SEECR: Secure Energy Efficient and Cooperative Routing Protocol for Underwater Wireless Sensor Networks

Underwater wireless sensor networks (UWSNs) is an emerging technology for exploration of underwater resources. Security plays an important role in the UWSNs environment because the environment of UWSNs is prone to different security attacks. This research proposes SEECR: Secure Energy Efficient and Cooperative Routing protocol for UWSNs. SEECR comprised of energy efficient and strong defense mechanism for combatting attacks in underwater environment. SEECR exploits cooperative routing for enhancing the performance of network. Considering the resource constrained UWSNs environment minimum computation is employed for implementing security so that SEECR remains suitable for underwater environment. In order to evaluate the performance of SEECR, this research compares the performance of SEECR with AMCTD: Adaptive Mobility of Courier Nodes in Threshold-optimized DBR - a well-known routing protocol for UWSNs environment. The performance of SEECR and AMCTD protocols are evaluated using different performance evaluation parameters such as number of alive nodes, transmission loss, throughput, energy tax and end-to-end delay. The results suggest an improved performance of SEECR over AMCTD. SEECR shows an improvement of 9% in terms of number of alive nodes, over 50% reduction in terms of transmission loss, up to 9% increase in throughput, up to 23% reduction in energy tax, and 25% reduction in end-to-end delay. Further, we observe that attack significantly degrades the performance of AMCTD whereas due to the embedded defense mechanism in SEECR the impact of attack is negligible.


I. INTRODUCTION
Underwater wireless sensor networks (UWSNs) consist of sensor nodes which are deployed under the water in order to sense the properties such as temperature, pressure and quality etc. [1]. The environment of UWSNs is very dissimilar from terrestrial wireless sensor networks (WSNs) in many ways. The electromagnetic signals cannot travel for long distance in UWSNs due to the high attenuation, scattering and the absorption effect [2]. In order to address this issue acoustic waves are used in UWSNs environment [3]. The propagation speed while using the acoustic waves in UWSNs environment is 1500m/sec which is much slower as compared to the radio The associate editor coordinating the review of this manuscript and approving it for publication was Longxiang Gao . waves based communication. In acoustic communication the end-to-end delay is high along with long propagation latency. The available bandwidth is very limited in acoustic communication (less than 100kHz) [2], [4], [5]. The sensor nodes in UWSNs environment are mostly considered static but due to the underwater activities these sensor nodes can move from 1 to 3m/sec. The sensor nodes in UWSNs environment are large in size and they consume more power due to which efficient utilization of energy is a critical factor in UWSNs communication. Charging and replacing of batteries in UWSNs environment is a challenging task [2], [6], [7]. The basic architecture of UWSNs is shown in Fig. 1.
The sensor nodes are deployed in UWSNs environment as shown in Fig. 1. The communication among sensor nodes under the water is through acoustic waves whereas the sink VOLUME 8, 2020 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ node communicates with on-shore server through radio communication.
Applications of UWSNs include pollution monitoring, temperature and pressure measurements, environmental monitoring, seismic monitoring, assisted navigation, unmanned underwater exploration, submarine navigation, underwater disaster detection and prevention, underwater exploration such as corals, minerals and rare metals etc. [8]- [12]. The challenges in UWSNs environment are deployment, fabrication, maintenance and recovery cost is very high as compared to the terrestrial WSNs. Other challenges include power harvesting and power consumption, localization, deployment of sensor nodes, time synchronization, battery lifetime, underwater data collection, replacing or repairing sensor nodes, capacity of communication channels and security etc. [8], [13].
The basic security requirements in UWSNs are authentication, confidentiality, integrity and availability [14], [15]. There are different attacks possible in UWSNs which can abrupt the normal operation of the networks. The most common attacks include black hole attack, wormhole attack and sinkhole attack. Attacks in UWSNs environment can be grouped into the following two categories. i) Attack on the sensor node in UWSNs ii) Attack on the protocol in UWSNs. The first type of attack damages the sensor node but such type of attack is least likely in the real environment because it is very difficult to physically access the sensor nodes in UWSNs environment. The second type of attack target the communication protocols used in UWSNs. Once the communication protocol is compromised then it affects the entire network [13]. There are security requirements and countermeasure mechanisms to combat these attacks in UWSNs [15].
In this research work, SEECR protocol is proposed that is secure, energy efficient and cooperative in nature. SEECR efficiently utilizes the energy of sensor nodes for maximizing their life time. Cooperation technique used in SEECR has vital role in efficient utilization of energy. Moreover, security mechanism has been incorporated in SEECR for combating security attacks. Security mechanism involves minimum computation considering the resource constrained environment of UWSNs. SEECR detects and eliminates all those active attacks which drop the packets.
The rest of the research paper is organized as follows: Section II discusses the current state of the art. Section III includes motivation and contributions. Section IV discusses underwater channel. Section V includes detail of the proposed routing protocol (SEECR) whereas Section VI includes details of simulation environment and performance evaluation parameters. Results are presented and discussed in Section VII. Section VIII concludes the work and presents directions for future work.

