| Algorithm 3: SVM with ANN |
|
Input Parameters: FR → Final Route ONPROP → Optimized Nodes Properties as a Training Data (T), C → In terms of communicative and non-communicating nodes, the target/category is Output: OR → Optimized Route for discovering a route from TX-Node to RX-Node and Malicious Nodes (M-Nodes) 1. Start Training 2. Set up the SVM training data. With RBF as the Kernel function, ONP is the total nodes optimised property. 3. For I = 1 → All Nodes 4. If Property of Node (I) == Real 5. Defined the Cat as a category of training data 6. Cat (1) = ONPROP (I) 7. Else 8. Cat (2) = ONPROP (I) 9. End—If 10. End—For 11. Train_Structure = SVMTRAIN (T, Cat, Kernel function) 12. OT= Train_Structure. Support-Vector //To find out the training data for ANN 13. Initialize the basic parameters of ANN—Number of Epochs (E) // Iterations used by ANN —Number of Neurons (N) // Used as a carrier in ANN —Performance: MSE, Gradient, Mutation, and Validation —Techniques: Levenberg Marquardt —Data Division: Random 14. For i = 1 → OT 15. If T belongs to communicating nodes property 16. Group (1) = Properties of training data according to the real nodes 17. Else if T belongs to non-communicating nodes property 18. Group (2) = Properties of training data according to the non-real nodes 19. Else 20. Group (3) = Extra properties of training data 21. End—If 22. End—For 23. Initialized the ANN using Training data and Group 24. MANET-Net = Newff () 25. Set the training parameters according to the requirements and train the system 26. VoIP = Train (MANET-Net, Training data, Group) Testing: 27. Current Node = Properties of the current node in MANET -Net 28. Authentication = simulate (MANET-Net, Current Node) 29. If Authentication = True 30. Genuine Nodes do not consider as a malicious 31. Else 32. M-Nodes = Malicious Node 33. End—If 34. Create an Optimized Route, OR = FR (Genuine Nodes) 35. Return: OR as an Optimized Route (OR) and Malicious Nodes (M-Nodes) 36. End—Function |