An intelligent and secure IoT-based framework for crop price prediction in smart agriculture

Internet of Things (IoT) has become a revolutionary technology that links sensors, devices, farm machinery, and
stakeholders to enhance efficiency and decision-making in farming. Smart agriculture will allow cost savings,
increase in yields, optimization of resources, and remote monitoring. Nonetheless, the extensive implementation
of IoT systems presents security risks of cyber attacks that can undermine the integrity of data and farm activities.
At the same time, climate variability, market demand, and uncertainty in the supply chain lead to economic risks
to rural farmers due to fluctuation in crop prices. This study proposes a Crop Price Prediction and Recommen
dation System for Rural Farmers (CPP-RSRF) approach. The major intention of this paper is to present a secure
IoT-based system for accurate crop price prediction. The proposed model implements dual stages, namely,
predicting crop prices and the attack detection process. For the crop price prediction stage, the proposed CPP-
RSRF technique primarily pre-process the input data using min-max scaling normalization to enhance further
investigation. Following that, a temporal convolutional network and transformer with a self-attention module are
employed to predict crop price. Along with that, the rectified adam is utilized for tuning parameters and boosting
the model's performance. For the attack classification stage, an information gain is used for selecting features and
determining the most informative features. In addition, the graph convolutional network is used for securing IoT
in the agricultural section via intrusion detection process. Experimental findings indicate that CPP-RSRF had
higher forecasting accuracy and intrusion detection performance than recent baseline models, and thus can be
used in secure smart agriculture applications