1. Introduction
WSNs consist of tiny sensor nodes made for data sensing, data computation, data transmission and data reception tasks. Unlike wired network scenarios, wireless sensor nodes transmit the sequence of environmental data from one location to another location via open multi-hop channels. In this case, each sensor node is intended to broadcast beacon messages to recognized neighbor sensor nodes in order to establish multi-hop channels. At the same time, each sensor node must hear the requests coming from other sensor nodes. Thus, the sensor nodes crucially spend significant amounts of energy in listening states. As the sensor nodes are vulnerable to resource limitations in real-time conditions, wireless network protocols and communication frameworks are expected with energy consideration functions.
Generally, WSNs manage various services in their protocol stack, such as mobility management, security management, power management, event management and quality management. The application-specific WSN architectures are deployed in order to provide various levels of network services. Wireless physical layer functions, WMAC functions and wireless routing protocols are the major considerations for achieving superior data transmission in WSNs. In this regard, configured physical layer parameters and nodes’ effective attributes initiate network operations. To an extent, wireless channel allotment policies (scheduled or random) are decided by WMAC procedures. Similarly, identifying optimal routes for multi-hop data communication along the channel from source node to destination node must be configured with the help of suitable wireless routing protocols. According to the deployment strategies of WSNs, the routing protocols are chosen to enable a multi-path routing process or a uni-path routing process.
Along the execution of layering functions in the network, each sensor node receives irregular energy distribution, computation load and memory utilization. Under the conventional or standard network functions, the lifetime of the WSN is unacceptably undersized. The lifetime of the WSN or sensor nodes can be increased through proper energy utilization policies and load distribution policies. These policies are expected in MAC and route management tasks. Initiating crucial experimental analysis over various WSN-based energy optimization techniques and load optimization techniques provides a new motivation for future research works. In this concern, the individual energy optimization rules established for WMAC and routing function in each node ensure the entire network’s lifetime and link availability. The importance of energy-sensitive communication protocols is seriously considered for research under WMAC and routing layer functions. As WMAC and routing jobs are more closely related to channel liveliness than other layers, the need for controlling energy wastage along the respective channel is a critical task. In the same manner, multi-path routing protocols are noted as better solutions than uni-path routing mechanisms against security threats. Apart from physical architectures, the sensor nodes require logical neighbor association rules to build usable network channels. Logical network configuration and successful data communication are confirmed through effective WMAC principles and resilient routing protocols, respectively. On the other hand, energy-efficient WMAC procedures and routing protocols are widely expected in each wireless sensor node to save the individual node’s energy. With this in mind, this experimental survey has been initiated from the study of the WSN’s characteristics, types, future-generation policies and real-time network problems. Accordingly, the major problems and solutions are described, as given in
Table 1.
As given in
Table 1, the general aspects and WSN characteristics are discussed. In addition, other research works [
17,
18,
19,
20,
21,
22,
23,
24,
25,
26,
27] discuss various resource allocation issues, routing protocols, smart sensor characteristics and energy problems of WSNs. The baseline understandings of WSNs and their application inspires researchers to focus on suitable WMAC and wireless routing strategies in real time. In the same manner, the articles used for experimental survey are categorized under WMAC and wireless routing protocols. Particularly, the related research works are classified under energy optimization policies, energy balancing policies, channel scheduling policies and machine learning techniques [
28,
29,
30,
31,
32,
33,
34,
35,
36,
37,
38,
39,
40,
41,
42,
43,
44,
45,
46,
47,
48,
49,
50,
51,
52,
53,
54,
55,
56,
57,
58,
59,
60,
61,
62,
63,
64,
65,
66,
67,
68,
69,
70,
71,
72,
73,
74,
75,
76,
77,
78,
79,
80,
81,
82,
83,
84,
85,
86,
87,
88].
The detailed literature discussions and relative observations mainly target the optimized WSN functions. Notably, energy-optimized solutions on the relative functions between WMAC and routing protocols of WSNs (channel allocation and route identification) are vastly explained in this article. Moreover, the experimental section of this article details a crucial set of energy optimization policies and load balancing policies in order to suggest better WSN strategies on the basis of WMAC and wireless routing protocols. As many research works propose resource-constrained communication protocols (channel control and route control), the need for classified results is mandatory to obtain crucial aspects through appropriate experimental conditions.
