN system. Finally, open challenges are mentioned and can be considered as research objectives for further research of the WSN system.

First-order Radio Mode
First-order Radio Model is used as stated in LEACH [12]. Radio dissipates 50 nJ/bit to transmit and receive for an amplifier 100 pJ/bit/m 2 , the energy loss due to channel transmission 'k' bit message and 'd' distance. The transmitting and receiving energy can be calculated as [8].
Transmitting Energy

Modified First order Radio Model
To optimize the energy consumption in the network, the clustering approach of nodes is implemented by [13,14]. Performance in terms of energy consumption is evaluated using first-order Radio Model.
Receiving energy is the same as LEACH.
The first order radio model is not accurate as it is not considering listening power consumption in the networks. It also assumes that the communication range of the sensor node is unlimited. A novel discrete model must be applied to calculate the energy calculation of the node. The total energy i and Ei are computed as follows [15] ∑ ∑

Distance Calculation
Longitudes and Latitudes can be received from GPS and distance between two nodes can be calculated. It is similar to calculation of two points in 2D plane by Euclidian distance formula [11].

Energy Consumption
The operations of sensor nodes are broadly classified into sensing, CPU processing, the transmission of data, receiving of data, nodes in idle state, and nodes in the sleep state. The maximum power consumption occurs when the sensor nodes are in the idle state and for transmitting and receiving data. The node which is active and not performing any task, is in the idle state. In this state, transmitters and receivers are active even though they are not performing any task. Sleep state is the state when the node is active but, not performing, sensing, nor processing, and the transmitter and receiver are in off mode. So, in the sleep mode very less energy is consumed [17]. Approximate energy consumption is shown in figure  6.

Design issues and Challenges
Now a day WSN system is applicable in real time and more complex applications. The technology is also growing simultaneously so there is need of more research in the existing design techniques to satisfy current applications need. It is a challenge to design the WSN as various factors need to be considered depending mainly on the type of application and type of target field. Accordingly, sensors and have to be designed, and other strategies need to be implemented. There are mainly two types of WSN systems flat-based WSN systems and cluster-based WSN systems [6] [7]. The design issues of both are different which are stated below.

Flat based WSN systems
Flat based WSN systems are generally deployed in smaller areas. The issues and challenges to design it are as follows:

Field Environment
Nowadays, WSN systems can be implemented on plain ground, hilly area, rainy forest, underwater, multi-storage buildings, city squats, etc. While designing WSN systems the consideration of field environment is very much important. The actual field decides the type of sensors, localization strategies, network topologies, routing protocols, information collection policies that are to be implemented [1] [2] [6].

Hardware constraints
The WSN Systems consists of four main components sensor nodes, base stations, communication media, and gateways. These components transmit the sensed information to the world via the network or internet. Sensor nodes are made up of four elements: sensing element, processing element, transceiver element, and battery power element. It also encompasses the location finding element, and mobilizer element [18] [19]. The architecture of the sensor node is shown in figure 7. These elements are packed in one unit called a sensor, as per the requirement of application and budget cost of the system, its size, processing power, battery power, and sensing capacity varies. The components of sensor nodes are shown in figure 8. The weight and size of the sensor also vary as per the application, undersea applications require heavy sensors, and air applications lightweight sensors. [18]. A few design challenges of the sensor nodes are mentioned below-  The base stations (BS) have to be localized in such a way that they can gather information from sensors and transmit them to the gateway. This affects communication in terms of a range of sensors. Communication media may be air, water, underground, flat ground, uneven ground, smoke etc. The additional units such as GPS modules and EAI Endorsed Transactions on Energy Web 01 2021 -03 2021 | Volume 8 | Issue 32 | e14 mobilizers are attached to the sensor node [18]. The GPS module is required in most of the sensors to get the current location and mobilizers to control the movement of mobile sensor nodes.

