Mitigating Intruder Detection System in Mobile Adhoc Network (MANET) using optimizer based ANN model

Mitigating Intruder Detection System in Mobile Adhoc Network (MANET) using optimizer based ANN model | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Mitigating Intruder Detection System in Mobile Adhoc Network (MANET) using optimizer based ANN model Sivanesan N, A. Rajesh This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3199495/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Mobile Adhoc Network (MANET) is an adoptable network with dynamic as well as decentralized network in nature. This research concentrates to resolve Denial-of-Service (DoS) attacks in MANET and illustrates the usual classification models, which may be unable to distinguish between legitimate DoS attacks as well as network problems. The routing path has been recognized as well as strengthen the environmental adoption with respect to several logic and statistical performances using Machine Learning (ML). The main aim in ML for recognizing the complexity pattern with AODV routing in MANET as well as decision making in accordance with accomplished result. In securing the MANET, there are several ML algorithms have been implemented. The lack of infrastructure in MANETs makes it extremely difficult to implement security measures. Moreover, the proposed sequential pattern with Artificial Neural Network (ANN) for AODV routing has generated better security from DoS attack. Thus, the security methods in MANETs majorly concentrating in mitigating intruder detection, eliminating malicious node as well as securing routing paths. MANET AODV Machine Learning (ML) Intruder detection sequential pattern. Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 1. Introduction Generally, the wireless networks have been classified with respect to infrastructure whereas the default with central access points as well as ad hoc consists of no access point. MANET doesn't have constant infrastructure but dynamic network in nature which can be implemented as multi-hop packet networks due to mobility in nature [ 1 ]. MANET is infrastructure less and each node may perform as source or destination node and even bridge in forwarding data packets in the node which is out of transmission range [ 2 , 3 ]. However, these devices or nodes are available in various transmission range, speed, packet size and data rates but certain available unique MANET characteristics like multihop, autonomous, dynamic topology, etc. [ 4 ]. There are certain restriction considered to the parameter in the network are packet loss, transmission range, security, etc. whereas the basic need in establishing an usual communication between the nodes. In this recent years, mobile communication growth get rapid because of ubiquity computation in MANET. The terminals of mobile set is located closer to communicating point that assist each other communication, sharing of resources or services and the generate limited period of computing time as well as within a defined space limit have generated a spontaneous adhoc network. Moreover, the network management is transparent to the user and the network type are independent in centralized administration. Hence, the user has the capability in entering and existing the network easily and the significant research area in MANET can be establish and maintain the ad hoc network usage of routing protocols [ 5 ]. The process of traffic path selection over a single or multi-network for sending and receiving data. This straightaway passes the logical addressed packet from their source towards overall destination via intermediate nodes. Based on the definite rules and recommendations, the data packets are routed is said to be routing protocol. Each routing protocol obtain its individual algorithm with respect to identification and maintenance of route. However, each routing protocol consists of data structure that stored the data route as well as alter the table in accordance with need of route maintenance. Hence, the routing metric is calculated through values that utilized by routing algorithm for determining the performance of routing of each node is compared with one another. Measurements may include data on bandwidth, delay, hop count, path cost, load, reliability, and communication costs. In contrast to link-state or topological databases, which can maintain all data, the routing table simply stores the optimum routes. The most significant development in the telecommunications industry right now is MANETs [ 6 ]. The Intruder Detection System (IDS) can detects intruders who attempt to steal data from the system. The main objective is in detecting communication attacks and send an alert to the administrator of network which perform as a significant system in detecting malicious data in the training mode [ 7 ]. In general, the intrusion has been detected through traditional methods namely firewall, authentication encryption as well as decryption, etc. that can be classified into three categories are Anomaly Detection System (ADS) Misuse Detection System (MDS) Hybrid Detection System (HDS) In ADS, the intrusions are checked at any sample deviation from the baseline is identified as the malicious node. In the case of MDS, the node used is executed through sample matching that have sorted in the database as well as make decisions. The HDS is combination of both ADS and MDS properties that minimize the flaws of detecting system which is robust detecting system than ADS and MDS to provide suitable decision making [ 8 ]. In MANET, the IDS may perform individually from its wired counterparts. If the advancement is done for IDS in MANET may solve the challenges needed. All unique activities have been tracked and recorded through node level agents are implemented in the non-collaborative intruder monitoring system [ 9 ]. The agents position is determined, if the nodes are mobile and challenge lies are highly significant. Nowadays, the ML is most widely used technology that enables computer systems in learning dynamically with no human intervention and also provides suitable action. It creates a model by manipulating complex data effortlessly, efficiently, and precisely. The generalized structure may assist ML in providing suitable patterns for improving device performance. There are numerous scientific applications, including manual data entry, medical diagnostics, automatic spam detection, data clearing, image recognition, noise reduction, etc. [ 10 ],[ 11 ]. According to the most recent research, ML is utilized in WSNs for addressing variety of issues. The usage of ML in WSNs not solely improves system efficiency and even avoids challenging issues like reprogramming, manually finding massive amounts of data, and collecting valuable data from dataset. ML methods are frequently useful in acquiring large amounts of data as well as producing valuable information [ 12 ]. Denial of service (DoS) attacks, aimed at preventing legitimate handlers from retrieving or using several network resources, have been common in network examination analysis. A resilience scheme contrary to these attacks has been proposed for AODV, and the efficiency of this scheme is demonstrated by Light Gradient Boosting Machine (LGBM), Gradient Boosting, Hist Gradient Boosting, Radom Forests, Decision Trees, Extra Tree Classifiers, Support Vector Classifiers, Bagging, K Neighbors, Extreme Gradient Boosting (XGB). Decision Tree (DT) Support Vector Machine (SVM), Naive Bayes (NB), etc. have been used for intruder detection [ 13 ]. This paper has structured with section II discusses the literature review in improving the accuracy in detecting intruder and classifying IDS prediction for AODV routing protocol in MANET using ML techniques. Section III discusses various sequential pattern mining in ANN model with various optimizer for identifying the high prediction in IDS classifications over MANET. Section IV deals with evaluation of various ANN model with different sequential pattern, activations and optimizer is determined with accuracy, sensitivity and specificity performance. Session V has concluded that ANN-Model 3 has better accuracy in predicting IDS classification in MANET while compared to other ANN model sequence pattern combinations. 2. Literature Review DoS attack has become a well-known attack that attracts the interest of several researchers and focuses on providing security. However, the solutions from the proposed researcher have only limited success against DoS attack impact in MANET. The literature review discusses IDSs are based on various detection techniques and ANN model approach in detecting malicious node in AODV routing protocol over MANET. Yuan et al., [ 14 ] compared MANET-mediated detection methods. Various attacks such as Black hole, gray hole, and war hole attack detection approaches are described. The weaknesses and strengths of these approaches were demonstrated. In this paper, we consider MANET using the AODV protocol. They analyzed the underlying AODV protocol used for route direction. The result is a Power overhead value of 0.74%. Panos et al., [ 15 ] proposed an AODV protocol monitoring DOS attacks. MANETs are improvements to the AODV protocol. Using the trust value for each node, they were able to identify malicious nodes. Here the author has worked on neural networks. They used Transmission Control Protocol (TCP) for that MANET rating. S. Sargunavathi et al. have illustrated an efficient trust based IDS by local data to data transmission for high dynamic MANET. In this research, blackhole as well as greyhole attacks have been addressed to select trusted nodes for ensuring security. Thus, the proposed method gets compared through conventional AODV protocol with several performance metrics [ 16 ]. Sultan Mohamad, Sayed Hesham, and Khan Manzoor are majorly focuses in designing and investigating using the ANN model for detecting intrusion in MANETs. The major goal of this study is for analyzing, simulating and evaluating the feedforward neural network benefits with back propagation over MANETs. The dataset extraction is generated through simulations manner for MANET has been utilized for evaluating the proposed method input parameters. Moreover, the RMSR is introduced as a metric for evaluating the proposed ANN model performance. Hence, the proposed ANN model is used in detecting DoS attacks in MANET. Thus, the accomplished result of the proposed model is associated with 4-15-10-1 network which generates an outcome of RMSE as 0.0452 is executed for 14 epochs with the train dataset and RMSE as 0.0512 is executed for 14 epochs with the test dataset [ 17 ]. Pravin kumar et al. has addresses the issue of packet loss in MANET by congestion control mechanism with NN to generate stable data transfer in MANET [ 18 ]. M.P. Arthur has introduced a cross layer-based IDS to assist in generating exchange routing data among various layers of two nodes from their neighbors. Additionally, the adoption of the multiclass SVM method is introduced for reducing the overhead as well as improving accuracy in classifying the attacks in MANET [ 19 ]. Basomingera et al. has explained the novel route cache mechanism with SVM method to progress IDS in non-clustering MANET. Thus, the simulation outcome illustrates that the rate of IDS has increased to 95% in MANET [ 20 ]. S. Pandey and V. Singh have recommended a secured AODV (S-AODV) routing techniques using ANN and SVM methods. The performance of the proposed S-AODV is evaluated through simulation results to determine that it obtains better performance than traditional ML with a routing protocol model under various traffic cases under DoS attack [ 21 ]. J. Kim et al. has introduced Long Short Term Memory - Recurrent Neural Network (LSTM-RNN) for accomplishing better classifier in IDS with 96.93% accuracy and 10% False Positive Rate (FPR) in the KDD dataset [ 22 ]. Yujie Fan et al. has illustrated the three-step process to create automatic malware detection using malicious sequential pattern mining. The three processes are Intrusion sequence extractor Miner of malicious sequential pattern ANN with kNN classifier for prediction In order to detect and predict the various threats and attacks through ANN and kNN classifiers whereas other such signature-based detection, as well as feature-based detection, is involved in this research for determining the better prediction of detecting intrusion [ 23 ]. 