Implementing the Markov Decision Process for Efficient Water Utilization with Arduino Board in Agriculture

Normally, the conventional irrigation system is implemented by setting a soil humidity threshold level through sensors in agriculture. If the current soil humidity is lower than expected soil humidity that the plants need for development, water is supplied to raise the current humidity. This threshold level based decision is quite slow and consumes a lot of water for watering plant and raising soil humidity. In this paper, the Markov Decision Process (MDP) with a reward function is proposed for efficient water utilization. Based on the prediction of soil humidity in the near future, the MDP makes a decision for watering. By doing so, the water can be supplied to plants in well-time for expected water demand. Compared to the threshold level based decision, the MDP can save more than 63% water and energy consumption. Moreover, the commercial Arduino board is used to implement this MDP with low cost in reality.