In the same manner, research problems are widely noted under energy-efficient channel management policies, medium utilization quality, energy-aware WMAC routines, node liveliness, node connectivity and optimized neighbor discovery processes. As the battery-powered wireless sensor nodes are vulnerable to unplanned energy depletions, the overall functions of the entire WSN cannot be expected as static and stable in real-time conditions. In this regard, the proposed article analyzes and compares recent works conducted regarding the issues mentioned. In addition, this article mainly finds the technical benefits, limitations and scientific facts of crucial research works accompanied by the energy-efficient WMAC principles, optimal medium utilization and energy sensitive wireless routing protocols of WSNs.
The study contributions of the article are listed, and details are given in the respective sections.
Discussing the types of WSNs, configuration details, resource management, channel rate allocation and future requirements.
Taking a comparative study and experiments on WMAC principles, channel allocation strategies and energy optimization issues.
Discoursing the functions, limitations and properties of various wireless routing protocols.
Energy optimization issues in wireless routing environment and protocol support.
Experimenting a detailed energy optimization scenario between different cases of WMAC functions and wireless routing protocol functions.
Many literature survey works are proposed under energy optimization policies for enabling feasible network communication between wireless sensor nodes. At any rate, the implications of multi-layer energy conservation policies (WMAC energy solutions, WMAC channel management, routing problems and load balancing problems) provide an integrated problem analysis and solution-making platform for future researchers. Crucially, the existing study articles consider mostly uniform energy-efficient solutions on a particular layer. The energy optimization solutions discussed on single-layer functions limits the relative energy-based interpretations between WMAC (channel allocation) functions and routing protocol functions. On the whole, the novelty and contributions of the proposed literature survey give diversified technical details on multi-layer network functions and energy considerations. On this basis, the research findings are taken in order to solve the energy optimization issues regarding wireless channel allotment schemes, WMAC protocol functions and wireless routing protocols.
The proposed review article has been classified under different sections.
Section 2 of the article consists of the technical discussions on WSN architectures, network problems, energy-sensitive WMAC strategies and energy-efficient routing protocols.
Section 3 includes practical investigations and performance comparisons between notable literatures.
Section 4 of this article provides a detailed conclusion on the review findings and future scopes.
3. Experiments, Comparative Investigations and Results
Experimenting the notable technical contributions is a challenging yet useful practice to observe the practical abilities and limitations of the developed frameworks. As illustrated in
Table 4, the related articles are taken under considerations such as energy-efficient WMAC policies and routing strategies [
84,
85,
86]. The successful development of the earlier research frameworks leads to noteworthy benefits for energy-saving plans in WSNs [
87,
88].
At any rate, this article was motivated by the intention to conduct an experimental comparison between crucial existing works developed based on energy-efficient WMAC polices, energy-efficient routing policies and multi-layer optimization policies (WMAC and wireless routing protocols), as shown in
Table 5.
Based on the illustration of
Table 5, the technical cohesions between each work deliver the implementation details and findings. In this experimental section, E1 [
28] is denoted for modified MAC protocols for managing energy distributions in the critical WSN environment. Notably, this work analyzed weak signal detection through CSMA/CA principles (CSMA/WSD). In this regard, this work found the variants of MAC models and the MiXiM-OMNet++ environment. On this basis, the proposed CSMA/WSD was implemented to classify weak signals and channel collisions. Under this mechanism, each packet loss event was evaluated, as given in
Figure 3.
In this process, the random determination of packet loss under collision state is evaluated for weak signal conditions. On successful detection of weak signals, the proposed CSMA/WSD helps to re-tune the data rate and finds handoff possibilities. Thus, the systems reduce the number of collision-based retransmissions and save the sensor node’s energy. In the same manner, E2 [
30] implemented a memetic algorithm–meta-heuristic suit for improving the ability of MAC-based sleep and wake operations. The proposed memetic algorithm implemented five steps to identify the possible number of active nodes in sensor networks to avoid the failure rate of data transmissions. The optimal selection of an active node on the wireless channel minimizes excessive consumption of energy.