II. LITERATURE REVIEW
Energy efficiency is a key consideration in wide range of applications [16]- [18]. The authors in [16] proposed a novel energy efficient key agreement mechanism for UWSNs. For reducing the communication overhead between sensor nodes UWSNs are gathered into the clusters. The key agreement mechanism proposed in this research can resist different security attacks which include replay attacks, sybil attack, spoofed attacks and node replication attacks etc. In the proposed mechanism geographical information and identity are added to the public as well as the private key of the sensor node in UWSNs in order to boost the capability of the mechanism to resist against different attacks. The results of the simulation revealed that the key agreement mechanism proposed in this research produced better results in terms of security as well as networks performance. Using the proposed mechanism the UWSNs can have high network connectivity and ensures that the energy consumption of the sensor node and cluster-head node is under 10J and 20J respectively.
The authors in [19] proposed key distribution scheme for peer to peer communication in mobile UWSNs environment. Two different mobility models are used in this research which are meandering and nomadic mobility model. In both the mobility models, structure used is hierarchical structure and communication is managed through distribution scheme known as Blom's key distribution scheme. Simulation results of this research indicated some minor connectivity issue due to mobility but the proposed scheme resolves connectivity issues on time. The key distribution scheme proposed in this research revealed better resiliency performance when some nodes are captured by the adversary and in that case very few links become compromised. Simulation results also revealed better performance of the proposed scheme in terms of security and energy consumption.
The authors in [20] proposed cryptographic mechanism for UASNs. The authors proposed encryption algorithm which is efficient in nature for protecting the integrity and confidentiality while considering the unique characteristics of UASNs environment. In this research some modifications are made in traditional AES-128 by using alternate approach instead of using S-Box. The proposed algorithm can resist against brute force as well as other attacks. It has been concluded in this research that the round key is not breakable using brute force attack. The performance of proposed cryptographic scheme is compared with AES-128, Blowfish and PRESENT. Simulation results revealed that as compared to AES-128 and other cryptographic schemes the proposed cryptographic approach is secure and energy efficient.
The authors in [21] proposed a design of secure routing approach for UASNs. The researchers in this research proposed an efficient signature schemes with no need of the online trusted third party. The scheme proposed in this research can tackle forgery attacks and improves the overall security. Moreover, a trapdoor scheme is presented which is used in routing messages for achieving anonymity among communicating nodes. The trap door design in the proposed scheme evades the overhead involved in maintaining huge number of pre-shared keys. The proposed routing scheme is compared with geographical information routing protocol based on partial network coding (GPNC) and level based adaptive geo-routing (LB-AGR). The performance parameters used are energy consumption, throughput and PDR. The results obtained from the simulation in this research revealed that the proposed routing scheme progresses security and the performance of the network become moderate.
The research conducted in [22] proposed distributed approach for combating routing attacks in UWSNs. The proposed approach can detect sinkhole attack and wormhole attack in UWSNs environment. Two different phases are used in the proposed approach such as silent monitoring phase and detection phase. According to the proposed approach of detection and mitigation each sensor node in UWSNs has to keep track of its neighbor by overhearing the messages sent and received by the neighbor. An analytical model is presented in this research in order to capture the interaction among different parameters. Simulation has been used for implementing the idea presented in this research and the obtained simulation results revealed that the suggested approach is efficient and correct.
The authors in [23] conducted detail study on denial of service (DOS) attack. In mobile UWSNs the DOS attack can be categorized into man in the middle attack, flooding attack and demolishing attack. The possible attacks in UWSNs environment are sybil attack, wormhole attack and selective forwarding attack. The influence of mentioned attacks on UWSNs is analyzed. They surveyed secure localization techniques for terrestrial WSNs. Simulation results revealed that there is dissimilarity in the performance of mobile UWSNs and mobile WSNs which indicates that the security mechanism suitable for WSNs are not appropriate for UWSNs environment.
The authors in [24] considered the requirements as well as security related issues of UWASNs. In this research it is proposed that same key should be utilized by both the sender and the receiver for the purpose of encryption and decryption as the size of the symmetric key is less as compared to the size of the asymmetric key. When all headers are added the entire message is entered in the algorithm known as message-authenticated code having a secret key which is shared. The message integrity code (MIC) which is an output value is attached to the entire message at its end. The purpose of the MIC is to ensure that each and every one bit of the complete message as well as the shared key is authentic. The encryption algorithm encrypts the message as well as the MIC. Receiver side calculates its MIC and compares it with the MIC received from sender side. If both the MIC are same the receiver accepts the message otherwise discards the entire message. The authors in this research revealed that minimum amount of overhead should be added to the data when applying security in UWASNs. They proposed to use cryptographic module validation program (CMVP) algorithm for the said purpose.
