An Energy-Efficient Clustering Approach for Wireless Sensor Networks to Reduce Hot-Spot Effect and Idle Listening Energy Consumption

Wireless sensor networks (WSNs) formed from the sensor nodes (SNs) that are basically small micro electro-mechanical (MEMS) as well as a base station (BS) or sink [1]. The primary goal of the SNs is to collect data from nearby nodes and forward it to the BS. In modern life, WSNs have become the integral part as in disaster management, weather forecast, industrial automation, heath care, surveillance and security, etc. [2-4]. SNs can efficiently implement a WSN via indirect or direct interface between each other where every node acts like a router [4]. A network architecture of WSN is categorized into three groups as location-based, hierarchical and fat routing protocols [5]. In general, SNs used an ad hoc fashion for the deployment [1, 6]. The basic differences between the ad hoc networks and WSN are [7]:

·The sensor nodes in WSN are much greater than the SNs in ad hoc networks.

·SNs are generally implemented in a larger number.

·SNs in WSN are open to failure.

·In WSN topology, frequent variations occur.

·Sensor nodes are constrained in memory, computation, and power.

A network is developed as clusters and offers better improvement in network energy as well as lifetime in hierarchical case. The other sensor nodes are known as member or common nodes, and each cluster is led by a cluster head (CH) [8]. SNs are implemented under crucial environments in most of WSN applications. While limited battery power can make the SNs as energy constrained. The lifetime of sensor nodes can be extended under these circumstances by giving the appropriate time to energy-aware routing protocol [9]. Static or dynamic routing strategies can be used by the routing methods [10]. Static routing allows data to be transmitted from sensor nodes to the destination using set pathways. Conversely, data trafficking to the destination is accomplished by dynamic routes in dynamic routing. WSN often uses dynamic routing since it allows data transfer even in the event of a router failure or death. The standard clustering of WSN is shown in Figure 1.

Hierarchical routing system was mainly anticipated for wireline networks [5, 11] because of its energy efficient and good scalability interface. Because of the aforementioned characteristics, Heinzelman [11, 12] developed the hierarchical routing strategy in WSNs. In hierarchical routing, the high-energy sensor nodes use data aggregation and the inert-cluster interface. Conversely, lower-energy SNs are using their specific clusters to gather sensing data. The idea of giving sensor nodes a unique role as cluster members or member nodes leads to improvements in network longevity, data packet collision avoidance, energy efficiency, and network scalability. The most well-known hierarchical routing techniques, including low-energy adaptive clustering hierarchy and hybrid energy-efficient distributed clustering, have made use of WSN protocols.

Extensive research work has been performed that can cover several LEACH based WSN protocols such as energy-efficient uneven clustering (EEUC) algorithm [13], power-efficient gathering in sensor information systems [14], two-level hierarchy LEACH [15], etc.

1.png

Figure 1. Structure of sensor node deployment and cluster formation in WSN [1]

LEACH [12] becomes the trending hierarchical clustering-based protocol for WSNs. It is an adaptive clustering protocol that utilizes equalized energy distribution and spontaneously organizes itself between the sensor nodes over rounds. The process works in rounds and every round includes the steady-state phase and clustering setup. The sensor nodes place themselves in a group having one node as representative during the clustering phase, known as cluster head (CH). On the other hand, the remaining SNs are named as regular/normal SNs. During the clustering stage, regular sensor nodes decide to either accept or reject a cluster for the existing round. It depends on the achieved signal strength. When utilizing single-hop communication for transmitting data from regular SNs to CH and from CH to BS, the steady-state phase is lengthier than the clustering phase. The member node of the cluster received the TDMA schedule from the cluster head, which further facilitates intra-cluster communication.

LEACH introduced the data compression with randomized rotation of cluster head before transferring it to BS for enhancing the WSN network lifetime. The percentage of CH in a network and the frequency with which the SN has continued to function as a CH in previous rounds are taken into consideration by the SN when making a decision about the CH for the current round during the cluster setup phase. For the current round, the sensor node selects a number between 0 and 1, and is promoted to the CH if its value is smaller than the predefined threshold T(n). The mathematical expression for the threshold T(n) is [16]:

$T_n=\frac{p}{1-p\left[rmode \frac{1}{p}\right]}, if \,\,\, n \in G$             (1)

$T_n=0, \,otherwise$              (2)

Figure 2 displays the radio model of LEACH as well as the intra cluster and direct CHs to BS transmission [12].