According to the model, the active sensor node selection process is illustrated in
Figure 4. Notably, the determination of this memetic approach considers the sensor node’s coverage quality and connectivity factors, the remaining energy in the sensor node and the duration of sleep–wake periods.
Figure 4 shows the modified genetic algorithm-based memetic node selection approach through population vectors and solution vector computations. In this case, population vectors are computed based on successful sensor node identification marks. To an extent, the solution vectors are computed on the basis of available nodes around consecutive neighbors. In this solution vector, each node’s neighbors are updated as child entries to create possible active channels. Under this MAC-based memetic approach, coverage costs, connectivity costs and energy costs are identified with maximum quantity to wake up the nodes from sleep mode. In this connection, each sensor node has been identified with possible solution vector entries (covering sleep nodes), coverage crossover points and sleep–wake mode factors. With this in mind, sensor nodes are searched to enable active node communication channels regularly. This work stated that the modified memetic approach reduces channel allocation time and energy consumption rate for channel allotment.
Comparing E1 and E2, both works minimize the cost of packet retransmission through the proper handling of collision cases and node availability issues, respectively. In the same manner, E3 [
37] provided the integrated channel management and routing solutions using energy-optimized underwater sensor networks. Compared to other types of WSNs (radio frequency), underwater WSNs use acoustic communication models. Acoustic signaling models are vulnerable to significant data loss, maximum propagation delay, restricted bandwidth, limited energy provisions and channel distortions. In comparison with E1 and E2, E3 takes critical channel (water medium) characteristics to establish energy optimization solutions. In the implementation phases, E3 found depth modelling procedures for numerous acoustic sensor nodes. In this work, the AUVNet simulator toll was used to observe the benefit of energy-efficient underwater MAC and routing protocols against other techniques such as the focused-beam routing protocol, distance-aware collision avoidance protocol and cluster-based protocol. In this scheme, multi-path clear-to-send (CTS) and request-to-send (RTS) packets are shared among the acoustic sensor nodes. On the basis of acoustic sensor placement, the score of any node was computed with respect to node distances (
and energy levels (
. The score values are calculated based on Equation (1).
To a certain extent, the appropriate node-level computations are useful for finding multi-path channels to route the packets. In the multi-layer network protocol tuning process, E4 [
39] developed quality of service (QoS)-MAC assisted multi-path routing protocols and data priority computation schemes. This work produced cross-layered packet analysis procedures, priority evaluation procedures, channel listening activities and flexible routing principles according to channel quality metrics. On this basis, this scheme used AODV protocol and QoS-MAC principles under a single point of concern. According to the system design, this work embedded a packet priority field in WMAC frames. This field was pointed with four priority levels, as shown in
Figure 5.
In this concern, E5 [
50] and E6 [
57] proposed load balanced sequence scheduling and a routing protocol, respectively, for reducing the sensor node’s energy consumptions. In E5, each sensor node adaptively used the internal buffer to process the data on demand. In multi-hop communication, each sensor node is activated through aggressive scheduling-based MAC models to hold the time division multiple access (TDMA) slots. The even distribution of load among sensor nodes shall forward the channel data using minimal requirement margins. At the same time, these TDMA slots were occupied by multiple sensor data streams. Similarly, E5 proposed an energy-efficient collection tree protocol for low-powered WSNs. The protocol functions executed in each sensor node collected and distributed the data streams based on energy-sensitive tree-like paths. In particular, WSNs are categorized under a low-powered wireless personal area network standard due to their resource-limited environment. In this regard, both works suggested restricted channel allocation policies and low-powered routing mechanisms in WSNs. Notably, E6 compared energy-efficient collection tree protocol with lossy routing protocols and conventional routing protocols.
Figure 6 illustrates the functions of E5.