Network topology and field coverage
Topology indicates the way of sensor nodes connected to base stations to transmit sensed information. WSN applications use various topologies such as bus, star, ring, tree, mesh, circular and grid topology etc. In the cluster architecture, one or more of the topologies can be used for sensing and routing information from the cluster head to the base station [2].
There are mainly two challenges that are field covered. Field coverage is the quality of the target field under surveillance by the sensors [20] [21]. It is divided into three subtypes which are shown in figure 9. a. The highest quality/blanket coverage-It offers complete coverage of the target field with the help of dense sensor nodes. It increases the cost of the system to implement and maintain the WSN system. b. The medium quality/barrier coverage It offers a border line or pathway coverage to identify intruders at the time of crossing the network that will not be possible by the system. The lowest quality/Sweep coverage It does not cover the complete target field in the same instance. Sensors can operate in active and sleep mode to save energy. Once the event is detected the sensor operates in active mode or as per time set else, it works in sleep mode. Sparse localization of sensor nodes in the network is made up of mobile sensor nodes and they can be used to cover the target field. It reduces the cost to implement and maintain the system due to the less number of nodes.

Connectivity
Continuous connectivity is essential for communication in the network. The efficiency of the network depends on the proper connectivity of network [22] [23] It has two forms which are shown in figure 10. If the sensor node operates in sleep mode, and its neighbouring node wants to send information, then it has to wait until the sleep mode node is in active mode or, find the alternative path. So, the sensor's active and sleep mode should be optimally scheduled to avoid other sensor node's waiting periods as it affects the connectivity of the network.
b. The spatial domain-Currently, available sensor nodes can operate in multiple energy levels. As per the energy levels of the sensor node, the communication range varies. So, there is a need to decide a policy to set a proper energy level of each sensor to maintain connectivity.

Node Types
There are two types of sensing nodes mentioned in the literature [6] [24] shown in figure 11.

Figure 11. Node types
Once the network topology is deployed, the lifetime will be the same throughout for the WSN system in the specific target field. MSNs are also the same as SSN. SSNs are fixed in a location, but in the case of MSN's, location changes depending on the movement of the autonomous, manual moving robots or movable trolleys on which they are placed. Applications like target finding and target tracking generally use MSN's.

Localization
Localization is analysing initial spatial coordinates of sensor nodes and deploying them at the appropriate location. The purpose of the localization is to cover the EAI Endorsed Transactions on Energy Web 01 2021 -03 2021 | Volume 8 | Issue 32 | e14 required target field for the surveillance. There are two ways to deploy sensor nodes in the field [22] as shown in figure 12.
Either manually, using the algorithmic technique or Global Positioning System (GPS), the spatial coordinates are decided initially in the fixed localization.
Sensor nodes are spread unevenly in random localization. If the target area is dangerous, for humans to enter then using a drone sensor nodes can be spread. The drawback of random localization is, it does not guarantee that sensor nodes will locate at equal distance from each other, and all are in essential range to communicate with BS. In most cases, these sensors are self-organized and allocate appropriate positions on the target field. Under the ocean, localization is still a challenge for the researchers to find the appropriate localization technique due to environmental parameters [25].

Figure 12. Localization
Using GPS is not often always productive for the localization, because GPS works well in an obstacle-free environment but, it does not work under the Deep Ocean and in a closed environment. So, there is a need to find a unique and generic technique to localize the sensor node in different environments.

Stationary and Mobile Base Station
In the literature, most of the researchers used stationary BS for their inventions. All nodes surround the BS in the target field and connected with clustering or simple network topology to sense and collect information [18] [22]. The drawback of the stationary BS is neighbour nodes of BS drains more energy as compared to other nodes because the sensed information by all sensors is routed through these nodes to BS. When these nodes are dead, the complete network collapses because the sensed information will not reach to BS.
So there is an alternative way that BS can be mobile so that neighbouring nodes of BS change every time, hence there is less possibility to drain more energy by the same nodes [26].
The challenge of stationary BS is, to enhance the lifetime of network by reducing the death rate of neighbouring nodes of BS. The challenges of Mobile BS are, it needs to define strategies for movement planning (predefined path or dynamic path), information gathering strategy current location awareness to network, the power consumption of BS, and sends collected information to the gateway, etc. [27].

