Energy efficient scheduling in IoT networks

The Internet of Things (IoT) is poised to be one of the most disruptive technologies over the next decade. It is speculated, that we shall have billions of devices with communication capabilities very soon. Minimizing energy consumption is one of the most important problems in such IoT networks mainly because IoT nodes are distributed in the field with limited, unreliable, and intermittent sources of power. Even though the area of reducing power for stand-alone machines is very rich, there are very few references in the area of co-operative power minimization in a system with many IoT nodes. We propose two algorithms in this paper, which are at the two ends of the spectrum: Local exchanges information between neighboring nodes, and Global uses a global server that has recent snapshots of the global state of the network. We show that both these algorithms reduce energy consumption by roughly 40% for settings that use data from real life IoT deployments (data from Barcelona city). We further show that if deadlines are tight, Local is preferable for smaller networks, and Global is preferable for larger networks. When deadlines are loose, Global is preferable if we need to follow hard real time semantics, otherwise Local is preferable.