2.png

Figure 2. Single hop and intra-cluster communication from Cluster head to the base station

Each CH delivers a message to member nodes during the steady-state phase. It also maintains their receivers on to listen mode as well as responds to respective cluster head accordingly. The subsequent mathematical expressions can define a radio transmission model [16]:

$E_{T X}(k)=E_{\text {elec }} k+\epsilon_{amk } k d^2, \quad if \quad d \leq d_o$           (3)

$E_{T X}(k)=E_{ {elec }} k+\epsilon_{ {amp }} k d^4, \quad if \quad d>d_o$              (4)

$E_{R X}(k)=E_{{elec }} k$             (5)

where, d₀ is a threshold distance, ϵ_amp is the energy consumed by the radio amplifier as a result, and E_elec is the energy needed for k bit transmission over distance d.

This research work has the following contributions:

i. This Paper concerns the energy efficiency of the network which is the most important criterion for applications where battery replacement of sensors is not possible.

ii. Proposed solution minimizes the energy consumption at sensors due to the hot-spot effect. In the proposed solution hot-spot effect is avoided which ultimately results to prolong the network lifetime.

iii. Proposed solution minimizes the energy consumption at sensors during the idle-listening phase in which sensors drain some amount of energy while not transmitting any data. For this sleep and wake-up approach is used to avoid idle-listening energy consumption.

The paper provides an energy-efficient clustering approach for Wireless Sensor Networks (WSNs) that can directly relate to manufacturing applications. To address the hot-spot effect requires careful consideration of node placement (based on distance from the sink) and communication protocols can mitigate the hotspot effect, manufacturing plays a role in producing reliable and energy-efficient nodes for deployment. Efficient sensor deployment and management can lead to improved process optimization, reduced downtime, and cost savings in manufacturing facilities.

The paper can provide insights into how the proposed clustering approach enhances the monitoring and control of manufacturing processes, leading to better product quality, and increased efficiency. The following approaches are used in this work:

i. Energy-efficient Hardware: Designing sensor nodes with energy-efficient components, low-power microcontrollers, and optimized power management systems can extend their battery life.

ii. Advanced Battery Technologies: Integrating improved battery technologies, such as rechargeable or energy-harvesting batteries, can help reduce the frequency of battery replacements.

iii. The clustering approach proposed in the paper involves designing a hybrid hierarchical structure where sensors are organized into clusters. This design aspect is crucial for optimizing network performance and energy consumption. The design of such clustering algorithms requires careful consideration of factors like network topology, communication overhead, and data aggregation strategies. For Coverage Optimization: Utilizing tools like coverage prediction models during the network design phase can help determine the optimal sensor placement to achieve uniform coverage and reduce potential hotspots.

iv. Power Management: Implementing advanced power management techniques such as duty cycling, where nodes alternate between active and sleep modes, can significantly reduce idle-listening energy, using adaptive sleep scheduling algorithms based on traffic patterns or network conditions can further optimize energy usage. Use of predictive models and redundancy strategies during the design phase to optimize sensor placement and ensure consistent coverage.

v. Energy-Efficient Protocols: This paper Employs energy-efficient communication protocols based on Low-Energy Adaptive Clustering Hierarchy (LEACH), which can help in reducing unnecessary data transmissions.

vi. Distance Aware approach: The paper focus on energy efficiency directly involves computing aspects. Clustering algorithms, which group sensors based on distance or other criteria, involve complex computation to determine optimal cluster heads and communication patterns.

The following are the main mechanisms used in this research:

i. Data Aggregation: Implementing data aggregation techniques at intermediate nodes can reduce the number of transmissions, conserving energy.

ii. Wake-up Mechanisms: Utilizing wake-up mechanisms that allow nodes to remain in a low-power state until an actual data transmission is required can reduce idle-listening.

Wireless sensor networks (WSNs) are becoming an essential and dynamic aspect of modern life for humans. The monitoring of the atmosphere is essential to the entire plan, including battle domains, climate monitoring, remises, and more. Nevertheless, the primary issue facing WSNs is energy management. In a wireless sensor network (WSN), non-rechargeable batteries often provide power to sensors because conventional nodes have a limited energy capacity. Energy harvesting techniques are a viable approach to address the battery-related problems with sensors. This study lowers the energy usage from idle listening and provides a workable remedy for the hotspot effect. Energy-efficient plans come in different forms, such as mechanical, thermal, wind, and solar, and they rely on the surroundings.

