A Hybrid Framework for IDS in IOT Data with RLSTM and GWO

One of the emerged and exciting area of present scenario is Internet of Things(IOT). Intruder Detection (ID) in IOT is one of the key concept and it needs immediate action. Security attacks on IoT networks can be detected and prevented intelligently by Deep Learning(DL). Popular technique which can capable of extract essential features of IDs is LSTM (Long Short-Term Memory Network) . However, LSTM requires a lot of iterations to attain good performance. RLSTM (recurrent LSTM) prevents backpropagated errors from growing out of control. In this model, at each virtual layer which are unfold affect by the errors. RLSTM has the benefit of being relatively insensitive to gap length, in contrast to regular RNNs, which become less effective as the gap length increases. The study , applied Grey Wolf Optimization (GWO) to the RLSTM to enhance the performance of model. A hybrid idea of the model done in two stages. The first involves training an RLSTM network to obtain initial weights, and the second involves utilizing GWO to optimize the RLSTM weights to enhance accuracy. The RLSTM was then utilized to create a very effective IDS for binary and multi-class classification scenarios. The performance of the model is tested over public data sets which includes UNSW-NB151 and BOT-IOT, to ensure that it is adaptable to diverse datasets. The proposed model is also tested against the standard approaches of ID and is attained better results.