The research conducted in [25] proposed a secure neighbor discovery wormhole resilient routing scheme for UWSNs. This research used RIPEMID-160 and direction of arrival estimation as authentication mechanism for the purpose of secure neighbor discovery. The proposed scheme can resist against wormhole attack, impersonation attack and route poisoning attack. The scheme proposed in this research uses four agencies such as security agency, routing agency, underwater gateway agency and vehicle agency. Simulation results revealed that the secure routing scheme proposed in this research produced better results as compared to basic neighbor discovery schemes in terms of route maintenance overhead, consumption of energy, PDR and detection of failure.
The authors in [26] offered a secure MAC protocol for UASNs environment in order to efficiently manage the environment of UASNs regarding energy efficiency, data reliability, authenticity, confidentiality of data, and the prevention against attacker. In this research cryptography algorithm is used in MAC protocol in order to provide security services such as data authenticity, confidentiality of data and protection against replay attack. CCM-UW mode based on AES and ARIA algorithms are used in order to carry out this research. The proposed MAC protocol is efficient in terms of consumption of energy and transmission time. The proposed MAC protocol for UWSNs is compared with existing MAC protocols. The results revealed that the proposed MAC protocol is secure and efficient as compared to the existing solutions.
The authors in [27] evaluated different schemes of digital signature for the purpose of end-to-end authentication in UWSNs environment in terms of consumption of energy. The results obtained in this research revealed that the schemes which perform better in WSNs do not necessarily perform better in UWSNs due to the exceptional characteristics of UWSNs environment. Different characteristics of digital signatures schemes are identified in order to make them suitable for UWSNs. Three different digital signature schemes are evaluated for different UWSNs scenarios in this research in terms of power consumption. These schemes are: i) zhangsafavi-naini-susilo (ZSS), ii) elliptic curve digital signature algorithm (ECDSA) iii) boneh-lynn-shacham (BLS). The signature generation time for ZSS is 229ms, ECDSA is 134ms and BLS is 302ms. The signature size of ZSS is 21 bytes, ECDSA is 40 bytes and BLS is 21 bytes. It has been revealed in this research that the use of aggregate and short signatures can effectively increase the energy efficiency in UWSNs.
The authors in [28] proposed a secure neighbor discovery scheme for UASNs. This research proposed a suite of protocols which performs secure discovery of neighbor resilient against wormhole attack in UASNs environment. The proposed protocol utilizes the direction of arrival (DoA) signals approach. The proposed protocol can resist against wormhole attack with very good probability and without the requirement of traditional hard requirements. The proposed solution in this research comprises of four protocols. Protocols such as B-NDP and MA-NDP are appropriate for UANs environment with little density and applications in which end-to-end delay and connectivity are of the main concern. On the other side protocols such as SDV-NDP and DV-NDP are appropriate for those applications in which there is high density of node and extraordinary requirements for the resilience of wormhole attack.
The research conducted in [29] proposed a security suite for UASNs comprised of both static and mobile sensor nodes. The proposed security suite is comprised of cryptographic primitives and secure routing protocols for achieving confidentiality and integrity while considering the constraints of UASNs environment. The authors proposed a protocol known as FLOOD. The FLOOD protocol was not secure and for protecting the confidentiality and integrity of the FLOOD protocol the authors proposed a secure flood (SeFLOOD) protocol. The experimental results revealed that the protocols suite proposed in this research is suitable for UASNs environment and it has affordable power consumption and communication overhead. The authors claim that so far their proposed protocol suite is the first practical, complete and efficient solution for providing confidentiality and integrity in the UASNs environment.
The authors in [30] proposed Tic-Tac-Toe AI-MINIMAX algorithm for implementing security in UWSNs. The focus of this research is to find the best and secure path for routing in UWSNs. The approaches of game theory are extensively utilized in this research in order to protect the sensor nodes in UWSNs from attacks and to protect the security of data in UWSNs communication. The implementation of Minmax algorithm includes two players such as min and max. While its implementation in UWSNs max is the sender and min is the attacker. The pre-requisite of Minmax algorithm is Tic-Tac-Toe. The Tic-Tac Toe considers all the available situations and chooses the optimal and secure move in UWSNs environment without any threat. The implementation of the proposed algorithm is done using the following procedures such as i) finding best move ii) current move better condition iii) GameOver state condition iv) Making our AI smarter. The authors predict that utilizing the AI models intelligent attacks can be mitigated up to some extent in the UWSNs environment.
The authors in [31] proposed a trust management model called a trust cloud model (TCM) for UWSNs. The purpose of the trust model is quantifying trust relationship among sensor nodes in UWSNs environment. After evaluating the quantified results the sensor nodes can determine whether the other node is trust worthy or not. Therefore, in the UWSNs environment only trust worthy sensor nodes will be selected for the transmission of data. The performance of the proposed TCM has been evaluated using the following three aspects: i) performance of detection of the malicious node: TCM performed well as compared to LCT and CBTM because both the LCT and CBTM did not take trust timelines into account. ii) performance of the calculation of trust value: It has been observed that in TCM the behavior of communication of normal nodes is good and their trust values rise with the passage of simulation time whereas in the presence of malicious nodes the trust values decreases due to the loss of packets by malicious nodes. iii) performance of the transmission of data: It has been observed that the rate of communication which is successful under the TCM is more as compared to the other two trust models.