3. Research Methodology This research purpose in proposing sequential ANN model using hidden dense layers modification are used in determining the intruder present in the MANET through various optimizer selections. The major goal of optimizer is initiating it’s parameter in the training data and associate with batch size and activation functions. Once the optimal modification is done through sequential pattern mining that can able to manage the iteration of epoch training to improve the accuracy of training model. The various attributes provided with dataset input from the sender user to receiver user in the MANET is shown in Fig. 1 . The MANET dataset get collected as well as verified with missing data is performed using is.null() function as missing imputation process and the data preprocessing is executed. In data preprocessing, normalization is done to define the target “attack_type” in term of ordinal as 0 to 4 shown in Table 1 is determined through labelEncoder(). This function is used for converting data type of the object in term of integer or float data type by considering a continuous variable as well as ID number variable has been dropped. Table 1 Assigning value for attack type Sl. No Attack Type Label Encoder Assigned 1 Grayhole 0 2 Normal 1 3 Blackhole 2 4 TDMA 3 5 Flooding 4 Figure 2 illustrates the correlation heat map in which the coloration among every attributes in the MANET dataset have been analyzed to understand the significance of the attributes present. The relationship of data_s and sch_r is 0.74 is highly correlated followed by send_code and sch_r is 0.70 as well as dis_to_ch and sch_r is 0.69. The standardScaler() function is used for scaling all attributes into single scale. In this research, there are 16 input variable get preprocessed by missing imputation, normalization and scaling all variable unit using standard scalar. The dataset get split into 80% as train dataset and 20% as test dataset. The ANN-Model 1 has trained using ANN algorithm with sequential pattern as well as setting the layer shapre of input layer as 32 as unit whereas the output dense layer as 5 and the activation for input as Rectified Linear unit (Relu) and output as sigmoid. The ANN-Model 2 is with sequence pattern and set the input layer shape as 32 as unit, the output dense layer as 5 and the activation for input as Leaky.Relu with alpha as 0.2 and output as sigmoid. The ANN-Model 3 with sequence pattern and set the input layer shape as 32 as unit and input layer as 16, the output dense layer as 5 and the activation for input as Relu and output as sigmoid. The ANN-Model 4 with sequence pattern and set the input layer shape as 32 as unit and input layer as 16, subsequent layer as 8, the output dense layer as 5 and the activation for input as Relu and output as sigmoid. Moreover, this model has used “relu” as an activation as well as fit with batch size as 5 and 10 epochs have been deliberated with adam optimizer. In addition, ANN-Model 5 has trained by ANN technique as well as setting layer shape of input as 32 whereas the hidden layer shape 16, output layer shape as 5 and optimizer as Stocastic Gradient Descent (SGD). In the case of ANN-Model 6, input layer shape is set input layer shape as 32 in which the hidden layer shape 16, output layer shape as 5 and optimizer as RMSprop in Fig. 3 . The best hidden layer has obtained from adequate sequence pattern and better optimization process is considered to accomplish high accuracy. 4. Working of ANN model building The ANN model has acquire from sequence data patterns and tried to accomplish as probable as an accurate prediction. Assume, the set of X data pairs containing the variables and the results, (m 1 , t 1 ), (m 2 , t 2 ), ….(m X , t X ) in which the m i is the input value and t i is the target value for i = 1, 2, 3, …, X. We would like to build a neural net F so that ideally as $$F\left({m}^{i}\right)= {t}^{i}$$ 3.1 Moreover, it allow for error \({\in }^{i}\) typically. Let n denotes the output of ANN is expressed in Eq. 3.2 and 3.3 . $${n}^{i}=F\left({m}^{i}\right)$$ 3.2 And $${t}^{i}={n}^{i}+{\in }^{i}$$ 3.3 However, the \({n}^{i}\) is based on the parameter with respect to weight and bias which turn as an optimizer issue. In this, setting the ANN in F that minimize the error function represent in Eq. 3.4 . $$E= \frac{1}{N}\sum _{i=1}^{X}{‖{t}^{i}-{n}^{i}‖}^{2}$$ 3.4 Where, N = Number of training patterns. When the ANN has become a two-way classification then N = 2. Based on the equation, E is a parameter function of F, and required in determining the weight values that minimize the error through differentiating E. When the research is focused on one term of the sum and it is expressed in Eq. 3.5 . $${‖t-n‖}^{2}= {\left({t}_{1}-{n}_{1}\right)}^{2}+{\left({t}_{2}-{n}_{2}\right)}^{2}+\dots +{\left({t}_{x}-{n}_{x}\right)}^{2}$$ 3.5 Thus, the values of input and output have been fixed, and only the parameter is consider to be calculated through weight and it can differentiate from both sides is expressed in Eq. 3.6 . $$\frac{\partial }{{\partial W}_{ij}}\left({‖t-n‖}^{2}\right)= -2\left(t-n\right).\frac{\partial n}{\partial W}$$ 3.6 There are highly specific and also verify the fits over the context of neural net. From a neural net, the output is defined as \({n}^{i}= {W}_{ij}{m}^{i}+b\) . Hence, the output is rely on weight and while differentiating both sides in accordance with \({W}_{ij}\) by chain rule as per Eq. 3.7 $$\frac{\partial }{{\partial \text{W}}_{\text{i}\text{j}}}\left({‖\text{t}-\text{n}‖}^{2}\right)= -2\left({\text{t}}_{\text{i}}-{\text{n}}_{\text{i}}\right){\text{m}}_{\text{i}}$$ 3.7 Where, \({\text{m}}_{\text{j}}\) is the i th coordinate position. The derivative has provided direction to the maximum for accomplishing the minimum point, and subsequently in opposite direction of this gradient. In addition, this derivative as close to 0 as possible for obtaining the minimum of error. The ANN method is a layer based network that involves one or more artificial neurons which generally includes input layer, hidden layer as well as an output layer. Figure 4 has illustrates the ANN structure that has capability in identifying, mapping with robust capability and even have capacity in processing data in parallel. a. Input Layer This layer is the top most layer of ANN in which collection of input data is in the form of text, image files, audio files and video files that have received initially. b. Hidden Layer This layer is the subsequent layer in ANN model that has potential in both perceptron or several hidden layers. Based on the input data, these hidden layers have been executed various mathematical operations as well as recognize the pattern of the ANN model. c. Output Layer Based on the hidden layer input generation from feature weights of middle layers with accurate and exact computations have capability in producing an adequate result in the output layer. There are several positive or negative weights have been related to every neurons for execute or prohibit an input with every connection to the artificial neurons. In order to control the performance of the artificial neurons, activation function plays a major role. This artificial neurons assist in collecting input signal through computation of net input signal as a function with its associative weights. The performed input from net input signals to activation functions have calculated the artificial neuron as an output signal which operate various stages in mathematical process whereas the unit numbers are arranged in layer numbers. The single unit is said to be neuron in which the input unit arranged from input layer consumes various inputs from the acquired data. After, the input data is further progressed to hidden layer unit that transform input data into output unit. Relu is a function and major benefit of Relu activation function and doesn’t activate all neuron simultaneously in which a neuron with negative value have converted into zero or it gets deactivated. Networks become sparse as well as computationally effective as a result. The gradient value of the graphs at the negative side is zero. It suggests that neurons terminate are never stimulated during back propagation. The sigmoid function is utilized to maintain issues with multi-class which maps the value of output from 0 to 1. It works best when utilised in the classifier's output layer. There isn't a guideline that may be used to select the activation function. The characteristics of the issue might assist in choosing a quicker converging network. Certain characteristics are based on the research as the Relu and sigmoid for activation functions. A procedure called optimization aims to decrease network error. This is essential for increasing the model's accuracy. The optimizer has three different iterations: Adam, SGD, and RMSProp. SGD is iterative gradient descent technique that uses iteration to search for an optimal error. These models produce predictions in every iteration that follows and predicted results are compared to the predictions. Error is defined as the difference between the projected value and the actual result. The internal parameters of the model as well as the weights of the network are updated using this error. The back propagation algorithm follows this updating process. SGD also finds it challenging to get away from the saddle points. The most popular choices for handling saddle points are AdaGrad, RMSprop, ADAM, and AdaDelta. To modify the gradient with the slop and quicken the SGD, Nesterov accelerating gradient is utilized. Due to its ability to execute more updates for infrequently parameters as well as fewer changes for frequent parameters, AdaGrad outperforms Nesterov accelerated gradient. As a result, SGD becomes faster, more scalable, and more resilient. In order to train ANN, it is utilized and the fundamental disadvantage in AdaGrad optimizer has minimized the model's ability for training through the sequential pattern better. Two optimizers like RMSProp as well as AdaDelta, have been created individually to address the problem with AdaGrad. Moreover, the optimizer like AdaDelta and RMSProp are identical whereas the main difference is AdaDelta which is not fixed as an early learning rate with constant. Adam is an optimizer which incorporate the beneficial