In addition, E7 [
75] and E8 [
79] produced notable solutions on energy-aware wireless routing protocols. Particularly, E7 generated trust value computation techniques using an average packet delivery ratio, route reply ratio, residual energy rate and number of retransmissions on the channel. In this scheme, local trust cost and global trust cost were computed for each sensor node. Local trust values of the sensor nodes were computed in the node itself. At the same time, global trust values were computed at border router points to optimize the data transmission rate. Under this trust evaluation scheme, each trusted sensor node formed a channel to avoid any discrepancies during the data transmission period. This practice reduced the number of retransmissions and packet drops on the channel. Thus, the cooperative routing methodology manages all active sensor nodes under trusted communities (
Figure 7). At the end, E8 discussed the particulars of routing protocol issues, connectivity problems and energy control mechanisms in detail. In the experimental sections, this work compared the AODV routing protocol and stateless real-time routing protocol for ensuring energy optimization qualities.
According to the technical observations, WMAC customization and channel allocation strategies [
28], sleep and wake strategies [
30], cross-layered implementations [
37], priority-based channel allocation [
39], balanced WMAC/routing solutions [
50,
57] and adaptive routing models [
75,
79] are observed for effective comparison practices. The diversified nature of the restricted choices from existing energy-saving policies provides the best understandings under experimental testbed conditions. On the execution of experimental study practices, this work was implemented using tools such as Network Simulator (NS-3.35) and the tool command platform.
The scenario of the WSN is created with a maximum of 300 sensor nodes around the geographical area (1000 m × 1000 m). Generally, NS-3.0 supports deploying more than 300 sensor nodes. At any rate, this experimental survey sets an assumption of 300 sensor nodes around a 1000 m2 area, which is significant in terms of network population. Additionally, the network configuration sets a mobility model for the sensor nodes in the prescribed region. The real-time deployment for this type of sensor network (300 sensor nodes with mobility features around a 1000 m2 area) provides enough challenges for data transmissions, energy harvesting schemes, channel management and route establishment tasks. NS-3.0 is a network scenario creator with simulated configurations of WSNs.
The network scenario has been assumed with the heterogeneous nature of sensor nodes where the internal components of nodes vary in terms of energy, transmission range and mobility constraints. In this regard, node characteristics such as initial energy (joules), transmission energy (joules), receiving energy (joules) and node coverage abilities (meters) are differently configured for each sensor node in the network (
Table 6).
Table 6 illustrates the implementation details of WSNs in the NS-3.0 environment. Consequently, the supportive packages of Python and C++ were used to implement the existing techniques. The performance metrics for evaluating the exiting techniques illustrated in
Table 4 (E1, E2, E3, E4, E5, E6, E7 and E8) are average energy consumption rate (joules), liveliness rate, successful data delivery rate, number of retransmissions (count), computational overhead (%), routing delay (milliseconds), scheduling time (milliseconds), packet drops due to downtime (count) and energy optimization rate. The definition of each performance metric is given as follows:
Average energy consumption rate (AECR): The average amount of joules spent by a sensor node throughout data transmission, collection and idle listening modes.
Liveliness rate: The availability rate of active appearance (data transmission, collection and idle listening) made by each sensor node against expected lifetime.
Successful data delivery rate (SDDR): The ratio between the quantity of packets delivered successfully by a sensor node against the packets dropped by the node.
Number of retransmissions: Total number of packets retransmitted by a sensor node during a simulation cycle.
Computational overhead: The excessive amount of packets (control messages, retransmitted data and other recovery messages) processed against the average number of network packets processed.
Routing delay: The time taken by the routing protocol to find the optimal route and deliver the data to the destination.
Scheduling time: Time taken by WMAC scheduler to make the unique channel period for sending the data through the multiple access medium.
Packet drops due to downtime (PDD): Number of packets dropped by an inactive node due to failure.
Energy optimization rate (EOR): The ratio between the amount of joules spent using the energy optimization policies and the amount of joules spent without using optimization policies.
The existing works are investigated under variable constraints as given in
Table 5. According to that base, the routing protocols are chosen as AODV and AODV–link-state (LS) models. Similarly, the traffic characteristics, mobility, coverage and energy levels of sensor nodes are configured as variable at different sensor nodes. In addition, the signaling models configured in this experiment are built with the functionalities of both electromagnetic and acoustic nature (underwater/underground).
