Mobile Ad-Hoc Networks (MANET) are becoming extremely popular as a result of their potential to offer inexpensive solutions to practical communication issues. MANETs are more vulnerable to security attacks due to their characteristics, including node mobility, lack of centralized control and limited bandwidth. Traditional cryptography techniques cannot entirely protect MANETs from new threats and weaknesses to address these security challenges. So, this paper brings an effective integration of cryptography with revolutionized deep learning technology to effectively detect the routing attacks in which following are the stages. a) Data Collection from popular repositories like AWID, DARPA, and UNSW-NB15 containing possible routing attacks in network and b) Preprocessing these with techniques like missing data, redundant data, noisy data and data cleansing c) feature extraction using autoencoder d) feature selection using Particle Swarm Optimization and finally e) classification using LSTM. Also, to secure the whole network along with security features of LST use False Advance Encryption Standard (Faa ES) as a cryptographic method as well. Experimental results were conducted over various state-of-art models under various measures (accuracy:0.97, precision:0.89, detection rate: 0.94).

MANET , Particle Swarm Optimization, False Advance Encryption Standard , LST