features of RMSprop as well as Adadelta. Adam is deemed a good choice since it get improves RMSProp as well as Adadelta. As a result, the optimizers used in this paper are Adam, RMSProp, and SGD. 5. Algorithm for Sequence pattern based ANN Sequence based ANN is an efficient algorithm for enumerating the set of layer sequence based on features weight considered through activation function. There are various sequence pattern and optimizer function involved in this experimental algorithm are discussed below. Step 1: Let initialize the network weight randomly in the training set and f represent the forward pass. Step 2: Let x be input network, O be the neural network weight during forward pass and assigning the sequence pattern with various dense layer combinations with activation function as Relu as well as leaky Relu in feed forward pass. Step 3: Considering the training output as T using activation function as sigmoid and calculating error = T-O as an output unit with five different layer. Step 4: Compute weight for all the features from input layers to hidden layers as a backward pass and even computing hidden layer to output layer. Step 5: Update the weight of best sequence in the ANN and again selection of various optimizer like adam, SGD and rmsprop involved to obtain high classification accuracy. Step 6: Repeat step 5 until the optimal accuracy of epochs have obtained from an optimizer. Step 7: Return The representation of epoch’s number determines the training dataset iteration is transverse. Each epoch provides training sample which the capability to update parameters of the internal model. The epoch is initiated at zero and incremental till epochs count, as well as several batches. Hundreds or thousands of epochs are typically chosen, allowing the network to effectively minimize the inaccuracy. The programmer must look at the learning curves of error as well as accuracy in choosing the best epoch value. These curves are utilized in identifying the model that has learned excessively, inadequately, or is well prepared for training. The training samples quality sent to the networks is known as the batch size. Several batches can be created from a training dataset whereas the algorithm is commonly mentioned as batch gradient descent while any training is given as a single batch. The SGD is batch size is one whereas the learning algorithm is said to be mini-batch gradient descent while the batch size is greater than one but smaller than the training size. The accuracy of the network is improved by the larger batch size which has a power of 2 range of 1 to 1024 provides several key considerations for choosing small batch sizes. Larger batch sizes reduce the model quality because they generally lead to sharp minimizes of the training function. The sequential pattern ANN has effectively extract the multifaceted characteristics between the applicable variables of classifying the intruder in the AODV routing. The selection of optimizer is significant for improving the optimum selection of route path in AODV of the MANET. This can be determined through better prediction with model accuracy. Thus, the model efficiency improved by optimizing the best obtained sequence in increasing the predicting IDS accuracy by enhancing the ANN model accuracy. 6. Results and Discussion This research experiment is conducted using an i5-7400 processor running at 3.00 GHz and a GTX 1050 GPU. The computer has 8.00 GB RAM and x64-ased processor. Python programming is used to train the models in the Keras environment and Tensorflow is employed to assess the models. The parameters involve the number of hidden units, activation function, number of layers, optimization, epoch’s number, and batch size used for validating the models. A crucial methodological decision in any empirical evaluation of optimizers is choosing the sequential search space associated with every optimizer. However, this necessitates the frequently untrue assumption that variables with identical designations should have comparable values among optimizer. Figure 5 illustrates the plotting of ANN sequence with dense layer of 32 units as input, 16 layer as hidden dense layer with an input activation as Relu. The output dense layer with five different units using sigmoid as an activation function in which the classification prediction of security status have been determined. Hence, the determined layer is accomplished with high accuracy in this respective sequence pattern than other sequence pattern like 32 unit with 16 and 8 dense layer as hidden layer in the ANN model. Figure 6 illustrates the optimizer iteration with loss function as binary cross entropy in which optimal results is obtained in the adam optimizer shown better accuracy while compared to other optimizer. Thus, the trained optimizer results in classifying the security of AODV routing protocol in MANET can be evaluated through confusion matrix metrics. Table 2 illustrates the ANN model with various sequence pattern and different optimizer whereas the input layer as dense unit as 32, following subsequent layer as 16 with activation as ReLu. It is classified with three different optimizers like Adam, SGD and RMSPROP. Table 2 ANN model with various sequential and optimizer Sl.No ANN Model Name Optimizer with different sequential pattern 1 ANN Model – 1 ADAM with I/P – Dense : 32, O/P – Dense : 5 I/p Activation: ReLu 2 ANN Model – 2 ADAM with I/P – Dense : 32, O/P – Dense : 5, I/p Activation: LeakyReLu 3 ANN Model – 3 ADAM with I/P – Dense : 32, Next I/P – Dense : 16,O/P – Dense : 5 I/p Activation: ReLu 4 ANN Model – 4 ADAM with I/P – Dense : 32, Next I/P – Dense : 16, Next I/P – Dense : 8 O/P – Dense : 5, I/p Activation: ReLu 5 ANN Model – 5 SGD with I/P – Dense : 32, Next I/P – Dense : 16,,/P – Dense : 5 I/p Activation: ReLu 6 ANN Model – 6 RMSPROP with I/P – Dense : 32, Next I/P – Dense : 16, O/P – Dense : 5 I/p Activation: ReLu Table 3 ANN model with various sequential and optimizer Classification ANN model Confusion matrix metrics Micro Precision Micro Recall Micro F1-Score Weighted Precision Weighted Recall Weighted F1-Score ANN Model – 1 96.28 96.28 96.28 96.10 96.28 96.45 ANN Model – 2 90.33 90.33 90.33 93.25 90.33 91.37 ANN Model – 3 96.75 96.75 96.75 95.90 96.75 96.89 ANN Model – 4 94.96 94.96 94.96 94.99 94.96 95.24 Table 3 illustrates the evaluation of ANN model with several sequence pattern with Adam optimizer whereas the input layer as dense unit as 32, following subsequent layer as 16 with input activation as ReLu except ANN model 2 in which the input activation as leaky ReLu but the out activation function used is sigmoid. Figure 7 illustrates the micro and weighted value to four different ANN models with adam optimizer in several sequence pattern. The accuracy is said to be micro F1-Score for ANN-Model 3 is 96.75% which is comparatively higher than other sequential pattern model. The proposed sequential pattern of ANN-Model 4 has better prediction in classifying IDS from the AODV routing protocol in MANET. Figure 8 has illustrate the sensitivity and specificity of various ANN models in which ANN models has sensitivity as 0.973 and specificity as 0.987 which is higher than other ANN models. Hence, it determines high true positive rate is better in ANN-Model 4 illustrates that IDS classification prediction is high in ANN-Model 4 than other model. Figure 9 has illustrates the accuracy of ANN model with various optimizer in which ANN-Model 3 as ADAM, ANN-Model 5 as sgd and ANN-Model 6 as rmsprop. However, the accuracy of adam optimizer in ANN model as 96.75% is higher than other optimizer ANN model. Hence, adam optimizer with sequence pattern of ANN-Model 3 has better accuracy in predicting IDS classification through AODV routing protocol in MANET. 7. Conclusion Providing security necessitates in maintaining unauthorized users from accessing data or other items and guarding against unauthorized changes to or destruction of user data. According to the traditional definition, security also includes availability, confidentiality, and integrity. The desirable features of wireless sensor network are in paying attention to many researchers to act based on various issues of security requirements. To solve the mentioned problems, ANN technique with different sequential pattern as the classifier in identifying IDS as well as route discovery for network shield from DoS attack. The MANET is trained by lazy predict classifier library for identifying the malicious node that appears in the network. The route has modified through elimination of malicious nodes from route as well as network is secured. Moreover, this proposed sequential ANN model with best optimizer has influences better understanding of data and resulted with high predictable accuracy as 96.75% which is higher than other ANN models. Thus, the evaluation of proposed ANN model produce high recognition in IDS as well as tracking shortest path in AODV routing protocol over MANET. Declarations Funding The authors have not received any funding for the work presented in the paper. Data Availability The data and source codes used to support the findings of this study can be obtained from the corresponding author upon request. Conflicts of Interest The authors declare that they have no competing interests. Authorship Contribution Statement Sivanesan N: Methodology, Formal Analysis, Writing of Original Draft, Validation. Rajesh A : Supervision, Reviews, Editing and Validation References T. Salam and M. S. Hossen, “Performance analysis on homogeneous LEACH and EAMMH protocols in wireless sensor network,” Wireless Personal Communications, vol. 113, no. 1, pp. 189–222, 2020. M. S. Hossen, “DTN routing protocols on two distinct geographical regions in an opportunistic network: an analysis,” Wireless Personal Communications, vol. 108, no. 2, pp. 839– 851, 2019. M. Singh, C. Kumar, and P. Nath, “Challenges and protocols for P2P applications in multi-hop wireless networks,” in in 2018 Second International Conference on Computing Methodologies and Communication (ICCMC), pp. 310–316, IEEE, 2018. T. Qiu, N. Chen, K. Li, D. Qiao, and Z. Fu, “Heterogeneous ad hoc networks: architectures, advances and challenges,” Ad Hoc Networks, vol. 55, pp. 143–152, 2017. M. Manjunath and D. H. Manjaiah, “Performance Comparative of AODV, AOMDV and DSDV Routing Protocols in MANET Using NS2 Alamsyah1,2,” Int. J. Commun. Netw. Syst., vol. 004, no. 001, pp. 18–22, 2018, doi: 10.20894/ijcnes.103.004.001.005. M. G. K. Alabdullah, B. M. Atiyah, K. S. Khalaf, and S. H. Yadgar, “Analysis and simulation of three MANET routing protocols: A research on AODV, DSR & DSDV characteristics and their performance evaluation,” Period. Eng. Nat. Sci., vol. 7, no. 3, pp. 1228–1238, 2019, doi: 10.21533/pen.v7i3.717. Feng, F., Liu, X., Yong, B., Zhou, R., Zhou, Q.: Anomaly detection in ad-hoc networks based on deep learning model: a plug and play device. Ad Hoc Netw. 84, 82–89 (2019) Liu, G., Yan, Z., Pedrycz, W.: Data collection for attack detection and security measurement in mobile ad hoc networks: a survey. J. Netw. Comput. Appl. 105, 105–122 (2018). Suma R, Premasudha BG, Ram VR. A novel machine learning-based attacker detection system to secure location aided routing in MANETs. International Journal of Networking and Virtual Organisations. 