Retraction Notice: A CNN and LSTM-based Data Prediction Model for WSN
Anand Kumar Dohare, Tulika · 2021 3rd International Conference on Advances in Computing, Communication Control and Networking (ICAC3N) · 2021
Due to the limited energy, the growth in network size and sensory data causes a slew of severe issues for wireless sensor networks. Data prediction methods are useful for reducing network traffic and increasing network lifespan, particularly when data correlation across sensory data is explored. In the event that these sensor nodes fail, data prediction may be utilised to retrieve aberrant or lost data. Current wireless sensor network prediction techniques do not fully use the spatial-temporal relationship between remote sensor hubs, resulting in greater prediction error. In a wireless sensor network, this article offers a new approach for multi step sensory data prediction. To begin, we present ANN (artificial neural networks) based on 1-D CNN that use pre-processed sensory input to extract abstract characteristics of various qualities. Then, using these abstract characteristics, one-step prediction is obtained. Finally, by repeatedly utilising chronicled information and the forecast outcomes of the preceding phase, the multi-step expectation is presented. The suggested multi-step prescient model can foresee multi step (mid and short term) tactile information after choosing appropriate hub mixes in which the spatial-fleeting connection is emphasised, according to experiment findings, and its performance is superior than other similar approach.