Abstract
This paper extracts the lessons learned in an on going project which aims at delivering agricultural intelligence to build a capable decision support system that improves agricultural practice and decision making in sub-Saharan Africa. By generalizing the research challenges experienced so far, this paper provides a framework for designing Wireless Sensor Networks for agricultural monitoring in developing regions, taking into account all the particularities of such environments. In this framework, new solutions and research ideas are proposed for sensor network design to address the special challenges of the developing world. These include a zone-based joint topology control and power scheduling mechanism, a multi-sink architecture with associated back-link/storage aware complementary routing and a parameter-energy-environment aware task scheduling approach.