III. MOTIVATION AND CONTRIBUTIONS
Research has been done on different issues in UWSNs such as the researchers in [32]- [38] conducted study on the recent applications, issues and challenges in UWSNs environment. Majority of the research done in UWSNs environment have energy efficiency as primary concern because the sensor nodes in UWSNs environment are operated on built-in battery having limited life time. The research conducted in [39]- [47] proposed different energy efficient routing protocol for UWSNs environment. Security is equally important to energy efficiency in the UWSNs environment. Compromising security in UWSNs cannot be tolerated in most of the cases. Therefore, this research work proposed a secure, energy efficient and cooperative routing (SEECR) protocol for UWSNs environment. The performance of SEECR is compared with AMCTD [48] a well-known routing protocol for UWSNs environment. This research implemented attack in both SEECR and AMCTD protocol in order to reveal the consequences of security attack in both the routing protocols. This research will help the research community to realize the impact of a security attack in UWSNs environment and must consider security mechanism while designing routing protocols for UWSNs environment. Secure solutions should have minimum computations considering the resource constrained environment of UWSNs.

IV. UNDERWATER WIRELESS CHANNEL
This section includes the attenuation, noise, signal to noise ratio (SNR) and path loss modeling in underwater wireless channel.

A. ATTENUATION IN UNDERWATER WIRELESS CHANNEL
The propagation speed of sound in water (c) is approximately 1500 m/s. The attenuation which takes place over the distance l, for a signal of narrow-band having carrier frequency f , is calculated in Eq.(1) [49].
where A 0 is a constant known as the normalizing constant, k is called the spreading factor, having value k = 1.5, and a(f ) is called the absorption coefficient which is modeled by using the Thorp's formula [50].
where a (f ) is calculated in dB/km and f is calculated in KHz. The value of total attenuation A (l, f ) can be calculated in Eq. (3).
10 log(A(l, f )) = k × 10 log(l) + l × 10 log(a(f )) (3) where k ×10 log(l) denotes spreading loss and l×10 log(a(f )) denotes absorption loss. where N t , N s , N w , and N th are the noise due to turbulence, shipping, wind, and thermal activities respectively. The whole spectral density of noise power is calculated in Eq. (4) [51].

C. SNR IN UNDERWATER WIRELESS CHANNEL
The signal to noise ratio (SNR) for underwater wireless channel is computed in Eq. (5) [51].