Single and Multiple Base Stations
Most of the researchers in literature used a single BS for the complete network. The drawback of a single BS is that it requires more transmission energy to send data from the sensor node to BS. If the target area is too large, multiple BS can be deployed to collect information. The multiple BS approach requires extra infrastructure to connect to the gateway [6]. There are two ways of multiple BS approaches as shown in figure 13.
Multiple BS in stationary approach collects data from the nearby group of sensor nodes. For the mobile multiple BS approach, a dynamic strategy needs to be implemented as per the movement of BS.
The movements of mobile multiple BS can be constrained made random, or predefined. According to the current location, BS needs to broadcast its location to all its neighbour nodes or the group of nodes that are allocated to it [28] [29].

Transmission Media
In WSN applications, mostly radio communication is used as a transmission medium. In the sensor node, power consumption is high, and it depends on the radio frequency [30] [31]. The major hurdle is the selection of frequency that is unregulated due to ISM (Industrial, Scientific, and Medical) band for communication. Other alternatives can be used such as Ultra High Frequency (UHF), Infrared, optical communication, Ultra-Wideband (UBW), etc. For the underwater environment, Acoustic Communication (AC) and Cognitive Acoustic (CA) are mainly used as, transmission media. It has its own challenges such as path loss, ambient noise, multipath effect, Doppler spread, etc.

Communication Energy
The radio is the main consumer of the energy of the sensor while transmitting and receiving data. The distance of the receiver and transmitter majorly affects the energy EAI Endorsed Transactions on Energy Web 01 2021 -03 2021 | Volume 8 | Issue 32 | e14 8 of the sensor node. Existing research shows that the energy required to transmit 1 bit over 100 meters is equivalent to processing approximately 3000 instructions. There are two solutions suggested in the literature. The first one is, to switch off the radio whenever the sensor node is not receiving and transmitting data. Second, minimize the size of data to be transferred. The Data can be processed, and duplication avoided at the sensor end only before transmitting [32].
In the literature, more focus is given on optimizing communication and, routing protocols to reduce the usage of energy though some improvements are still expected.
These are the design issues and challenges of Flat based WSN systems, to the researchers to improve the system performance, and to extend the network lifetime.

Cluster-based WSN system
In the large target areas, due to the limitations of BS in terms of energy wastage for transmission, the clusterbased WSN approach is more popular. The WSN network is divided into different groups and works independently in a distributed manner. After collecting the data, it is routed to BS. The major design issues and challenges are mentioned below.

Cluster Formation
Due to the limited sensor energy, processing power, and memory of sensor nodes, designing the WSN system is a more challenging task. Traditional systems are easy to design for small target areas, but for large target areas, its performance degrades in terms of energy for data transmission to long distances. To transmit the information from the sensor node to BS, it directly depends on the distance from the Sensor node to BS. So, there are more chances of the nodes located at long a distance from BS to drain off early. Hence LEACH (Low-Energy Adaptive Clustering Hierarchy) protocol is designed and suggested a novel technique to divide target area into different parts [12]. Each part is called a cluster. The clusters work as a separate flat-based WSN system, the only difference is that instead of BS, there is a Cluster Head (CH). CH collects the data from all its cluster nodes and forwards it to BS after aggregating. The opted path may be static or dynamic. Some researchers in their literature have suggested the movable BS, which moves to each CH to collect the information [33].
Using the cluster technique overall network lifetime increases efficiently. To design the cluster WSN system, different parameters need to be considered, which are as follows-

Number of nodes in a cluster
There are no fixed criteria for the number of nodes that should be in one cluster. It mainly depends upon the area of the target and sensing range of the sensor nodes. It is also influenced by the processing power and memory of CH. The number of nodes in one cluster is decided by the static allocation of cluster or dynamic allocation of the cluster as per the size of the target field. It is still a challenge to find the optimum number of nodes in one cluster depending upon the type, and size of the application. It also depends upon the volume of data transmission [6] [34].