A novel clustering routing protocol to extend the lifetime of WSN has been proposed through this research. The research paper explains the structure of a new network model and presents an updated version of the original energy consumption scheme. This scheme can find out the optimal number of clusters for the minimization of total energy consumption. The developed algorithm follows the concept of balanced energy consumption involving the node dormancy mechanism, the energy balance strategy using the clustering routing technique, and the introduction of distance variance.

The clustering process involves a cluster head selection considering distinct variants accessible with SNs at run time such as residual energy, hop count and initial energy. Such advanced algorithm increases the energy efficiency of the WSN and thus offers better lifetime, stability, energy and throughput in the network. Moreover, remaining energy and node positioning are important in the clustering formation process. Finally, this protocol is simulated using an initial energy of 0.5 J for heterogeneous networks.

According to the achieved results, the developed algorithm can lessen the decay rate that further reduces the energy consumption of network. Also, it enhances the throughput along with the prolonged network lifetime. It is the updated LEACH version. In LEACH-C, the BS creates clusters, while in LEACH every node self-configures them into cluster.

The BS accepts the entire data about the location/energy of entire nodes implemented in the system. By performing thus, BS finds the CH numbers as well as assembles network into several clusters. Nevertheless, because of insufficient organization between nodes, the CHs number changes from round to round. In LEACH-C the CHs number in every round matches an optimal estimated value. A centralized routing method is one in which for a SNs set, BS measures the average energy of a network consisting energy level greater than average. ACH will be chosen from the nodes set to assure that nodes chosen should have adequate energy to be a CH.

The system has two sub clusters as well as they are additionally separated as the anticipated CHs number. In this manner, the nodes are consistently dispersed to assure that load is ultimately distributed. By employing the minimal spanning tree technique, the BS selects the least energy routing paths and uses CH to distribute the clustering data to every node in the network.

Although, because of centralized system, interface overhead will enhance in the CH reselection, as BS takes the reselection decision. Furthermore, each cluster will transmit the request; therefore, a high energy consumption occurred.

This paper is further organized as follows: Section 2 discusses about the related work. Section 3 explains about the routing problems in a network model. Section 4 describes the LEACH based system model followed by Section 5 demonstrating the results and discussion for both the hardware and software implementations. In last, Section 6 presents the conclusion.

This paper describes an advanced version of LEACH protocol that depends on clustering routing technique. It considers a threshold value of an optimal distance among sink and node. A node act as a forwarder node if it is placed under the optimal distance and will not contribute in clustering. The entire clusters deliver data to such nodes via cluster heads that is to be further transferred to the BS. In clustering, the remaining nodes participates according to the probabilistic approach named as LEACH protocol. This paper suggested a novel scheduled probabilistic clustering routing approach. Probabilistic approach can perform the clustering. SNs still dissipate some energy during the no transmitting phase. Sleep-wake up method can lessen the energy consumption at SNs. In this approach, SNs remain on sleep mode or inactive mode up to these have no data transmission/reception. Therefore, energy drainage because of idle listening is decreased.

Pseudo code for hot-spot effect

Step 1: Disperse sensors at random across the sensing field.

Step 2: Assume an ideal distance, Opt_Dist, for i=1 to n (number of nodes). Repetition of steps 3 and 4 is required.

Step 3: Calculate the "d" distance between the sink and node.

Step 4: Flag(i)=1 if (d<=Opt_Dist) otherwise Flag(i)=0.

Step 5: If Flag(i)=0, apply LEACH-based clustering based on the Probabilistic Approach for all nodes.

Step 6: Finish

Pseudo code for Idle listening

Step 1: Place sensors in the sensing field at random.

Step 2: For i=1 to R (number of rounds) Steps 3 through 6 should be repeated.

Step 3: Choose the cluster head using a probabilistic technique and residual energy.

Step 4: Counting the number of nodes (i=1 to n).

Step 5: Communicate data since opt_time if Node[i]!=Node[i].Sleep state; n = n+1;

Step 6: If Node[i].receiveData= true; if (Node[i].State=sleep) Node[i].State=Wakeup;

Step 7. Finish.

The energy harvesting schemes for WSN have some framework for the suggested algorithm, which are as follows:

·Implementation of a set consisting N SNs in a two-dimensional structure as A × A with large area.

·Implemented sensor nodes carry the characteristics in a random manner.

·Homogeneous techniques are used for initial energy distribution, Eo.

·Base stations as well as all the connecting nodes are in stationary states.