D. PATH LOSS IN UNDERWATER WIRELESS CHANNEL
Signals in underwater wireless channel practice frequency as well as link length-dependent path loss that is quite complex as compared to the radio channels and it is modeled in Eq. (6) [52].

V. SEECR: SECURE ENERGY EFFICIENT AND COOPERATIVE ROUTING PROTOCOL FOR UNDERWATER WIRELESS SENSOR NETWORKS
The proposed SEECR protocol utilizes multi-hop networking in UWSNs environment by utilizing the cooperation technique. In the proposed scheme, data packets which are generated from source node are forwarded to the destination node or sink node via hop by hop. Deployment of the relay node is done at the joint which contains two consecutive hops that accepts the arriving packets and after that amplifies them, and then retransmits these packets towards the destination. The proposed scheme detects common active routing attacks which drop packets and eliminates attacker nodes from the network efficiently.
A. NETWORK MODEL Fig. 2 shows a network model of UWSNs, where S is the source node, R1 and Rb are relay nodes, D is the destination/sink node and A1 and A2 are attacker nodes deployed in UWSNs. Rb is the best relay node selected among the available relay nodes such as R1 and Rb. The line dark in color reflected in Fig. 2 is the communication path which is direct and the dotted lines show the routes which are cooperative and these cooperative routes are used when the direct path is either unavailable or infeasible. The relay node will be checked whether it is attacker node or not, if it is attacker node then it will not be selected for the transmission of data.

B. CONFIGURATION AND INITIALIZATION
In this phase configuration and initialization is done. Each sensor node broadcast its depth as well as residual energy information to the neighbors using hello packets for the initialization of network operation. In this phase each sensor node knows about their neighbor sensor nodes. Sink node sends the hello packet to all sensor nodes. Each sensor node computes its weight using Eq. (7).
where W i is the weight of node i, T l reflects the path loss, R i represents the residual energy of node i, D w represents depth of water and D i shows the depth of node i.

C. MOVEMENT SCHEME OF COURIER NODES
For eliminating flooding process the value of depth threshold of the sensor nodes is marked to 60m. When the numbers of dead nodes is increased by 20% then for increasing the number of depth based threshold neighbors, the depth threshold is set to 40m. When the number of dead nodes in UWSNs environment increase to 75% then the depth of the courier nodes are adjusted for improving the network lifetime as follows: where N d is the number of dead nodes C 1d , C 2d , C 3d , C 4d are the depth of courier nodes 1, 2, 3 and 4 respectively, as also utilized in AMCTD.

D. ELIMINATING ATTACKER NODES
For detecting and eliminating attacker node each sensor node stores packet which are sent and received by neighbor sensor nodes. The packets are stored in Q j and Q k . The incoming packets P in of sensor node are stored in Q j and the outgoing packets P out of sensor node are stored in Q k . After the packets are stored, both the values Q j and Q k will be compared. If both the values are not equal and the sensor node is not a sink node then there are chances that the sensor node is an attacker node. The value of A i (attack indicator) will be incremented by 1. If the value of A i reaches x for a sensor node then that node will be detected as an attacker node and it will be eliminated so that it cannot contribute in the routing process. The value of x is set to 3. The value of x can be adjusted according to the environment. The value of x should be set carefully. If the value of x is very high the attacker node will be there in network for more time, if the value of x is very low then there is possibility of removing genuine sensor node as attacker node.
The process of attacker node detection and elimination has to be performed by all sensor nodes in order to ensure that the attacker node cannot participate in any activity of the network. Since the sensor nodes are positioned underwater in UWSNs and physical access to the sensor nodes is difficult and time consuming due to which after the detection of attacker node(s) it can still physically exist in the premises of UWSNs but the attacker node(s) will not be able to participate in any operation of the network.

Q j = P in
Store incoming packets P in of a sensor node in Q j Q k = P out Store outgoing packets P out of a sensor node in Q k where S k represents sink and A i is the attack indicator.  Where S represents source, Rb represents best relay node, and D represents destination.
Y sd = H s * g sd + n sd (8) where Y sd as shown in Fig.3 is the signal that is directly conveyed from the source to the destination node, H s reflects channel among the source and destination node or sink node, g sd reflects the information which is broadcasted from the source to the destination node, and n sd reflects the ambient noise which is added to the channel H s .
Y sr = H s * g sr + n sr (9) Y sr shows the signal transmitted among the source node and the relay node, g sr is the information which is conveyed from the source towards the relay node, n sr reflects the ambient noise which is added to the channel H s , Y rd = Y sr * g rd + n rd (10) During network operation when the destination node do not receive data packets from the source node then the relay node starts transmission and send data packets. Y rd represents the data which is transmitted from the relay node towards the destination node, g rd represents the information which is forwarded by the relay node towards the destination node, and n rd reflects the noise which is added on the channel H s to the information Y sr .