2020;22(1):17-41. Bi J, Yuan H, Zhou M. Temporal prediction of multiapplication consolidated workloads in distributed clouds. IEEE Transactions on Automation Science and Engineering. 2019 Feb 21;16(4):1763-73. Bi J, Yuan H, Zhang L, Zhang J. SGW-SCN: An integrated machine learning approach for workload forecasting in geo-distributed cloud data centers. Information Sciences. 2019 May 1;481:57-68. Wang J, Gao Y, Yin X, Li F, Kim HJ. An enhanced PEGASIS algorithm with mobile sink support for wireless sensor networks. Wireless Communications and Mobile Computing. 2018 Dec 2;2018. N. Khanna and M. Sachdeva, ''A comprehensive taxonomy of schemes to detect and mitigate blackhole attack and its variants in MANETs,'' Comput. Sci. Rev., vol. 32, pp. 24–44, May 2019. X. Yuan, C. Li, and X. Li, "DeepDefense: Identifying DDoS attack via deep learning", Proc. IEEE Int. Conf. Smart Comput. , pp. 1-8, May 2017. C. Panos, C. Ntantogian, S. Malliaros, and C. Xenakis, ‘‘Analyzing, quantifying, and detecting the blackhole attack in infrastructure-less networks,’’ Comput. Netw., vol. 113, pp. 94–110, Feb. 2017. Sargunavathi, S. and Martin Leo Manickam, J., 2019. Enhanced trust based encroachment discovery system for Mobile Ad-hoc networks. Cluster Computing, 22(2), pp.4837-4847. https://link.springer.com/article/10.1007/s10586-018-2405-7. Sultan, Mohamad & Sayed, Hesham & Khan, Manzoor., An Intrusion Detection Mechanism for MANETs Based on Deep Learning Artificial Neural Networks (ANNs), International Journal of Computer Networks & Communications (IJCNC) Vol.15, No.1, January 2023. Kumar, P., Tripathi, S. and Pal, P., 2018, March. Neural network based reliable transport layer protocol for MANET. In 2018 4th International Conference on Recent Advances in Information Technology (RAIT) (pp. 1-6). IEEE. Arthur, M.P., 2018, September. An SVM-based multiclass IDS for multicast routing attacks in mobile ad hoc networks. In 2018 International Conference on Advances in Computing, Communications and Informatics (ICACCI) (pp. 363-368). IEEE. Basomingera, R. and Choi, Y.J., 2019, January. Route cache based SVM classifier for intrusion detection of control packet attacks in mobile ad-hoc networks. In 2019 International Conference on Information Networking (ICOIN) (pp. 31-36). IEEE. Pandey, S. and Singh, V., 2020, July. Blackhole attack detection using machine learning approach on MANET. In 2020 International Conference on Electronics and Sustainable Communication Systems (ICESC) (pp. 797-802). IEEE. Thu H L T Kim J, Kim J et al. Long short term memory recurrent neural network classifier for intrusion detection. Platform Technology and Service (PlatCon), International Conference on. IEEE, 2016. Y. Fan, Y. Ye and L. Chen, "Malicious sequential pattern mining for automatic malware detection", Expert Systems with Applications, vol. 52, pp. 16-25, 2016. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-3199495","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":222238379,"identity":"1be81dd7-e13d-4b01-aef1-672305cdbfb2","order_by":0,"name":"Sivanesan N","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA+UlEQVRIiWNgGAWjYJCCDwwMEjC2DRAzNh4goINxBpKWNJBAAzFa4OAwmMSrxVz68MHmgj8Wif2zmw9/+LnjvN3a9sNAW2psonFpsexLS2ye2SaROOPOsTTJ3jO3k7edSQRqOZaW24BDi8EZHvPHvA0Sxgw3cswYeNtuJ5sdAGphbDiMT4thM88fCWP5G/mfP/5tO5dsdv4hMVrYJOQMbuQwSPO2HbAzu0HAFssetsRm3jYJOcMbaWbSsm3JCWY3gLYk4PGLOQ/zQaDD6njkbiQ//vi2zc7e7Hz6wwcfamxwOwxdIBGsMgGHcqxa7PEoHgWjYBSMghEKALFAYowzmAmlAAAAAElFTkSuQmCC","orcid":"","institution":"Vels Institute of Science, Technology \u0026 Advance Studies (VISTAS)","correspondingAuthor":true,"prefix":"","firstName":"Sivanesan","middleName":"","lastName":"N","suffix":""},{"id":222238381,"identity":"f78316cc-b884-461b-9215-4584ae88899b","order_by":1,"name":"A. Rajesh","email":"","orcid":"","institution":"Vels Institute of Science, Technology \u0026 Advance Studies (VISTAS)","correspondingAuthor":false,"prefix":"","firstName":"A.","middleName":"","lastName":"Rajesh","suffix":""}],"badges":[],"createdAt":"2023-07-24 13:00:10","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3199495/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3199495/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":40968831,"identity":"f0c26772-c6f8-4ca1-9ae0-3e03a37b1157","added_by":"auto","created_at":"2023-08-02 16:57:56","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":119631,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMANET dataset from an open source as input\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-3199495/v1/d5610e614270d035ad6e72fb.png"},{"id":40968581,"identity":"84847f5d-7e67-412f-bf52-abe9b0c5293d","added_by":"auto","created_at":"2023-08-02 16:49:56","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":406032,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCorrelation of MANET dataset attribute except target attribute\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-3199495/v1/3342e5249820b8189629529c.png"},{"id":40968964,"identity":"921eccf3-1a63-40f1-a212-20bf139b41cc","added_by":"auto","created_at":"2023-08-02 17:05:56","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":89520,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eProposed architecture for sequential pattern with optimized ANN model for IDS\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-3199495/v1/f291e7ed740d0519bb098172.png"},{"id":40968578,"identity":"2838adf4-ee05-400a-9270-a2b16d028e42","added_by":"auto","created_at":"2023-08-02 16:49:56","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":135364,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eArchitecture of ANN\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-3199495/v1/01b5f0c7fead413a44edfb4e.png"},{"id":40968586,"identity":"21591f7f-174a-412b-ba32-fbcfc1af014e","added_by":"auto","created_at":"2023-08-02 16:49:56","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":200938,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eANN with sequence dense layer of 32*16*5\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-3199495/v1/835bd48d77bac5b2011bea61.png"},{"id":40968584,"identity":"e7b78609-2fdb-4e4f-abec-2a36e8170d6e","added_by":"auto","created_at":"2023-08-02 16:49:56","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":229787,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eEpochs for adam optimizer with input dense 6 as sequential selection of ANN model\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-3199495/v1/153d5a2e94c51c215f9d6a54.png"},{"id":40968833,"identity":"ddac9de2-cbc4-4ab3-b6f5-ac6a23166e3d","added_by":"auto","created_at":"2023-08-02 16:57:56","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":41491,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMicro and weighted value for various ANN models\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-3199495/v1/8d032327f42dd7f4beef9a4d.png"},{"id":40968830,"identity":"8eaf5c02-7859-4a9b-abbd-30b10c78576f","added_by":"auto","created_at":"2023-08-02 16:57:56","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":32695,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSensitivity and specificity score for various ANN models\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-3199495/v1/db0e333e7bbf561a48700127.png"},{"id":40968582,"identity":"9fb2adad-d4e9-42f4-b1aa-eb50659fb463","added_by":"auto","created_at":"2023-08-02 16:49:56","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":19761,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAccuracy for different optimizer ANN models\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"9.png","url":"https://assets-eu.researchsquare.com/files/rs-3199495/v1/54e664b100d710dbc244036b.png"},{"id":41269769,"identity":"6b4f4420-4742-4aa5-b87a-49918aaffc28","added_by":"auto","created_at":"2023-08-08 21:37:25","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1625153,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3199495/v1/d0c59222-b80d-46e4-a130-63b1270307ba.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Mitigating Intruder Detection System in Mobile Adhoc Network (MANET) using optimizer based ANN model","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eGenerally, the wireless networks have been classified with respect to infrastructure whereas the default with central access points as well as ad hoc consists of no access point. MANET doesn't have constant infrastructure but dynamic network in nature which can be implemented as multi-hop packet networks due to mobility in nature [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. MANET is infrastructure less and each node may perform as source or destination node and even bridge in forwarding data packets in the node which is out of transmission range [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. However, these devices or nodes are available in various transmission range, speed, packet size and data rates but certain available unique MANET characteristics like multihop, autonomous, dynamic topology, etc. [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. There are certain restriction considered to the parameter in the network are packet loss, transmission range, security, etc. whereas the basic need in establishing an usual communication between the nodes. In this recent years, mobile communication growth get rapid because of ubiquity computation in MANET. The terminals of mobile set is located closer to communicating point that assist each other communication, sharing of resources or services and the generate limited period of computing time as well as within a defined space limit have generated a spontaneous adhoc network. Moreover, the network management is transparent to the user and the network type are independent in centralized administration. Hence, the user has the capability in entering and existing the network easily and the significant research area in MANET can be establish and maintain the ad hoc network usage of routing protocols [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. The process of traffic path selection over a single or multi-network for sending and receiving data. This straightaway passes the logical addressed packet from their source towards overall destination via intermediate nodes. Based on the definite rules and recommendations, the data packets are routed is said to be routing protocol. Each routing protocol obtain its individual algorithm with respect to identification and maintenance of route. However, each routing protocol consists of data structure that stored the data route as well as alter the table in accordance with need of route maintenance. Hence, the routing metric is calculated through values that utilized by routing algorithm for determining the performance of routing of each node is compared with one another. Measurements may include data on bandwidth, delay, hop count, path cost, load, reliability, and communication costs. In contrast to link-state or topological databases, which can maintain all data, the routing table simply stores the optimum routes. The most significant development in the telecommunications industry right now is MANETs [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe Intruder Detection System (IDS) can detects intruders who attempt to steal data from the system. The main objective is in detecting communication attacks and send an alert to the administrator of network which perform as a significant system in detecting malicious data in the training mode [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. In general, the intrusion has been detected through traditional methods namely firewall, authentication encryption as well as decryption, etc. that can be classified into three categories are\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eAnomaly Detection System (ADS)\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eMisuse Detection System (MDS)\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eHybrid Detection System (HDS)\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eIn ADS, the intrusions are checked at any sample deviation from the baseline is identified as the malicious node. In the case of MDS, the node used is executed through sample matching that have sorted in the database as well as make decisions. The HDS is combination of both ADS and MDS properties that minimize the flaws of detecting system which is robust detecting system than ADS and MDS to provide suitable decision making [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. In MANET, the IDS may perform individually from its wired counterparts. If the advancement is done for IDS in MANET may solve the challenges needed. All unique activities have been tracked and recorded through node level agents are implemented in the non-collaborative intruder monitoring system [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. The agents position is determined, if the nodes are mobile and challenge lies are highly significant.