·Using CSMA and TDMA techniques, this model takes into account the data interface from a cluster member to CH or CH to sink node.

·Harvesting device module is connected to every sensor node.

·When the deployment is in the initial time, every SN has the equal energy amount E0.

·The energy harvesting rate of each SN varies and ranges from ½ to 0; Umax.

·All SNs have the ability to collect energy from the ambient resource or battery.

·In the majority of the rounds in this constructed model, the energy harvesting rate is lower than the energy consumption rate.

·The entire energy harvesting in WSNs operation has been categorized into the round r.

Because ambient resources are unpredictable, energy prediction is a significant challenge for energy harvesting models. Therefore, the energy harvesting techniques of WSNs mostly depend on energy resource management. Exponential Weighted Moving Average (EWMA) is one trending method for estimating the resource that generates uncontrolled energy. The EMWA method is based on the identical time slots that happen throughout the day, which are indicated by the slot duration.Such system depends chiefly on the intensity of sunlight. Kansal [63] discovered a diurnal scale graph depicting that the energy harvesting rate depends on the collected data from sunlight. Nevertheless, the EWMA prediction model also depends on the different seasonal conditions that require to maintain a historical summary. The mathematical equation for estimated energy (E) and harvesting energy (H) is shown in Eq. (9). Such represent the total by using a weighting factor ranges as 0\a\1. In this, a represents high value, while the last harvesting energy is smaller and vice versa. In this expression, n refers the time slots and d represents as present-day.

$E(d, n)=\alpha E(d-1, n)+(1-\alpha) H(d-1, n)$               (8)

Deployment Strategy

The "Atmel Sam3x8e ARM Cortex-M3" microprocessor was embedded on an "Arduino Due" microcontroller in our network [23]. For a railway control case, a related system utilizing the Xbee mote and depending on Arduino has been depicted. While numerous commercial nodes are static for the physical as well as MAC layers, working with the Arduino microcontroller offers a simple programming interface for the entire layers. The SN is connected to the system by the RF transceiver. The RF module typically functions in one of three modes: send, receive, as well as rest. More energy is consumed in the transmit mode. The medium access algorithm used may affect how much energy is used in the other modes.

Data Transmission

Every cluster head utilizes a non-persistent carrier sense multiple access (CSMA) MAC approach to broadcast an advertisement message (ADV) inside the range Rc. Such message is very small consisting the header and node id of several messages for a declaration purpose. Each non-CH node chooses the CH that requires the shortest transmission range in order to estimate its cluster for the current round. Each node transmits the signal to the join-REQ message and onward to the CH via the non-persistent CSMA MAC protocol. Following cluster formation, CH establishes a TDMA schedule for data transmission from cluster members to CH. The number of sensor nodes in a given cluster determines the TDMA utilization rate. There may have been a collision if the system had more cluster members. The cluster's TDMA scheduling can be used to prevent this collision. Sleep wake protocol can also be provided using TDMA method. It stipulates that each CH gives cluster members a certain amount of time to schedule. After finishing their work, cluster members go into sleep mode. This method can spare a node's excess energy. Another name for this approach is the duty cycle technique.

3.png

Figure 3. Small embedded scenario for WSN

Following are the hardware details of created embedded scenario (Figure 3):

·REES52 Bluetooth Transceiver

·ROBODO MO41 USB to RS232 PL2303 TTL converter Adaptor for ARDUINO Nano Raspberry Pi

·MicroSD Class 10-8 GB memory card

·9V Rechargeable battery

·REEs52 Arduino Uno R3

·ACS712 30A Hall Current Sensor Module

·Voltage sensor module (25 V)

TDMA schedule can decide the data transmission via CH after setup phase completion. Such phase is categorized as the inter-clustering, intra-clustering and sub phase communication. The CH gathers data from the entire cluster member at the time of intra clustering transmission. Furthermore, least amount of energy is used by the transmission phase. Each member of the cluster has the ability to disable radio energy up to the transmission time allotted by SNs in order to minimize energy consumption in their nodes. The CH gathers and stores data about the received signal from cluster members. Once all the data has been received, the cluster head processes it to create a single signal by compressing it. Data aggregation techniques are procedures that remove redundant and duplicate data.

Afterwards, data aggregating cluster head transfer data to the BS in inter clustering phase by utilizing the single- or multi-hop routing. Another popular protocol is fixed spreading code or CSMA that transforms the data from CH to BS. When CH needs to transfer the information, it should sense the channel first. It requires to wait if it is busy. Else, the cluster head instantaneously transmits information utilizing the base station by spreading code to the BS.