Cluster Count
The number of clusters in a WSN system depends on the target field size, the density of nodes in the field, and the processing power of CH. There should be an optimum number of clusters in the network because each cluster has a CH, and it needs to do a lot of activities such as, gathering data from all the nodes from a cluster, aggregate the data, and transmit it to the next hop or BS. Hence, the CH may drain off early so, the selection process of the new CH needs to be initiated, which requires a lot of processing power, and energy. As the distance is directly proportional to the drain of energy the distance of the furthest node, from the CH, in one cluster, needs to be considered in the formation of a cluster. To find the optimal size of the cluster is still a research problem because the optimal cluster size directly affects the performance of the network and the overall energy of the network [6] [35].

CH selection
The minimum distance from all the nodes and the maximum energy of the sensor node is the main criterion of the selection of the CH, from the available nodes in the cluster. The CH in a cluster usually drains more energy as it is involved in the CH selection process, gathering data from sensor nodes, aggregation, and transmitting information to next-hop or BS. CH cannot be the same throughout the lifetime of the network or cannot be predefined; as it drains more energy so it can drain off early. Hence, the dynamic selection of CH is preferable. There are many CH selection algorithms suggested by researchers as election algorithms. If the energy of CH goes below the threshold level then, the new CH selection process must be initiated immediately and a new CH is decided for a cluster. The Previously selected CH now works as a normal node for sensing data. This selection of the new CH process continues till the last alive node in the cluster [36].
Random technique and probabilistic technique also can be used to process the selection of the CH. Each technique has its pros and cons, and so the selection of CH to increase the overall lifetime of the cluster still remains a research issue. Each technique has its pros and cons, and so the selection of CH to increase the overall lifetime of the cluster still remains a research issue. The sensor node has two ranges one for sensing and the other for communication with further sensor nodes and BS. Searching Range (SR) of the sensor node is a range or radius of the node in which it can sense the required parameter from the assigned area. Communication Range (CR) / radio range is a radius of a sensor node in which it can communicate with each other [18]. The sensing range and communication range of the sensor node is shown in figure 14.

Data collection and Dissemination
After sensing data, the sensor node transmits data with or without processing to CH as per the communication range. Sensed data is stored in the memory if the CH is not in the communication range of the sensor node due to some reason, and transmitted whenever CH will be in range. If the CH is too far than its communication range, then through the static or dynamic path, data will be transferred through other sensor nodes of the same cluster to CH. Generally, CH is always in the communication range of the sensor node of the same cluster [32].
Most of the sensor nodes send sensed data without filtering at their end. CH has to process the received data and aggregate it. But at the end of the sensor node itself, if data is processed and only required data is transmitted to CH then CH may save a lot of time and energy. This technique also reduces the packet size for the transmission of the sensor node to CH and ultimately less energy will be required for the transmission. So it can increase the overall lifetime of the cluster and ultimately of the network.

Aggregation
In a cluster, all the sensor nodes send sensed data to CH. CH has a lot of duplicate data sent by all sensors. To reduce the data duplication and convert it into a required and transferrable format is called aggregation. In some applications where confidentiality is a prime factor, data transmitted in the encrypted form [6].
Still, there is wide scope to find novel techniques of data aggregation and encryption because it is directly proportional to several bits transmitted in the network and it consumes more energy.

Intra Cluster Communication
Generally, single-hop communication is adopted in intracluster communication. Sensor nodes send their data to CH if it is in the range of direct communication, otherwise, multi-hop communication strategy needs to use. If multi-hop communication is adopted and filtering of data is not performed by the sensor node, then, more energy is wasted to transmit huge unstructured and duplicate data. The sensor node should have a mechanism to compare, last sent data, and current sensed data. If it is the same, then it should be avoided to transmit [33] [37]. Only one message needs to be transmitted; 'No Change' so that CH realizes that there is no variation in the data and, last received data can be considered for the processing or, avoided. So, it saves energy. Thus, it is not advisable to apply multi-hop communication in the same cluster. Star network topology is preferable for single-hop communication.

