On Optimal Crowd-Sensing Task Management in Developing Countries

In developing countries, crop field productivity is particularly vulnerable to spreading diseases, including viruses and fungi. This is mostly due to the lack of skilled plant pathologists as well as to the scarce fund and poor infrastructure (e.g., roads, power and water lines) availability. The PlantVillage project through its mobile application named Nuru provides an AI digital assistant to recognize plants and their diseases through image analysis. Through the use of Nuru endowed smartphones, farmers can participate in a mobile crowd-sensing framework to improve their crop production. The crowd sensing framework also contributes to early detection of the outbreak of spreading diseases across geographical regions, and consequent adoption of appropriate countermeasures to ensure food security. As devices are often granted in a limited number by countries' government or charities, we propose a Farmer to Farmer (F2F) cooperation to achieve the required Quality of Information (QoI) for the system. In particular, only a selected crew of farmers receive smartphones to monitor their own farm as well as some other farmers' one. We formulate two variants of the problem of mobile device deployment and task assignment and propose related solutions. We evaluate the proposed approaches through simulations and apply them to a test-bed in Kenya.