This system utilized an Arduino UNO R1 as well as a PC with Windows 7 as well as E-Prime 2.0.10.182 (Intel Core 2 Quad Q6600, 2.4 Ghz). Sequential port associations were set to 128,000 baud. The outer DAQ was a CED Miniature 1401 mk II (from Cambridge Hardware Gadgets, UK) inspecting at 10000 Hz. For this test, we utilized an Arduino Uno R3. The connection to USB is handled by an Atmega16U2 chip. The board was attached to a Sony laptop running EPrime 2.0.10.242 and equipped with an Intel Core i5 processor. For every preliminary, the preliminary length shipped off the Arduino was 100 ms. The Arduino only waited for the specified 100 milliseconds while examining the conditions of the two connected buttons as well as then transmitted back the time it really waited. For each of the following baud rates, we conducted 500 consecutive trials: 14,000, 38,400, 57,600, 115,200, and 128,000 (all set in E-Prime, the Windows instrument controller, and the Arduino code). We deducted the time the Arduino stated waiting from the response latency calculated by E-prime, which is the time among transmitting as well as obtaining from the serial port, in order to calculate the delay caused by the USB communication.

Figure 4 is for sequential flow of proposed algorithm.

4.png

Figure 4. Sequential flow of proposed protocol to overcome hot-spot effect

All through the reenactment, a 100×100 organization setup with 100, 200 as well as 300 hubs is thought about where every sensor hub is relegated an underlying energy of 0.5 J, how much transmission energy is 50 nJ/bit, communicate enhancer energy (Eamp) is 100 pJ/bit. In our reproductions, all sensor hubs are expected to do detecting activity at a proper rate as well as consistently have information to transmit when they get messages from the sink hub. It is likewise expected that entire information transmitted by the past hubs are totaled into an information fragment with a consistent size of 4000 pieces. The channel has the maximum baud rate of 38400.

Wireless sensor networks feature a tremendous scope with enormous number of applications. Several applications between these protocols are energy constraint. WSN has some prevailing techniques that have the potential to attain the energy efficiency. Clustering becomes an appropriate approach that optimize a system. The LEACH procedure is used by the clustering approach, which produces very good energy-efficient results. Although these protocols work extremely well, these systems nonetheless have some serious problems. In idle listening, energy consumption is significantly dependent. In an idle state, SNs dissipate some energy for listening data apart from data transmission. Keeping nodes in sleep mode can reduce this energy consumption during no data transmission mode. Also, when they need to transmit data, it makes them awaken. This work provides an advance form of LEACH-based probabilistic clustering approach. It is a hybrid system to solve the issues of sleep-wake up and hot-spot effect to attain the high energy efficiency. Findings depict that the developed system provides effective results with great performance. Clustering is the process of identifying a natural link between certain things or data, or the grouping of comparable objects. It is used in a variety of fields. Especially in sensor networks, clustering can be used to overcome a number of issues. It reduces the number of nodes involved in long-distance communication by using clusters to send processed data to base stations. This has a direct impact on the total energy dissipation of the system. In addition to sensor networks, data mining and VLSI-CAD have made extensive use of clustering. Clustering is used by a classical analytic VLSI placer to efficiently place standard cells. In this paper, we investigate numerous examples of the clustering problem and find optimal solutions to a few intriguing cases. There are numerous uses for these issue situations in sensor networks, particularly in energy management and master node selection.

The Watch Dog Protocol is used to address the Idle listening problem, it introduces a mechanism where nodes alternate between active and sleep states. During active periods, nodes monitor the channel for incoming data while in sleep period, nodes power gets down. To support this concept, a small embedded scenario has been created. Results show that voltage drop (With Watch Dog) at the battery occurred after a longer time than applied without the Watch Dog concept.

The paper also bridges the gap between manufacturing, design, and computing by addressing practical challenges in deploying WSNs in manufacturing environments. It offers insights into how the design of sensor networks impacts manufacturing processes, contributing to more efficient and optimized manufacturing operations. It may also discuss the hardware design considerations for sensors in manufacturing, considering aspects such as power efficiency, robustness, and communication capabilities. Overall, the paper's focus on an energy-efficient clustering approach for WSNs aligns well with the scope of covering manufacturing, design, and computing. It offers practical solutions to challenges faced in manufacturing settings, where effective sensor network deployment can lead to significant improvements in production processes and overall operational efficiency.