The experiment starts with the performance validations of the existing techniques (
Table 4) using the metric AECR of each sensor node. In this experiment, the cross-layered techniques (WMAC and routing) consume minimal AECR compared with other single-layered solutions. In this concern, the observation has been conducted with the minimal AECR of E3, E4 and E5. The existing works E3, E4 and E5 consider the impactful factors of medium, power limitations and equilibrium in load distribution. As these works are developed to consider multi-layered network functions (MAC and routing issues) with evenly distributed load management policies, they produce optimal AECR between 1.75 J and 1.85 J. At the same time, the AECR of E2, E6 and E7 fall closely under E1. The reason behind this observation is that the specified approaches consider energy optimization as the main problem. In this regard, E2 used heuristic MAC management strategies to effectively organize the data transmission slots (sleep and wake up node selection). In contrast, E6 and E7 focused on balanced load management on routing procedures to reduce the AECR.
In this observation, each existing technique initiated various energy optimization solutions regarding MAC principles or routing protocols. At any rate, the successful engagement of MAC and routing protocol principles of E3 assures minimal AECR. At the same time, other techniques experience a slight hike in AECR where the number of sensor nodes is 300. The variations among the techniques are not huge in AECR, yet the range between 1.57 J and 1.95 J shows significant impacts in resource-limited sensor nodes and network lifespan reduction (
Figure 8).
Figure 9 describes the average liveliness (active) rate of sensor nodes around the WSN. As the network population is increased in the prescribed geographical area (1000 m × 1000 m) due to the increasing number of nodes, the frequency of a node’s activity increases to manage the neighbor discovery process, data transmission, data collection, route updating process and idle listening process. Hence, each sensor node consumes more energy to accomplish the requested tasks and gradually falls at the critical stage of residual energy.
At this point, the need for energy optimization techniques is essential to keep the node live to handle the data transmissions in a dense network field. The illustration given in
Figure 9 implicates the fall of the average active conditions of sensor nodes in the network. The proven performance of cross-layered techniques (E3, E4 and E5) shows a better liveliness rate from 0.93 to 0.85 as the number of sensor nodes increases from 50 to 300.
In this experiment, E3, E4 and E5 diversely manage their MAC principles by considering channel quality metrics and network dynamics. Significantly, E3 managed both timeline-based channel allocation and route consistencies throughout the increasing number of sensor nodes. In the same way, E5 developed the load distribution and aggressive scheduling procedures. Based on these reasons, E3 and E5 compactly maintained the overall node liveliness rate better than other works. In the next level, E4 achieved an even better liveliness rate (0.83) under a highly populated WSN (300 sensor nodes). In this experiment, other techniques found active node selection procedures (E2-0.81 and E6-0.82) and load-optimized energy control procedures, respectively. This kind of practice improves network lifetime and the active state of nodes (liveliness rate). By contrast, other existing techniques such as E1, E7 and E8 hold the sensor nodes in an active condition for a more limited period of time than the expected case (20 to 25% of limited downtime). These existing techniques mainly concentrate on the WMAC-based energy efficiency than the node’s overall behaviors (routing, advertising, discovery and listening).
From the observations, this article classifies the cross-layered energy-efficient strategies from other techniques, as given in
Figure 10. The performance of each cross-layered technique is evaluated using SDDR for multiple data sessions. As denoted, the data communication sessions are populated (20 to 220) and executed for multiple test cycles. Let us assume the number of sessions increase as the number of sensor nodes increases to handle the increasing rate of throughput. In this comparison, the integrated principles of WMAC, routing assistance and scheduler policies of E3 ensure a better SDDR than other works. In any event, E4 manages the SDDR at a higher rate than E3 (94.7%) during initial sessions. The priority calculation and QoS management tasks of E4 give optimal results in SDDR around the network. In contrast, the stability in data delivery needs proper organization of routing strategies at any cost. Accordingly, E3 attains optimal SDDR during the moments of more populated sessions than other works.