Inter-cluster communication
The cluster-based WSN system has multiple clusters and respective CHs. CH transmits data directly to BS like star topology if, BS is in the communication range of it. But, if the BS is far away then, the data should be transmitted through other CH, which is called multi-hop communication. The path may either be static or, dynamic. The static path is predefined. The drawback of the static path for routing is, the CHs which are on routing path drain more energy. Generally, dynamic route-finding techniques are used for multi-hop communication. Dynamic path selection increases the overall lifetime of the network, but it requires extra energy for the process. Even though a lot of novel ideas have been invented to reduce the use of energy, the network, still needs a better generic and optimized technique that consumes less energy [33].

Routing Protocols
Routing protocols decide the way of transmission of data to BS. There mainly two cases as shown in figure 15.  The network, which has a movable BS, proves more challenging to track the BS. The BS may follow the predefined path to cover all the CH in all the clusters to receive data from CH. The BS needs to broadcast a signal to its neighbour CH nodes once it changes its location, then CH initiates data transmission as BS is in the communication range.
BS changes its location again after receiving the data. For the data collection, multiple BS can be used for a large target area. All or some of the BS can follow the static or dynamic path for moving [1]. Due to the environmental condition and the type of the target area, movable BS cannot be applied to all applications, in such cases, stationary BS is preferred. It also follows static and dynamic route selection to transmit the data. Static path selection is easy to implement but the drawback is the CH nodes which are near to BS, drains more energy, as they need to route more packets as compared to other outer CHs.
In dynamic path selection, at every hop, the decisionmaking mechanism is required to select the next-hop as per its energy and the distance from BS. The path may not be the same at every routing. Dynamic route selection is widely used in cluster-based WSN. Dynamic route selection requires processing and energy at every hop and consumes more energy, so the invention of an optimized technique is essential.

Optimized clustering algorithms
In stationary BS, dynamic path selection is a challenging task. It is a non-deterministic polynomial (NP-Hard) problem. There is no effective linear solution to it, so, an optimization technique helps to improve the network lifetime and routing efficiency [22].
Several optimization techniques like fuzzy logic, genetic algorithm, neural network, reinforcement learning, and swarm intelligence are used in the literature. Among all swarm intelligence technique is mostly used by a researcher which includes Ant Colony Optimization (ACO), Particle Swarm Optimization (PSO), Bee Colony Optimization (BCO) [38] [39].
Many of the researchers have implemented hybrid optimization techniques for the routing data but still, there is a need to find a generic solution, which will be applied to most of the applications.

Data load management
The transmission of data in the network must be managed and routed optimally. The CH node near to BS, due to heavy load traffic of data, drains off more energy as compared to distant CH nodes. Some researchers suggest that the size of clusters that are near to BS must be small as shown in figure 16. So, CHs of those clusters will require less energy to process aggregation and transmission of data. Some researchers have pointed out that, a separate group of routing nodes should be located near to the BS so that, they will only involve in receiving and transmitting of data to BS as shown in figure 17. By observing the overall network energy utilization of nodes, mostly CH nodes drain off early, so there is a need to find some alternative solutions on load management so that, the overall network energy level will be maintained [24].

Relay node
The concept of a relay node is when any node fails to sense or transmit data in the routing path due to failure or other reason, standby or relay node takes its position.

Collision Avoidance
Collision avoidance strategies need to apply to CH for two reasons. First is the flooding of data by sensor nodes to CH and second is at the same time when another CH sends data to transmit to the next hop or BS. Such instances may invite collisions. To avoid the collisions unique time slots need to be provided for each sensor node, or neighbouring CH. The highest priority can be assigned to data transmission of CH to CH communication, as it effects on latency time [33].

Figure 16. Architecture of unequal clustering in WSN
The relay node may be static or movable. The network should have the mechanism to identify the location of the failed node so that, the relay node can take its place manually or automatically, a movable relay node [40]. The applications in which there are limitations for the installation of the node, it may face the problem of locating relay nodes also.

Latency Reduction
The total time needed to transmit information from source CH to BS is called latency time. Due to the improper load management at CH, channel bandwidth, availability of route nodes, non-optimal path selection, the latency time may increase. Routing paths should be always shortest, optimal, and obstacle-free to reduce latency time [33].