\u003c/p\u003e \u003cp\u003eNowadays, the ML is most widely used technology that enables computer systems in learning dynamically with no human intervention and also provides suitable action. It creates a model by manipulating complex data effortlessly, efficiently, and precisely. The generalized structure may assist ML in providing suitable patterns for improving device performance. There are numerous scientific applications, including manual data entry, medical diagnostics, automatic spam detection, data clearing, image recognition, noise reduction, etc. [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e],[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. According to the most recent research, ML is utilized in WSNs for addressing variety of issues. The usage of ML in WSNs not solely improves system efficiency and even avoids challenging issues like reprogramming, manually finding massive amounts of data, and collecting valuable data from dataset. ML methods are frequently useful in acquiring large amounts of data as well as producing valuable information [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. Denial of service (DoS) attacks, aimed at preventing legitimate handlers from retrieving or using several network resources, have been common in network examination analysis. A resilience scheme contrary to these attacks has been proposed for AODV, and the efficiency of this scheme is demonstrated by Light Gradient Boosting Machine (LGBM), Gradient Boosting, Hist Gradient Boosting, Radom Forests, Decision Trees, Extra Tree Classifiers, Support Vector Classifiers, Bagging, K Neighbors, Extreme Gradient Boosting (XGB). Decision Tree (DT) Support Vector Machine (SVM), Naive Bayes (NB), etc. have been used for intruder detection [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e].\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eThis paper has structured with section II discusses the literature review in improving the accuracy in detecting intruder and classifying IDS prediction for AODV routing protocol in MANET using ML techniques. Section III discusses various sequential pattern mining in ANN model with various optimizer for identifying the high prediction in IDS classifications over MANET. Section IV deals with evaluation of various ANN model with different sequential pattern, activations and optimizer is determined with accuracy, sensitivity and specificity performance. Session V has concluded that ANN-Model 3 has better accuracy in predicting IDS classification in MANET while compared to other ANN model sequence pattern combinations.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e"},{"header":"2. Literature Review","content":"\u003cp\u003eDoS attack has become a well-known attack that attracts the interest of several researchers and focuses on providing security. However, the solutions from the proposed researcher have only limited success against DoS attack impact in MANET. The literature review discusses IDSs are based on various detection techniques and ANN model approach in detecting malicious node in AODV routing protocol over MANET.\u003c/p\u003e \u003cp\u003eYuan et al., [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e] compared MANET-mediated detection methods. Various attacks such as Black hole, gray hole, and war hole attack detection approaches are described. The weaknesses and strengths of these approaches were demonstrated. In this paper, we consider MANET using the AODV protocol. They analyzed the underlying AODV protocol used for route direction. The result is a Power overhead value of 0.74%. Panos et al., [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e] proposed an AODV protocol monitoring DOS attacks. MANETs are improvements to the AODV protocol. Using the trust value for each node, they were able to identify malicious nodes. Here the author has worked on neural networks. They used Transmission Control Protocol (TCP) for that MANET rating. S. Sargunavathi et al. have illustrated an efficient trust based IDS by local data to data transmission for high dynamic MANET. In this research, blackhole as well as greyhole attacks have been addressed to select trusted nodes for ensuring security. Thus, the proposed method gets compared through conventional AODV protocol with several performance metrics [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eSultan Mohamad, Sayed Hesham, and Khan Manzoor are majorly focuses in designing and investigating using the ANN model for detecting intrusion in MANETs. The major goal of this study is for analyzing, simulating and evaluating the feedforward neural network benefits with back propagation over MANETs. The dataset extraction is generated through simulations manner for MANET has been utilized for evaluating the proposed method input parameters. Moreover, the RMSR is introduced as a metric for evaluating the proposed ANN model performance. Hence, the proposed ANN model is used in detecting DoS attacks in MANET. Thus, the accomplished result of the proposed model is associated with 4-15-10-1 network which generates an outcome of RMSE as 0.0452 is executed for 14 epochs with the train dataset and RMSE as 0.0512 is executed for 14 epochs with the test dataset [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. Pravin kumar et al. has addresses the issue of packet loss in MANET by congestion control mechanism with NN to generate stable data transfer in MANET [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. M.P. Arthur has introduced a cross layer-based IDS to assist in generating exchange routing data among various layers of two nodes from their neighbors. Additionally, the adoption of the multiclass SVM method is introduced for reducing the overhead as well as improving accuracy in classifying the attacks in MANET [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. Basomingera et al. has explained the novel route cache mechanism with SVM method to progress IDS in non-clustering MANET. Thus, the simulation outcome illustrates that the rate of IDS has increased to 95% in MANET [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. S. Pandey and V. Singh have recommended a secured AODV (S-AODV) routing techniques using ANN and SVM methods. The performance of the proposed S-AODV is evaluated through simulation results to determine that it obtains better performance than traditional ML with a routing protocol model under various traffic cases under DoS attack [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. J. Kim et al. has introduced Long Short Term Memory - Recurrent Neural Network (LSTM-RNN) for accomplishing better classifier in IDS with 96.93% accuracy and 10% False Positive Rate (FPR) in the KDD dataset [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. Yujie Fan et al. has illustrated the three-step process to create automatic malware detection using malicious sequential pattern mining. The three processes are\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eIntrusion sequence extractor\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eMiner of malicious sequential pattern\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eANN with kNN classifier for prediction\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eIn order to detect and predict the various threats and attacks through ANN and kNN classifiers whereas other such signature-based detection, as well as feature-based detection, is involved in this research for determining the better prediction of detecting intrusion [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e].\u003c/p\u003e"},{"header":"3. Research Methodology","content":"\u003cp\u003eThis research purpose in proposing sequential ANN model using hidden dense layers modification are used in determining the intruder present in the MANET through various optimizer selections. The major goal of optimizer is initiating it\u0026rsquo;s parameter in the training data and associate with batch size and activation functions. Once the optimal modification is done through sequential pattern mining that can able to manage the iteration of epoch training to improve the accuracy of training model. The various attributes provided with dataset input from the sender user to receiver user in the MANET is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe MANET dataset get collected as well as verified with missing data is performed using is.null() function as missing imputation process and the data preprocessing is executed. In data preprocessing, normalization is done to define the target \u0026ldquo;attack_type\u0026rdquo; in term of ordinal as 0 to 4 shown in Table \u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e is determined through labelEncoder(). This function is used for converting data type of the object in term of integer or float data type by considering a continuous variable as well as ID number variable has been dropped.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eAssigning value for attack type\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSl. No\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAttack Type\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLabel Encoder Assigned\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGrayhole\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNormal\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBlackhole\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTDMA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFlooding\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e illustrates the correlation heat map in which the coloration among every attributes in the MANET dataset have been analyzed to understand the significance of the attributes present. The relationship of data_s and sch_r is 0.74 is highly correlated followed by send_code and sch_r is 0.70 as well as dis_to_ch and sch_r is 0.69. The standardScaler() function is used for scaling all attributes into single scale.