Now in these days location aware application development and services are in fashion. These applications are providing services according to the locality of users. Therefore, location estimation is a branch of communication and computing which provides ease in service distribution. In our proposed work, different techniques of location estimation methods are investigated for accurate positioning information with less resource consumption. In search of efficient techniques, hybrid predictive data models are targeted for investigation. We first re-implement and analyse the two best available predictive technique i. e. SMOreg and PSO. After successfully executing the simulation of these techniques, we observed that these methods are able to predict the locations of mobile node, but these methods consume too much time during historical data analysis. Therefore a new adaptive technique is required to design by which in less training time maximum mobility patterns can be learned. With the same motivation, here we are proposing our adaptive localization approach; the main advantage of our technique over other exiting technique is that it consumes only meaningful mobility patterns for position approximation. The obtained results demonstrate PSO provide low performance as compared to SMOreg and SMOreg provides less accurate results with respect to proposed algorithm. Therefore the proposed algorithm is adoptable due to higher learning capability with less number of training cycles.

WSN ANN BPN SMOreg PSO Epoch Cycle Generation.