Node Authentication
Any malicious new node can be the part of a network and harm the network or steal the information. So a strong authentication strategy must be applied to a new node entry in the network. Only authentic nodes should take part in the communication and sense of data. Any intruder can disturb the network operation [41] [42].

Routing Node Failure
The causes of node failure are lack of energy, physical damage, malicious attack, not in a coverage area, failure of communication link etc. If any routing path node fails due to the lack of energy or other reasons, the system should find an alternative path for transmission. In static path selection, if any routing node fails, then part of the network will be isolated even though the isolated network works fine. There should be a mechanism to assign the responsibility of routing to its neighbour node or replace the node [43].

Network lifetime
As the sensor node has limited energy and there is no provision to recharge it, it is essential to utilize the available energy efficiently. While designing intra-cluster and inter-cluster communication, dynamic path selection, aggregation, and CH selection, the energy parameter must be considered, and optimization techniques need to be applied. It will help to extend the overall lifetime of the network [44] [45].

Open Research Challenges in WSN
As WSN is the most demanding technology due to its huge number of applications, more researchers are focusing to optimize the existing techniques and trying to find novel techniques for the problems. There are many open issues in WSN some of them are stated below.

Energy Level of Sensor
The variations of radio performance depend upon the energy level of the sensor node. The reasons for radio variations are due to anisotropy, variations in direction and, the heterogeneous energy level of sensors [17] [33].
Anisotropy is at various directions' path loss, observed while transmitting a signal by the sensor node. Path loss is due to incremental changes in the propagation direction of the transmitter. At a certain period, the remaining energies of all nodes are different due to the power consumption of sensor nodes for operations. The energy level of sensor nodes affects communication and data transmission.

Realistic connectivity model
It is assumed that the communication range is approximately twice more than the sensing range of the sensor node in literature [6]. While designing protocols for routing, it is also assumed that the link quality is the same for the nearest possible distance and maximum possible distance as per the communication range of the sensor node. These assumptions are not always correct. Link symmetry is one of the features of WSN and weakens as the distance increases. It effects of packet loss, packet delay, and overall transmitting and receiving speeds. Communication link quality should not be associated with distance [46]. RSSI (Received Signal Strength Indicator) and LQI (Link Quality Indicator) are the two standard link quality metrics to be considered. Signal strengths can be checked by the RSSI parameter at the receiver end while receiving the packet. The strength and quality of the packet which are received can be checked by LQI.

Transfer of Energy and Energy Harvesting
Mainly the energy consumption of the sensor node depends upon the processing load, receiving, and transmission of the data. So every node is having a different level of residual energy. It is an open research challenge to transfer node's energy from one node to another [20]. So that the node which is about to die can EAI Endorsed Transactions on Energy Web 01 2021 -03 2021 | Volume 8 | Issue 32 | e14 get energy from the other node which has a good amount of energy. It may broadcast the energy by predicting the energy level of the respective nodes, whether it is below the threshold level or not. If it is possible to transfer then, the overall network lifetime will increase.
If it is possible to charge the nodes wirelessly, from a remote place by broadcasting from fixed or, mobile energy sources then, the energy issue of WSN will permanently be overcome. Energy harvesting is one more research challenge. Each sensor node needs to be equipped with; energy harvesting equipment and gets charged generating power from the natural resources as a requirement arises.

Mobility of sensor node in WSN
The mobile sensor node can cover more areas than static sensor nodes. The mobility of WSN can be constrained, random, or predefined. Adequate coverage is achieved in constrained movements [11] [22]. The Random movement moves the MSN in any direction, but it is assumed that to transmit the data, the MSN will always be in the communication range. The Predefined moves of the MSN, is a fixed path which is set at the time of deployment of the system. MSN's mobility reduces the number of SSN's in the network, so there is a need to focus on the research in the mobility of the sensor nodes, its communication, and routing strategies. It improves the quality of coverage, connectivity & overall lifetime of the network.