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eIn this research, there are 16 input variable get preprocessed by missing imputation, normalization and scaling all variable unit using standard scalar. The dataset get split into 80% as train dataset and 20% as test dataset. The ANN-Model 1 has trained using ANN algorithm with sequential pattern as well as setting the layer shapre of input layer as 32 as unit whereas the output dense layer as 5 and the activation for input as Rectified Linear unit (Relu) and output as sigmoid. The ANN-Model 2 is with sequence pattern and set the input layer shape as 32 as unit, the output dense layer as 5 and the activation for input as Leaky.Relu with alpha as 0.2 and output as sigmoid. The ANN-Model 3 with sequence pattern and set the input layer shape as 32 as unit and input layer as 16, the output dense layer as 5 and the activation for input as Relu and output as sigmoid. The ANN-Model 4 with sequence pattern and set the input layer shape as 32 as unit and input layer as 16, subsequent layer as 8, the output dense layer as 5 and the activation for input as Relu and output as sigmoid. Moreover, this model has used \u0026ldquo;relu\u0026rdquo; as an activation as well as fit with batch size as 5 and 10 epochs have been deliberated with adam optimizer. In addition, ANN-Model 5 has trained by ANN technique as well as setting layer shape of input as 32 whereas the hidden layer shape 16, output layer shape as 5 and optimizer as Stocastic Gradient Descent (SGD). In the case of ANN-Model 6, input layer shape is set input layer shape as 32 in which the hidden layer shape 16, output layer shape as 5 and optimizer as RMSprop in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. The best hidden layer has obtained from adequate sequence pattern and better optimization process is considered to accomplish high accuracy.\u003c/p\u003e"},{"header":"4. Working of ANN model building","content":"\u003cp\u003eThe ANN model has acquire from sequence data patterns and tried to accomplish as probable as an accurate prediction. Assume, the set of X data pairs containing the variables and the results, (m\u003csup\u003e1\u003c/sup\u003e, t\u003csup\u003e1\u003c/sup\u003e), (m\u003csup\u003e2\u003c/sup\u003e, t\u003csup\u003e2\u003c/sup\u003e), \u0026hellip;.(m\u003csup\u003eX\u003c/sup\u003e, t\u003csup\u003eX\u003c/sup\u003e) in which the m\u003csup\u003ei\u003c/sup\u003e is the input value and t\u003csup\u003ei\u003c/sup\u003e is the target value for i\u0026thinsp;=\u0026thinsp;1, 2, 3, \u0026hellip;, X. We would like to build a neural net F so that ideally as\u003cdiv id=\"Equ1\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ1\" name=\"EquationSource\"\u003e\n$$F\\left({m}^{i}\\right)= {t}^{i}$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e3.1\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eMoreover, it allow for error \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\({\\in }^{i}\\)\u003c/span\u003e\u003c/span\u003e typically. Let n denotes the output of ANN is expressed in Eq.\u0026nbsp;\u003cspan refid=\"Equ2\" class=\"InternalRef\"\u003e3.2\u003c/span\u003e and \u003cspan refid=\"Equ3\" class=\"InternalRef\"\u003e3.3\u003c/span\u003e.\u003cdiv id=\"Equ2\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ2\" name=\"EquationSource\"\u003e\n$${n}^{i}=F\\left({m}^{i}\\right)$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e3.2\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eAnd\u003cdiv id=\"Equ3\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ3\" name=\"EquationSource\"\u003e\n$${t}^{i}={n}^{i}+{\\in }^{i}$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e3.3\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eHowever, the\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\({n}^{i}\\)\u003c/span\u003e\u003c/span\u003e is based on the parameter with respect to weight and bias which turn as an optimizer issue. In this, setting the ANN in F that minimize the error function represent in Eq.\u0026nbsp;\u003cspan refid=\"Equ4\" class=\"InternalRef\"\u003e3.4\u003c/span\u003e.\u003cdiv id=\"Equ4\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ4\" name=\"EquationSource\"\u003e\n$$E= \\frac{1}{N}\\sum _{i=1}^{X}{‖{t}^{i}-{n}^{i}‖}^{2}$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e3.4\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eWhere,\u003c/p\u003e \u003cp\u003eN\u0026thinsp;=\u0026thinsp;Number of training patterns.\u003c/p\u003e \u003cp\u003eWhen the ANN has become a two-way classification then N\u0026thinsp;=\u0026thinsp;2. Based on the equation, E is a parameter function of F, and required in determining the weight values that minimize the error through differentiating E.\u003c/p\u003e \u003cp\u003eWhen the research is focused on one term of the sum and it is expressed in Eq.\u0026nbsp;\u003cspan refid=\"Equ5\" class=\"InternalRef\"\u003e3.5\u003c/span\u003e.\u003cdiv id=\"Equ5\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ5\" name=\"EquationSource\"\u003e\n$${‖t-n‖}^{2}= {\\left({t}_{1}-{n}_{1}\\right)}^{2}+{\\left({t}_{2}-{n}_{2}\\right)}^{2}+\\dots +{\\left({t}_{x}-{n}_{x}\\right)}^{2}$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e3.5\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eThus, the values of input and output have been fixed, and only the parameter is consider to be calculated through weight and it can differentiate from both sides is expressed in Eq.\u0026nbsp;\u003cspan refid=\"Equ6\" class=\"InternalRef\"\u003e3.6\u003c/span\u003e.\u003cdiv id=\"Equ6\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ6\" name=\"EquationSource\"\u003e\n$$\\frac{\\partial }{{\\partial W}_{ij}}\\left({‖t-n‖}^{2}\\right)= -2\\left(t-n\\right).\\frac{\\partial n}{\\partial W}$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e3.6\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eThere are highly specific and also verify the fits over the context of neural net. From a neural net, the output is defined as \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\({n}^{i}= {W}_{ij}{m}^{i}+b\\)\u003c/span\u003e\u003c/span\u003e.\u003c/p\u003e \u003cp\u003eHence, the output is rely on weight and while differentiating both sides in accordance with \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\({W}_{ij}\\)\u003c/span\u003e\u003c/span\u003e by chain rule as per Eq.\u0026nbsp;\u003cspan refid=\"Equ7\" class=\"InternalRef\"\u003e3.7\u003c/span\u003e\u003cdiv id=\"Equ7\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ7\" name=\"EquationSource\"\u003e\n$$\\frac{\\partial }{{\\partial \\text{W}}_{\\text{i}\\text{j}}}\\left({‖\\text{t}-\\text{n}‖}^{2}\\right)= -2\\left({\\text{t}}_{\\text{i}}-{\\text{n}}_{\\text{i}}\\right){\\text{m}}_{\\text{i}}$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e3.7\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eWhere,\u003c/p\u003e \u003cp\u003e \u003cspan class=\"InlineEquation\"\u003e \u003cspan class=\"mathinline\"\u003e\\({\\text{m}}_{\\text{j}}\\)\u003c/span\u003e \u003c/span\u003eis the i th coordinate position. The derivative has provided direction to the maximum for accomplishing the minimum point, and subsequently in opposite direction of this gradient. In addition, this derivative as close to 0 as possible for obtaining the minimum of error.\u003c/p\u003e \u003cp\u003eThe ANN method is a layer based network that involves one or more artificial neurons which generally includes input layer, hidden layer as well as an output layer. Figure\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e has illustrates the ANN structure that has capability in identifying, mapping with robust capability and even have capacity in processing data in parallel.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003ea. Input Layer\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThis layer is the top most layer of ANN in which collection of input data is in the form of text, image files, audio files and video files that have received initially.\u003c/p\u003e \u003cp\u003e \u003cb\u003eb. Hidden Layer\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThis layer is the subsequent layer in ANN model that has potential in both perceptron or several hidden layers. Based on the input data, these hidden layers have been executed various mathematical operations as well as recognize the pattern of the ANN model.\u003c/p\u003e \u003cp\u003e \u003cb\u003ec. Output Layer\u003c/b\u003e \u003c/p\u003e \u003cp\u003eBased on the hidden layer input generation from feature weights of middle layers with accurate and exact computations have capability in producing an adequate result in the output layer.\u003c/p\u003e \u003cp\u003eThere are several positive or negative weights have been related to every neurons for execute or prohibit an input with every connection to the artificial neurons. In order to control the performance of the artificial neurons, activation function plays a major role. This artificial neurons assist in collecting input signal through computation of net input signal as a function with its associative weights. The performed input from net input signals to activation functions have calculated the artificial neuron as an output signal which operate various stages in mathematical process whereas the unit numbers are arranged in layer numbers. The single unit is said to be neuron in which the input unit arranged from input layer consumes various inputs from the acquired data. After, the input data is further progressed to hidden layer unit that transform input data into output unit.\u003c/p\u003e \u003cp\u003eRelu is a function and major benefit of Relu activation function and doesn\u0026rsquo;t activate all neuron simultaneously in which a neuron with negative value have converted into zero or it gets deactivated. Networks become sparse as well as computationally effective as a result. The gradient value of the graphs at the negative side is zero. It suggests that neurons terminate are never stimulated during back propagation. The sigmoid function is utilized to maintain issues with multi-class which maps the value of output from 0 to 1. It works best when utilised in the classifier's output layer. There isn't a guideline that may be used to select the activation function. The characteristics of the issue might assist in choosing a quicker converging network. Certain characteristics are based on the research as the Relu and sigmoid for activation functions.