During this moment, the performance of E5 starts at the lowest SDDR (94.4%) during initial sessions and manages with the average performance between E3 and E4 as it is modelling both balanced power assumptions and balanced scheduling possibilities. Finally, E5 achieves the SDDR of 93.5% against densely populated sessions. Consequently, the changes in the number of retransmissions are crucially noted for the existing techniques such as E3, E4 and E5. The number of retransmissions is closely connected with SDDR and the node’s liveliness rate. The SDDR is indirectly proportional to the retransmission rate and directly proportional to the liveliness rate. According to that, E3 produces the minimal number of data retransmissions from 45 to 95 (number of packets) as the increasing number of sessions at each test cycle. In the same manner, the retransmission rate of E4 reaches a higher point for the maximum number of sessions (116). Similarly, E5 produces a moderate load in data retransmission compared to other works (between 55 and 105). The implications of data retransmission for varying number of transmission sessions are given in
Figure 11. In the same manner, E6 shows notable contributions in both routing and WMAC policies. E6 achieves balanced routing and medium management policies in WSN. In this case, the observations of E6 performance using AECR, liveliness rate and SDDR are denoted in
Figure 8 and
Figure 9. The effort of E6 on these metrics is crucial after the implications of E3, E4 and E5.
With this in mind,
Figure 12 shows the comparative cases of E3, E6, E7 and E8 with respect to computation overhead. In this association, the assumption on the computation overhead of the existing techniques E1, E2, E4 and E5 are directly mapped with the number of retransmissions taken for each session. In any event, the comparison given in
Figure 12 relates the existing techniques E3, E6, E7 and E8 in terms of routing behaviors. These techniques commonly achieve energy-optimized routing solutions in WSNs. The efficient and energy-optimized routing models need to produce minimal computation overhead irrespective of the node’s location. Additionally, the processes involved in each sensor node vary depends upon the mobility of sensor node.
The challenges taken under mobile sensor nodes lead us to solve new problems such as link breaks, node failures, data retransmission, frequent route updates and energy loss. In this preference, the computation overhead of related routing principles are denoted in
Figure 12 against the changing velocity of sensor nodes. As the sensor nodes’ velocity changes from 15 m/s to 45 m/s, the excessive load initiated in the sensor node’s processor slowly increases and it increases energy AECR definitively. The comparison with energy-optimized routing practices is noted in terms of the excessive number of processes raised in the distributed environment. As discussed, the additional processes in each sensor node are created as a result of managing real-time issues such as node failures, link breaks, excessive route updates, etc.
At this point, the computation overhead of E8 varies from 210 packets to 321 packets over the changing velocity of sensor nodes. Similarly, other techniques such as E7 have 290 packets as the computation load in the front end for a maximum velocity of 45 m/s. On the other hand, E6 produces a rate between 175 and 270 of overhead. Among these techniques, E8 was developed for the hierarchical routing model and it is not flexible with random networks. Hence, an excessive load in E8 is observed. In this case, E6 and E7 target load balance and cooperative routing procedures, respectively. The comparison between E6 and E7 shows the better contribution of the cooperative routing technique (E7) in computation reduction. As E7 uses an authentic and distributed cooperative model for computing twin-trust costs (local trust value and global trust value), the elimination of irregular nodes is easy in the network. In addition, the excessive packet transmission to unethical nodes or inefficient nodes is ignored in the network. At any rate, the performance of E6 directly deals with low-powered computation procedures in order to limit transmissions.
In this manner, the technical pitches and purposes used for establishing these existing techniques make notable performance variations. In this comparison, E3 falls in both channel and efficient routing practices in order to reduce the overloaded tasks at different layers. Accordingly, the production of computation overhead in E3 is maintained between 170 and 250 which is the minimum compared to other techniques. The experimental contributions of this article are extended to analyze the practical betterment of routing delay (routing protocol) and scheduling time (channel allotment). The motive of understanding the routing-based research techniques and channel slot management techniques leads to separate comparative observations.
Hence, the implications are noted as given in
Table 7 and
Table 8, respectively.
