Exploring Different Architectures to Support Crop Farmers with a Mobile Application on Pesticide Control

Abstract

The MobiCrop app, which is a distributed mobile application has been proposed to aid crop farmers with timely decision making on the applicability of pesticides (i.e., which pesticide to apply, when, where, and how to apply them). Due to the vast amount of pesticide and crop data, the application is designed following the three-tier architecture technique which comprises the mobile devices, a cloud-hosted middleware, and cloud-based database. The idea is to enable the mobile device to retrieve the needed pesticide data from the back-end and when necessary, part of the data can be stored on the mobile through caching for offline accessibility. However, constantly updating the mobile cache through data polling is costly for the wireless bandwidth and energy usage on the mobile. Also, it is difficult to update the stale cache data when there is no wireless connectivity. Hence, this work explores three architectural designs of the MobiCrop app which arethe: 1) the standalone (network independent), 2) distributed architecture through data offloading, and 3) distributed architecture through data
partitioning.