Heterogeneous Network
In the literature, most of the work is proposed on homogeneous WSN systems where all the nodes are of the same configurations [22]. There are some applications where different sensor nodes deployment is essential. So, there is a need to focus on a heterogeneous system and its operations like communication strategy between sensor nodes, data routing strategy, path selection, security measures, etc.

Environment with obstacles
In the existing research, most of the researchers have presumed a plane and obstacle-free target field. It is not always possible in reality to have an obstacle-free target field. The researchers have also assumed circular sensing and a communication range [1]. The range may get changed due to environmental factors such as; shadowing effects, interference in the network, signal strength, and energy of a node, leading to the connectivity issues in the network, so existing proposed protocol fails in such situations. More work is expected on the presence of obstacles and variations in sensing and communication ranges to provide robust solutions.

Optimal sleep and wake-up of sensor nodes
Many researchers have proposed sleep and wake-up node scheduling. The Sensors that are not required for a short period, some are put in the active mode and some in the sleep mode in this scheduling. In the sleep mode, it shuts itself, and either low power timer or, low power trigger sensing device is on [22] [33]. It automatically emanates in the wake-up mode as expected event triggers at a particular time. Centralized and distributed are the two methods to control scheduling. Centralized scheduling is easy to control but challenging for a dynamic environment. Distributed scheduling works well in a dynamic environment but it always leads to imprecise results [48].
Optimal wake-up and sleep schedules are also studied based on, coverage, probability, periodic and random in the literature. Optimal wake-up and sleep scheduling still a research problem in WSN for efficient utilization of resources and extend the lifetime of the network.

Probability of Node failure
Failure of the node directly affects the area coverage of the target field. In the target field, where manual localization is not possible, we need to spread the sensor nodes with the help of a drone. There is a chance of physical damage to the components and hardware failures [22].
Another reason for the failure of nodes in environmental factors such as; increase in temperature, pressure, humidity, rain, storm, etc. The failure of the nodes influences the coverage area. In the literature, the node failure issue is less discussed. As it is directly affecting coverage of the target field, it should be focused and investigated more.

Optimal clustering technique
Most of the researchers have proposed even-sized clusters in the WSN. But in reality, it is not always possible. The drawbacks of even-sized clusters are the clusters and CHs which are near to BS, drains more energy, compared to the outer cluster, as they need to transmit their data and the outer cluster's data too. Hence, the size of the clusters near BS can be reduced. The architecture of unequal clustering in WSN is shown in figure 18. It can extend the overall lifetime of the network [47].

Un-even initial energy level of sensor nodes
Most of the researchers assumed the same initial level battery power for all the sensor nodes. But, it is not EAI Endorsed Transactions on Energy Web 01 2021 -03 2021 | Volume 8 | Issue 32 | e14 probable in realistic applications. At the deployment level, it is not always possible to have the same battery level due to battery potentials and other initial operations performed. Hence, new approaches need to investigate by considering the un-even initial energy level of sensor nodes that will give different realistic results.

Cryptographic Security
Some researchers have proposed cryptographic techniques to encrypt the data for transmission. The traditional ad hoc networks cryptographic techniques cannot be used in WSN due to the limitations of power, processing, and memory. Some lightweight novel techniques need to be implemented which require minimum energy and network bandwidth for the processing and are more secure to transfer.

Large streaming applications
Nowadays, WSN applications are connected to IoT (Internet of Things) applications which demands huge data transmission in the network. Also, ecological monitoring and multimedia applications need to generate and transmit a huge amount of data. It is a big challenge to the researches to design communication, and routing strategies to support a considerable amount of data transmission.

Conclusions and future directions
Designing a robust, low cost and energy-aware WSN system is itself a challenge due to the involvement of different design parameters. All the design parameters are surveyed and critically reviewed of flat-based and clusterbased WSN systems. The pros and cons of existing parameters are discussed wherever necessary, and the future scopes are mentioned briefly. Lastly, the current open research issues are discoursed by considering the future of WSN technology.
In the future, each sensor node should be able to generate power for processing to extend the lifetime. To implement energy harvesting strategies, using natural resources such as; heat, light, and vibrations to WSN.