F. RELAY STRATEGY AND ROUTING PHASE
A source sensor node S has n encompassing sensor nodes in its vicinity and it depends on the condition of the channel in order to figure out the neighbors most suitable for transferring its data towards the sink node. The source node selects the relay node from its neighborhood by equating their weights. The sensor node having the maximum value of W i is selected for the transmission of data. If the residual energy of the source is more than or equal to the residual energy of the relay node, then direction transmission will be done otherwise the transmission will be through relay node.
where R s is residual energy of source node and R r is residual energy of relay node. The technique known as amplify and forward is considered at the relay node which uses an amplification factor on the signal received from the source and prior to forwarding the signal towards the destination node.

G. COMBINING STRATEGY
The destination sensor node D utilizes signal to noise ratio combining (SNRC) technique as combining technique in order to combine the signals which are arriving from the source S and relay R. The SNRC is where the signals combined at the receiver have a weight equal to the SNR seen at each array element. SNRC definitely works better than equal ratio combining (ERC) because it considers the small scales fluctuations and weights those signals less (low SNR) while combining. SNRC can be calculated as: (11) where Y d shows the output signal which is combined at the destination node D, X 1 represents weight of direct path and X 2 represents weight of relay path.

H. WEIGHT UPDATING PHASE
When the number of dead nodes is less than or equal to 20% then weight of sensor nodes will be same as the weight calculated in the initialization phase in Eq. (7). When the number of dead nodes in UWSNs environment is more than 20% and less than 75% then statements following if condition will be used to calculate the weight. If the numbers of dead nodes are more than or equal to 75% of the sensor nodes then statement following else condition will execute and the weight will be computed according to the statement following else condition as follows: If N d > 20 and N d < 75 The complete working of SEECR is reflected in a flowchart as shown in Fig. 4.

VI. SIMULATION ENVIRONMENT AND PERFORMANCE EVALUATION PARAMETERS
SEECR and AMCTD protocols are evaluated in two different scenarios such as with and without attack. The attack scenario contains eight attacker nodes whereas there is no attacker node deployed in without attack scenario. The numbers of sensor nodes deployed are 225 whereas the numbers of sink nodes deployed at the surface of the water is set to 10. The total number of rounds in simulation is set to 9000 as shown in Table 1.

A. PERFORMANCE EVALUATION OF SEECR
For evaluating the performance of proposed SEECR, it is compared with AMCTD using different performance evaluation parameters. The performance of SEECR and AMCTD are evaluated in different scenarios such as with and without attack.

B. PERFORMANCE EVALUATION PARAMETERS
The following parameters are used in order to evaluate the performance of SEECR and AMCTD protocols in the scenarios such as with and without attack.

1) NUMBER OF ALIVE NODES
The number of alive nodes is calculated by subtracting the number of dead nodes from the total number of nodes during the entire simulation.

2) TRANSMISSION LOSS
It is the transmission loss between the source and sink node during a single round. The transmission loss is calculated in decibels (dB).

3) THROUGHPUT
It is the entire number of packets which reach the sink node.

4) ENERGY TAX
It is the energy consumed while forwarding the data from the source node towards the sink node. It is calculated in joules.

5) END-TO-END DELAY
It is the time between packet generation and packet reach the sink. It is calculated in milliseconds.