\u003c/p\u003e \u003cp\u003eA procedure called optimization aims to decrease network error. This is essential for increasing the model's accuracy. The optimizer has three different iterations: Adam, SGD, and RMSProp. SGD is iterative gradient descent technique that uses iteration to search for an optimal error. These models produce predictions in every iteration that follows and predicted results are compared to the predictions. Error is defined as the difference between the projected value and the actual result. The internal parameters of the model as well as the weights of the network are updated using this error. The back propagation algorithm follows this updating process. SGD also finds it challenging to get away from the saddle points. The most popular choices for handling saddle points are AdaGrad, RMSprop, ADAM, and AdaDelta. To modify the gradient with the slop and quicken the SGD, Nesterov accelerating gradient is utilized. Due to its ability to execute more updates for infrequently parameters as well as fewer changes for frequent parameters, AdaGrad outperforms Nesterov accelerated gradient. As a result, SGD becomes faster, more scalable, and more resilient. In order to train ANN, it is utilized and the fundamental disadvantage in AdaGrad optimizer has minimized the model's ability for training through the sequential pattern better. Two optimizers like RMSProp as well as AdaDelta, have been created individually to address the problem with AdaGrad. Moreover, the optimizer like AdaDelta and RMSProp are identical whereas the main difference is AdaDelta which is not fixed as an early learning rate with constant. Adam is an optimizer which incorporate the beneficial features of RMSprop as well as Adadelta. Adam is deemed a good choice since it get improves RMSProp as well as Adadelta. As a result, the optimizers used in this paper are Adam, RMSProp, and SGD.\u003c/p\u003e"},{"header":"5. Algorithm for Sequence pattern based ANN","content":"\u003cp\u003eSequence based ANN is an efficient algorithm for enumerating the set of layer sequence based on features weight considered through activation function. There are various sequence pattern and optimizer function involved in this experimental algorithm are discussed below.\u003c/p\u003e \u003cp\u003eStep 1: Let initialize the network weight randomly in the training set and f represent the forward pass.\u003c/p\u003e \u003cp\u003eStep 2: Let x be input network, O be the neural network weight during forward pass and assigning the sequence pattern with various dense layer combinations with activation function as Relu as well as leaky Relu in feed forward pass.\u003c/p\u003e \u003cp\u003eStep 3: Considering the training output as T using activation function as sigmoid and calculating error\u0026thinsp;=\u0026thinsp;T-O as an output unit with five different layer.\u003c/p\u003e \u003cp\u003eStep 4: Compute weight for all the features from input layers to hidden layers as a backward pass and even computing hidden layer to output layer.\u003c/p\u003e \u003cp\u003eStep 5: Update the weight of best sequence in the ANN and again selection of various optimizer like adam, SGD and rmsprop involved to obtain high classification accuracy.\u003c/p\u003e \u003cp\u003eStep 6: Repeat step 5 until the optimal accuracy of epochs have obtained from an optimizer.\u003c/p\u003e \u003cp\u003eStep 7: Return\u003c/p\u003e \u003cp\u003eThe representation of epoch\u0026rsquo;s number determines the training dataset iteration is transverse. Each epoch provides training sample which the capability to update parameters of the internal model. The epoch is initiated at zero and incremental till epochs count, as well as several batches. Hundreds or thousands of epochs are typically chosen, allowing the network to effectively minimize the inaccuracy. The programmer must look at the learning curves of error as well as accuracy in choosing the best epoch value. These curves are utilized in identifying the model that has learned excessively, inadequately, or is well prepared for training.\u003c/p\u003e \u003cp\u003eThe training samples quality sent to the networks is known as the batch size. Several batches can be created from a training dataset whereas the algorithm is commonly mentioned as batch gradient descent while any training is given as a single batch. The SGD is batch size is one whereas the learning algorithm is said to be mini-batch gradient descent while the batch size is greater than one but smaller than the training size. The accuracy of the network is improved by the larger batch size which has a power of 2 range of 1 to 1024 provides several key considerations for choosing small batch sizes. Larger batch sizes reduce the model quality because they generally lead to sharp minimizes of the training function.\u003c/p\u003e \u003cp\u003eThe sequential pattern ANN has effectively extract the multifaceted characteristics between the applicable variables of classifying the intruder in the AODV routing. The selection of optimizer is significant for improving the optimum selection of route path in AODV of the MANET. This can be determined through better prediction with model accuracy. Thus, the model efficiency improved by optimizing the best obtained sequence in increasing the predicting IDS accuracy by enhancing the ANN model accuracy.\u003c/p\u003e"},{"header":"6. Results and Discussion","content":"\u003cp\u003eThis research experiment is conducted using an i5-7400 processor running at 3.00 GHz and a GTX 1050 GPU. The computer has 8.00 GB RAM and x64-ased processor. Python programming is used to train the models in the Keras environment and Tensorflow is employed to assess the models. The parameters involve the number of hidden units, activation function, number of layers, optimization, epoch\u0026rsquo;s number, and batch size used for validating the models. A crucial methodological decision in any empirical evaluation of optimizers is choosing the sequential search space associated with every optimizer. However, this necessitates the frequently untrue assumption that variables with identical designations should have comparable values among optimizer.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e illustrates the plotting of ANN sequence with dense layer of 32 units as input, 16 layer as hidden dense layer with an input activation as Relu. The output dense layer with five different units using sigmoid as an activation function in which the classification prediction of security status have been determined. Hence, the determined layer is accomplished with high accuracy in this respective sequence pattern than other sequence pattern like 32 unit with 16 and 8 dense layer as hidden layer in the ANN model.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e illustrates the optimizer iteration with loss function as binary cross entropy in which optimal results is obtained in the adam optimizer shown better accuracy while compared to other optimizer. Thus, the trained optimizer results in classifying the security of AODV routing protocol in MANET can be evaluated through confusion matrix metrics.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e illustrates the ANN model with various sequence pattern and different optimizer whereas the input layer as dense unit as 32, following subsequent layer as 16 with activation as ReLu. It is classified with three different optimizers like Adam, SGD and RMSPROP.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eANN model with various sequential and optimizer\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSl.No\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eANN Model Name\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eOptimizer with different sequential pattern\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eANN Model \u0026ndash; 1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eADAM with I/P \u0026ndash; Dense : 32, O/P \u0026ndash; Dense : 5 I/p Activation: ReLu\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eANN Model \u0026ndash; 2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eADAM with I/P \u0026ndash; Dense : 32, O/P \u0026ndash; Dense : 5, I/p Activation: LeakyReLu\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eANN Model \u0026ndash; 3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eADAM with I/P \u0026ndash; Dense : 32, Next I/P \u0026ndash; Dense : 16,O/P \u0026ndash; Dense : 5\u003c/p\u003e \u003cp\u003eI/p Activation: ReLu\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eANN Model \u0026ndash; 4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eADAM with I/P \u0026ndash; Dense : 32, Next I/P \u0026ndash; Dense : 16, Next I/P \u0026ndash; Dense : 8\u003c/p\u003e \u003cp\u003eO/P \u0026ndash; Dense : 5, I/p Activation: ReLu\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eANN Model \u0026ndash; 5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSGD with I/P \u0026ndash; Dense : 32, Next I/P \u0026ndash; Dense : 16,,/P \u0026ndash; Dense : 5\u003c/p\u003e \u003cp\u003eI/p Activation: ReLu\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eANN Model \u0026ndash; 6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eRMSPROP with I/P \u0026ndash; Dense : 32, Next I/P \u0026ndash; Dense : 16, O/P \u0026ndash; Dense : 5\u003c/p\u003e \u003cp\u003eI/p Activation: ReLu\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eANN model with various sequential and optimizer\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eClassification ANN model\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"6\" nameend=\"c7\" namest=\"c2\"\u003e \u003cp\u003eConfusion matrix metrics\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMicro Precision\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMicro Recall\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMicro F1-Score\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eWeighted Precision\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eWeighted Recall\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eWeighted F1-Score\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eANN Model \u0026ndash; 1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e96.28\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e96.28\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e96.28\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e96.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e96.28\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e96.45\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eANN Model \u0026ndash; 2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e90.33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e90.33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e90.33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e93.25\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e90.33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e91.37\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eANN Model \u0026ndash; 3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e96.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e96.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e96.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e95.90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e96.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e96.89\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eANN Model \u0026ndash; 4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e94.96\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e94.96\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e94.96\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e94.99\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e94.96\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e95.24\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e illustrates the evaluation of ANN model with several sequence pattern with Adam optimizer whereas the input layer as dense unit as 32, following subsequent layer as 16 with input activation as ReLu except ANN model 2 in which the input activation as leaky ReLu but the out activation function used is sigmoid.