Role of Neural Network in Mobile Ad Hoc Networks for Mobility Prediction

The MANETs differ from traditional networks in a lot of aspects, such as high channel error rates, unusual channel features, frequent link breaks, and intense link layer contentions. These characteristics significantly reduce network connectivity, which affects overall network latency, network overhead, network throughput (i.e. the amount of data successfully transferred via a MANETs in a predetermined amount of time), and packet delivery ratio (PDR). For effective network resources preparation and organization in MANETs, the mobility forecast of MN and units is essential. This effectiveness would allow for better planning and higher overall quality - of - service, including reliable facility availability and efficient management of energy. In this research, we suggest to use ELMs, which are renowned for their ability to approximate anything, to model and forecast the mobility of each node in a MANET. Mobility-aware topology control methods and location-assisted routing both leverage mobility prediction in MANETs. It is assumed that each MN taking part in these protocols is aware of its current mobility data, including location, velocity, and movements direction angle. This approach predicts both the locations of future nodes and the distances between subsequent nodes. The interaction or relationship between the Cartesian longitude and latitude of the erratic nodes is better captured by ELMs than by multilayer perceptron's, resulting in mobility prediction that is based on several conventional mobility models that is more precise and realistic.