Table 7 shows the performance of energy-efficient routing strategies (E3, E6, E7 and E8). As discussed, E3, E6, E7 and E8 mainly focus on the implementation of energy-efficient routing protocols, and routing delay calculation is an important task. With this in mind, the successful elimination of overhead and retransmission leads to minimal routing delay (milliseconds (msec)). Thus, the performance of E3 is optimal in terms of routing delay production (maximum 553 msec) during the change in the node’s velocity. In this experiment, this article takes the node’s velocity as changing parameter to validate the routing delay. Since the dynamic velocity rate (meter/seconds, m/s) increases the possibilities of node failures and link breaks continuously in the network, the routing delay produced from each protocol varies rapidly. In this case, E6 produces a moderate routing delay compared to other techniques such as E7 and E8. The existing techniques E7 and E8 deal with secure and hierarchical routing models; therefore, the production of routing delay is higher than E6 (load balanced and low-powered scheme).
Similarly,
Table 8 gives the identifications of channel scheduling strategies (E1, E2, E3 and E5) and their performance. As E3 is noted, it is a cross-layered solution for energy optimization in WSN, and it is not effective in scheduling processes. At the same time, the development of E1, E2 and E5 are channel allocation and timeline scheduling tasks to reduce frame allocation time on the channel. In this regard, E1 and E2 effectively process the timeline slot for scheduling through modified MAC policies and memetic–heuristic scheduling policies, respectively. These techniques perform better than aggressive scheduling policies in dynamic WSNs (E5).
Table 9 illustrates the routing benefits of using a standard AODV protocol and hybrid AODV protocol with link-state features. This work identifies the limitations of standard AODV as frequent updates and overhead during network changes. Generally, an AODV protocol performs optimally for reactive updates yet takes maximum overhead for a more dynamic network. At the same time, AODV takes maximum time to update the route information for the whole network regularly. The implementation of both link-state routing models and distance vector models gives fast route updates and reactive route updates, respectively. Especially, the varying velocity of sensor nodes affects the established wireless links frequently. In this case, the uni-protocol system struggles to initiate either global updates or frequent neighbor updates. However, the idea behind AODV-LS supports both reactive and proactive updates (global updates and frequent neighbor updates).
Figure 13 compares the efforts of all existing techniques using the metric as the number of packets dropped due to energy shortage (downtime) at a sensor node. This is an important evaluation that relates the amount of residual energy maintained by the node and the active participation of that node itself in data communication. Apart from the continuous oscillations in packet drops produced by each technique, the optimal results are observed for E1, E3, E4 and E5 under various test cycles. As noted in
Figure 13, the existing techniques E3 and E4 attain the drops between 35 and 60 as optimal compared to other techniques. At the end, the EOR was taken for performance analysis of all experimented techniques. According to this observation, the higher rates of E2, E3 and E4 show the suitable nature of energy optimization in WSN. On the scope of energy-efficient routing methodologies, many recent works are developed in the research arena [
89,
90].
Finally, the practical evaluations are crucially mounting on the considerations of packet drops due to the node’s downtime and EOR (
Table 10). EOR is the definite attainment of each work towards an energy optimization goal. In this experiment, better energy harvesting (saving) solutions are produced from E3, E4 and E5. Particularly, the effective reduction in computation overhead and retransmission rate gives optimal attainment in EOR. Thus, the EOR of E3 varies between 0.365 and 0.467 during various iterative simulations. In the same manner, E4 attains an EOR from 0.321 to 0.563 as it is using a channel adaptive quality evaluation procedure to initiate data transmission compared to E3. On the next level, E5 attains a better EOR (0.443) due to its load balancing principles compared to other works. In addition, the practical comparison between more recent energy-efficient routing protocols gives diversified solutions to the research community. In this regard, this proposed review article extends the evaluation of E9 [
51], E10 [
54], E11 [
55] and E12 [
56] under the considerations of energy optimization and multi-hop routing principles in WSNs.
The most recent ideologies on novel energy-optimized routing protocol development and relevant discussions lead to future-generation WSN energy models [
91,
92]. In this concern, E9 developed swarm-intelligence-based chimp optimization solutions and hunger game searching principles to find energy-efficient multi-hop routing paths. This work followed the natural habits of chimps to optimize path-finding problems with minimal overhead. Particularly, the first phase of the chimp optimization algorithm initiated the network formation under a hierarchical structure (base station and clusters). The clusters were formed as driving nodes, chaser nodes, barrier nodes and attacker nodes using chaotic cost computations. In the same manner, the second phase of this work provided a hunger search-based path selection approach for initiating an energy-efficient multi-hop routing process in underwater sensor networks.