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e illustrates the micro and weighted value to four different ANN models with adam optimizer in several sequence pattern. The accuracy is said to be micro F1-Score for ANN-Model 3 is 96.75% which is comparatively higher than other sequential pattern model. The proposed sequential pattern of ANN-Model 4 has better prediction in classifying IDS from the AODV routing protocol in MANET.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003e has illustrate the sensitivity and specificity of various ANN models in which ANN models has sensitivity as 0.973 and specificity as 0.987 which is higher than other ANN models. Hence, it determines high true positive rate is better in ANN-Model 4 illustrates that IDS classification prediction is high in ANN-Model 4 than other model.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003e has illustrates the accuracy of ANN model with various optimizer in which ANN-Model 3 as ADAM, ANN-Model 5 as sgd and ANN-Model 6 as rmsprop. However, the accuracy of adam optimizer in ANN model as 96.75% is higher than other optimizer ANN model. Hence, adam optimizer with sequence pattern of ANN-Model 3 has better accuracy in predicting IDS classification through AODV routing protocol in MANET.\u003c/p\u003e"},{"header":"7. Conclusion","content":"\u003cp\u003eProviding security necessitates in maintaining unauthorized users from accessing data or other items and guarding against unauthorized changes to or destruction of user data. According to the traditional definition, security also includes availability, confidentiality, and integrity. The desirable features of wireless sensor network are in paying attention to many researchers to act based on various issues of security requirements. To solve the mentioned problems, ANN technique with different sequential pattern as the classifier in identifying IDS as well as route discovery for network shield from DoS attack. The MANET is trained by lazy predict classifier library for identifying the malicious node that appears in the network. The route has modified through elimination of malicious nodes from route as well as network is secured. Moreover, this proposed sequential ANN model with best optimizer has influences better understanding of data and resulted with high predictable accuracy as 96.75% which is higher than other ANN models. Thus, the evaluation of proposed ANN model produce high recognition in IDS as well as tracking shortest path in AODV routing protocol over MANET.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors have not received any funding for the work presented in the paper.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data and source codes used to support the findings of this study can be obtained from the corresponding author upon request.\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflicts of Interest\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they \u0026nbsp;have \u0026nbsp;no competing interests.\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthorship Contribution Statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSivanesan N: \u0026nbsp;\u003c/strong\u003eMethodology, Formal Analysis, Writing of Original Draft, Validation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRajesh A\u003c/strong\u003e: \u0026nbsp;Supervision, Reviews, Editing and Validation\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eT. Salam and M. S. Hossen, \u0026ldquo;Performance analysis on homogeneous LEACH and EAMMH protocols in wireless sensor network,\u0026rdquo; Wireless Personal Communications, vol. 113, no. 1, pp. 189\u0026ndash;222, 2020.\u003c/li\u003e\n\u003cli\u003eM. S. Hossen, \u0026ldquo;DTN routing protocols on two distinct geographical regions in an opportunistic network: an analysis,\u0026rdquo; Wireless Personal Communications, vol. 108, no. 2, pp. 839\u0026ndash; 851, 2019. \u003c/li\u003e\n\u003cli\u003eM. Singh, C. Kumar, and P. Nath, \u0026ldquo;Challenges and protocols for P2P applications in multi-hop wireless networks,\u0026rdquo; in in 2018 Second International Conference on Computing Methodologies and Communication (ICCMC), pp. 310\u0026ndash;316, IEEE, 2018. \u003c/li\u003e\n\u003cli\u003eT. Qiu, N. Chen, K. Li, D. Qiao, and Z. Fu, \u0026ldquo;Heterogeneous ad hoc networks: architectures, advances and challenges,\u0026rdquo; Ad Hoc Networks, vol. 55, pp. 143\u0026ndash;152, 2017.\u003c/li\u003e\n\u003cli\u003eM. Manjunath and D. H. Manjaiah, \u0026ldquo;Performance Comparative of AODV, AOMDV and DSDV Routing Protocols in MANET Using NS2 Alamsyah1,2,\u0026rdquo; Int. J. Commun. Netw. Syst., vol. 004, no. 001, pp. 18\u0026ndash;22, 2018, doi: 10.20894/ijcnes.103.004.001.005.\u003c/li\u003e\n\u003cli\u003eM. G. K. Alabdullah, B. M. Atiyah, K. S. Khalaf, and S. H. Yadgar, \u0026ldquo;Analysis and simulation of three MANET routing protocols: A research on AODV, DSR \u0026amp; DSDV characteristics and their performance evaluation,\u0026rdquo; Period. Eng. Nat. Sci., vol. 7, no. 3, pp. 1228\u0026ndash;1238, 2019, doi: 10.21533/pen.v7i3.717.\u003c/li\u003e\n\u003cli\u003eFeng, F., Liu, X., Yong, B., Zhou, R., Zhou, Q.: Anomaly detection in ad-hoc networks based on deep learning model: a plug and play device. Ad Hoc Netw. 84, 82\u0026ndash;89 (2019)\u003c/li\u003e\n\u003cli\u003eLiu, G., Yan, Z., Pedrycz, W.: Data collection for attack detection and security measurement in mobile ad hoc networks: a survey. J. Netw. Comput. Appl. 105, 105\u0026ndash;122 (2018).\u003c/li\u003e\n\u003cli\u003eSuma R, Premasudha BG, Ram VR. A novel machine learning-based attacker detection system to secure location aided routing in MANETs. International Journal of Networking and Virtual Organisations. 2020;22(1):17-41.\u003c/li\u003e\n\u003cli\u003eBi J, Yuan H, Zhou M. Temporal prediction of multiapplication consolidated workloads in distributed clouds. IEEE Transactions on Automation Science and Engineering. 2019 Feb 21;16(4):1763-73. \u003c/li\u003e\n\u003cli\u003eBi J, Yuan H, Zhang L, Zhang J. SGW-SCN: An integrated machine learning approach for workload forecasting in geo-distributed cloud data centers. Information Sciences. 2019 May 1;481:57-68. \u003c/li\u003e\n\u003cli\u003eWang J, Gao Y, Yin X, Li F, Kim HJ. An enhanced PEGASIS algorithm with mobile sink support for wireless sensor networks. Wireless Communications and Mobile Computing. 2018 Dec 2;2018.\u003c/li\u003e\n\u003cli\u003eN. Khanna and M. Sachdeva, \u0026apos;\u0026apos;A comprehensive taxonomy of schemes to detect and mitigate blackhole attack and its variants in MANETs,\u0026apos;\u0026apos; Comput. Sci. Rev., vol. 32, pp. 24\u0026ndash;44, May 2019.\u003c/li\u003e\n\u003cli\u003eX. Yuan, C. Li, and X. Li, \u0026quot;DeepDefense: Identifying DDoS attack via deep learning\u0026quot;, \u003cem\u003eProc. IEEE Int. Conf. Smart Comput.\u003c/em\u003e, pp. 1-8, May 2017.\u003c/li\u003e\n\u003cli\u003eC. Panos, C. Ntantogian, S. Malliaros, and C. Xenakis, \u0026lsquo;\u0026lsquo;Analyzing, quantifying, and detecting the blackhole attack in infrastructure-less networks,\u0026rsquo;\u0026rsquo; Comput. Netw., vol. 113, pp. 94\u0026ndash;110, Feb. 2017. \u003c/li\u003e\n\u003cli\u003eSargunavathi, S. and Martin Leo Manickam, J., 2019. Enhanced trust based encroachment discovery system for Mobile Ad-hoc networks. Cluster Computing, 22(2), pp.4837-4847. https://link.springer.com/article/10.1007/s10586-018-2405-7.\u003c/li\u003e\n\u003cli\u003eSultan, Mohamad \u0026amp; Sayed, Hesham \u0026amp; Khan, Manzoor., An Intrusion Detection Mechanism for MANETs Based on Deep Learning Artificial Neural Networks (ANNs), International Journal of Computer Networks \u0026amp; Communications (IJCNC) Vol.15, No.1, January 2023.\u003c/li\u003e\n\u003cli\u003eKumar, P., Tripathi, S. and Pal, P., 2018, March. Neural network based reliable transport layer protocol for MANET. In 2018 4th International Conference on Recent Advances in Information Technology (RAIT) (pp. 1-6). IEEE. \u003c/li\u003e\n\u003cli\u003eArthur, M.P., 2018, September. An SVM-based multiclass IDS for multicast routing attacks in mobile ad hoc networks. In 2018 International Conference on Advances in Computing, Communications and Informatics (ICACCI) (pp. 363-368). IEEE. \u003c/li\u003e\n\u003cli\u003eBasomingera, R. and Choi, Y.J., 2019, January. Route cache based SVM classifier for intrusion detection of control packet attacks in mobile ad-hoc networks. In 2019 International Conference on Information Networking (ICOIN) (pp. 31-36). IEEE.\u003c/li\u003e\n\u003cli\u003ePandey, S. and Singh, V., 2020, July. Blackhole attack detection using machine learning approach on MANET. In 2020 International Conference on Electronics and Sustainable Communication Systems (ICESC) (pp. 797-802). IEEE.\u003c/li\u003e\n\u003cli\u003eThu H L T Kim J, Kim J et al. Long short term memory recurrent neural network classifier for intrusion detection. Platform Technology and Service (PlatCon), International Conference on. IEEE, 2016.\u003c/li\u003e\n\u003cli\u003eY. Fan, Y. Ye and L. Chen, \u0026quot;Malicious sequential pattern mining for automatic malware detection\u0026quot;, Expert Systems with Applications, vol. 52, pp. 16-25, 2016.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"MANET, AODV, Machine Learning (ML), Intruder detection, sequential pattern. ","lastPublishedDoi":"10.21203/rs.3.rs-3199495/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3199495/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eMobile Adhoc Network (MANET) is an adoptable network with dynamic as well as decentralized network in nature. This research concentrates to resolve Denial-of-Service (DoS) attacks in MANET and illustrates the usual classification models, which may be unable to distinguish between legitimate DoS attacks as well as network problems. The routing path has been recognized as well as strengthen the environmental adoption with respect to several logic and statistical performances using Machine Learning (ML). The main aim in ML for recognizing the complexity pattern with AODV routing in MANET as well as decision making in accordance with accomplished result. In securing the MANET, there are several ML algorithms have been implemented. The lack of infrastructure in MANETs makes it extremely difficult to implement security measures. Moreover, the proposed sequential pattern with Artificial Neural Network (ANN) for AODV routing has generated better security from DoS attack. Thus, the security methods in MANETs majorly concentrating in mitigating intruder detection, eliminating malicious node as well as securing routing paths.\u003c/p\u003e","manuscriptTitle":"Mitigating Intruder Detection System in Mobile Adhoc Network (MANET) using optimizer based ANN model","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2023-08-02 16:49:51","doi":"10.21203/rs.3.rs-3199495/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"72bef5dd-0206-4a2f-ae9a-9fdb2cdb9234","owner":[],"postedDate":"August 2nd, 2023","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2023-08-08T21:29:16+00:00","versionOfRecord":[],"versionCreatedAt":"2023-08-02 16:49:51","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-3199495","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3199495","identity":"rs-3199495","version":["v1"]},"buildId":"J0_U0BvcaRcwD8yVFaRlm","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}