VII. RESULTS AND DISCUSSION
The results obtained in this research from different performance evaluation parameters are discussed in this section of the research paper. The results obtained are presented in the form of both figures and tables. Fig. 5 shows the number of alive sensor nodes during the entire simulation. Initially the number of sensor nodes deployed is set to 225 which includes all the source and relay nodes. At the end of simulation the number of alive nodes in SEECR with and without attack is 111 and 112 respectively, the number of alive nodes in AMCTD with and without attack is 68 and 82 respectively. The results obtained show that attack significantly affects the energy consumption due to which the numbers of alive nodes is significantly less in AMCTD with attack scenario. Fig. 5 further revealed that stability is better in SEECR as compared to AMCTD in both the scenarios such as with and without attack. SEECR improves the overall stability of the network whereas attack causes more instability in AMCTD protocol with attack scenario. Moreover, due to the embedded security defense mechanism in SEECR there is almost negligible impact of attack on SEECR as we can see in the scenario of SEECR with attack. Table 2 shows the number of alive nodes after every 1000 rounds in different scenarios. It is clear from the results  obtained in Table 2 that the performance of SEECR is better than AMCTD in terms of number of alive nodes due to the energy efficient approach adopted in SEECR protocol. The percentage of alive nodes is calculated for the entire simulation. Moreover, it can be seen in Table 2 that the overall percentage of alive nodes in SEECR with and without attack is 60.9% and 63.2% respectively, which shows very little impact of attack and the little impact is due to the computation done for the detection as well as elimination of attacker nodes from the network. On the other side it is clear from Table 2 that the overall percentage of alive nodes in AMCTD with and without attack is 51.7% and 56.6% respectively. There is a significant degradation in the performance of AMCTD routing protocol in the presence of attack. The attack consistently degrades the performance of AMCTD protocol due to no defense mechanism. Moreover, SEECR protocol beats AMCTD protocol in both the scenarios such as with and without attack. Fig. 6 shows the transmission loss of SEECR and AMCTD protocols in different scenarios such as with and without attack. The results obtained revealed that the transmission loss of SEECR with and without attack is significantly less as compared to the transmission loss of AMCTD with and without attack. The energy efficient approach utilized VOLUME 8, 2020  by SEECR protocol significantly reduced the transmission loss in SEECR protocol. Moreover, SEECR protocol beats AMCTD routing protocol in terms of transmission loss. Table 3 shows the transmission loss of SEECR and AMCTD protocols after every 1000 rounds in different scenarios. The results mentioned in Table 3 indicates that the transmission loss in SEECR protocol with and without attack is significantly less as compared to the transmission loss of AMCTD protocol with and without attack. Moreover, the overall percentage of transmission loss in SEECR protocol with and without attack is 43.7% and 43.1% respectively, whereas the overall percentage of transmission loss in AMCTD with and without attack is 100% and 98.8% respectively. It is clear from Table 3 that as compared to AMCTD there is more than 50% reduction of transmission loss in SEECR protocol in both the scenarios such as with and without attack. Fig. 7 shows the throughput of SEECR and AMCTD protocols in different scenarios such as with and without attack. The results obtained show that the throughput of SEECR protocol is more as compared to the throughput of AMCTD protocol in both the cases such as with and without attack. The better throughput in SEECR protocol than AMCTD protocol is due to the energy efficient and secure mechanism utilized in K. Saeed et al.: SEECR Protocol for UWSNs  SEECR protocol. It is obvious from Fig. 7 that SEECR beats AMCTD in terms of throughput. Table 4 shows throughput of SEECR and AMCTD protocols after every 1000 rounds in different scenarios. The results mentioned in Table 4 revealed that the throughput of SEECR protocol is much better as compared to the throughput of AMCTD protocol. Moreover, it has been revealed in the results obtained that attack significantly degrades the performance of AMCTD protocol whereas the effect of attack is negligible on SEECR protocol due to the strong defense mechanism embedded in SEECR protocol. The overall throughput percentage of SEECR protocol with and without attack is 42.4% and 42.6% respectively, whereas the throughput percentage of AMCTD protocol with and without attack is 32.9% and 37.6% respectively, which shows significant degradation in the performance in AMCTD protocol in the presence of attack. Fig. 8 shows the energy tax of SEECR and AMCTD protocols in different scenarios such as with and without attack. The results obtained show that the energy tax of SEECR protocol is much less as compared to the energy tax of AMCTD protocol in both the scenarios such as with and without attack. Moreover, it is obvious from Fig. 8 that attack significantly  degrades the performance of AMCTD protocol in terms of energy tax whereas the effect of energy tax on SEECR protocol is negligible. Table 5 shows energy tax of SEECR and AMCTD protocols after every 1000 rounds in different scenarios. The results obtained in Table 5 revealed that energy tax of SEECR protocol is much less as compared to the energy tax of AMCTD protocol in both the cases such as with and without attack. Moreover, it is also clear from Table 5 that in attack scenario the energy tax of AMCTD protocol increases. The overall percentage of energy tax of SEECR protocol with and without attack is 77.3% and 76.1% respectively, whereas the percentage of energy tax of AMCTD protocol with and without attack is 100% and 88.3% respectively. It can be concluded from Table 5 that the energy tax in SEECR protocol is up to 23% less as compared to AMCTD protocol. Fig. 9 shows end-to-end delay of SEECR and AMCTD protocol in different scenarios such as with and without attack. Results reflected in Fig. 9 shows that there is major difference in the end-to-end delay of SEECR and AMCTD protocols. The end-to-end delay is less in SEECR protocol as compared to the end-to-end delay of AMCTD protocol. Table 6 shows end-to-end delay of SEECR and AMCTD protocols after every 1000 rounds in different scenarios.   The result obtained in Table 6 indicates that the overall percentage of end-to-end delay in SEECR protocol with and without attack is 70.2%, whereas the percentage of end-toend delay in AMCTD with and without attack is 95.6% and 100% respectively. It can be concluded from Table 6 that the end-to-end delay is 25% less in SEECR protocol as compared to that in AMCTD protocol.