E10 proposed energy classification and channel assessment techniques using a greedy approach for minimizing the overload of the node’s energy resources. According to the strategies, each sensor node’s energy levels are monitored with adaptive internal buffer management policies on the reception of data packets. Similarly, the routing protocol used for this mechanism found the greedy-based route selection with sufficient node resources to avoid packet losses. In this connection, E11 and E12 considered coverage problems and locality problems, respectively. Particularly, E11 proposed link stability evaluation protocols and grid-level stimulated network models to achieve coverage optimization in WSNs. In this concern,
Figure 14 illustrates the functions of link stability evaluation and coverage problem analysis models (holes or inactive nodes).
Finally, E12 proposed zone-based routing protocols and energy optimization policies throughout the WSNs. In this concern, each sensor node was constructed under various zonal locations with allotted energy resources to be associated with other zonal sensor nodes. The proposed routing protocol running in each sensor node evaluated residual energies to proceed data routing into other zones. Thus, various recent routing protocols were proposed under energy consideration platforms. These works are compared as illustrates below.
Figure 15 depicts the performance comparison between E9, E10, E11 and E12 in terms of AECR against changing number of sensor nodes. In this experiment, four different types of energy-efficient routing protocols are compared. E9 has been experimented for underwater sensor networks. Compared to other sensor networks, underwater sensor networks use acoustic sensors with minimal energy resources. In this concern, E9 developed hierarchical energy optimization solutions and chimp optimization policies in order to search active nodes to enable flawless communication. In this case, the network clustering process and node searching processes consume significant energy (1.93 J). On the other hand, protocol implemented in E12 consumes maximum energy as it is related to zonal computational policies. At any rate, E10 and E11 are optimally designed for improving the quality of energy-saving mechanisms via the greedy approach and link stability validation approach.
In this comparison, E11 ensures network coverage optimality and link stability concerns to operate successful data transmission from source to destination. In this analysis, E11 provides more stable and optimized channel circumstance to reduce packet drops and retransmissions. Thus, the AECR of E11 attains a minimal value (1.77 J). At the same time, E10 secures 1.88 J which is better than other works such as E9 and E12. Similarly,
Figure 16 shows the calculation of the average routing delay in milliseconds.
The routing delay of E10 and E11 are minimal compared to E9 and E12. The category of E9 and E12 fall under clustered or zonal network architecture. Under these network management policies, the route construction to deliver the data from source to destination has to follow clustered or zonal rules. This makes the routing delay a little higher than flat network architectures (E10 and E11). In this regard, E11 strongly builds stable links and coverage assurance in the WSN for data transmission. Once this process has been successfully developed, it reduces the routing delay during data transmission. Thus, it minimizes the delay (495 ms). In the same manner, E10 produces 510 ms of routing delay, which still better than E9 and E12.
Figure 17 illustrates the computation overhead of E9, E10, E11 and E12. As discussed, E9 and E12 produce more computation overhead (additional packet transmission) than E10 and E11. The observed results of E9 (145) and E12 (170) are closely related to AECR produced by each technique.
In contrast, E10 and E11 managed the computation overhead with minimal rates compared to E9 and E12. The reason for the optimized overhead of E10 and E11 is the energy stability and link stability in the network. In this manner, E10 produces 140 additional transmissions and E12 produces 110 additional transmissions in order to stabilize network communications. The observations are gathered against the number of actively participating sensor nodes for each iteration. Finally,
Figure 18 illustrates the average packet delivery ratio (PDR) attained by each work. In this experiment, E10 and E11 obtained 0.93 to 0.94 of average PDR for a maximally populated network (300 sensor nodes). On the other hand, E9 and E12 sent packets at the PDR of 0.9. In general, the difference in PDR produced by each system is not crucial, yet the delay and energy consumptions significantly vary due to network changes. Hence, the comprehensive experimental analysis and technical discussion described in this article clarify the crucial efforts of energy-efficient WMAC policies and routing strategies to extend the lifetime of sensor nodes.