VIII. CONCLUSION AND FUTURE WORKS
Routing attacks in UWSNs environment is an important factor which needs to be addressed. This research work proposed SEECR protocol for UWSNs environment. The proposed routing protocol efficiently utilizes the energy consumption and has built-in defense mechanism against common active attacks in UWSNs environment. The performance of SEECR protocol has been compared with AMCTD protocol in terms of different performance evaluation parameters. The results obtained revealed that SEECR protocol beats AMCTD protocol in terms of all performance evaluation parameters. Moreover, due to the built-in defense mechanism in SEECR protocol there is negligible impact of attack on SEECR protocol in UWSNs environment.
This research work is focused on the importance of energy efficient security mechanism in routing protocol for UWSNs environment. The results produced in this research revealed that despite of computation for attacker node detection and elimination the proposed solution is still suitable for UWSNs environment. The proposed solution in this research will encourage the research community to design secure solutions for UWSNs as well as other environments. Some possible research directions in this area can be designing other secure routing solutions and using artificial intelligence models for mitigating attacks in UWSNs environment. In the future we have a plan to introduce other energy efficient secure solutions for different environments.

IX. CONFLICTS OF INTEREST
The authors declare that there is no conflict of interest regarding the publication of this paper.
KHALID SAEED received the M.S. degree (Hons.) in computer engineering from the University of Engineering and Technology at Taxila, Pakistan, in 2011. He is currently pursuing the Ph.D. degree in computer science with the University of Engineering and Technology at Peshawar, Peshawar, Pakistan. His research interests include security in underwater wireless sensor networks, wireless sensor networks, cloud computing, mobile ad-hoc networks, delay tolerant networks, and information and communication technologies. Till date he has more than 20 publications to his credit in IEEE conferences and reputed journals.
WAJEEHA KHALIL received the Ph.D. degree from the University of Vienna. She is currently an Assistant Professor with the Department of Computer Sciences and Information Technology, University of Engineering and Technology at Peshawar, Peshawar. Her research interests include distributed computing, human-computer interaction, and parallel computing.
SHEERAZ AHMED received the Ph.D. degree in electrical engineering from COMSATS Islamabad, Pakistan, in the domain of underwater and body area sensor networks. He is currently a Professor with Iqra National University, Peshawar, Pakistan. His research interests include UWSNs, WBANs, smart grids, VANETs, FANETs, and so on. Till date he has more than 160 publications to his credit in IF journals and IEEE conferences. He has an experience of more than 20 years in teaching, research, and administrative positions.
IFTIKHAR AHMAD received the master's degree in computer science from the University of Freiburg, Germany, and the Ph.D. degree in theoretical computer science from Saarland University, Saarbrücken, Germany. He is currently an Assistant Professor with the Department of Computer Science and Information Technology, University of Engineering and Technology at Peshawar, Pakistan. He is also leading the National Center of Big Data and Cloud Computing, Machine Learning Group, University of Engineering and Technology at Peshawar. His research interests include theoretical computer science, machine learning, graph theory, and blockchain and cryptocurrencies.
MUHAMMAD NAEEM KHAN KHATTAK is currently working as the Chairman at the Department of Mechanical Engineering, University of Engineering and Technology at Peshawar, Peshawar. He has his specialization in technology and quality management. In particular, his areas of interest include total quality management and technology and innovation management. He has produced several publications internationally in the fields of his interest. VOLUME 8, 2020