A down-to-earth case for autonomy in massively plural sensor systems

The role of autonomy in wireless sensor networks remains something of a controversial issue. Although most of the early work in the field, and the subsequent research agenda, has implicitly assumed the use of autonomous sensor nodes, few if any of the practical implementations that have followed have featured autonomy at node level. Systems deployed so far have been relatively small, and their operation posed few problems which require autonomy. At this scale, “planned”, non-autonomous systems are seen to be highly successful, so much so that the largest operation WSN so far achieved was designed using similar principles. In this paper, the authors look forward to larger networks, with 100,000 or more nodes, and it is argued that for these networks it is unlikely that “planned” architectures will be feasible. Rather, a return to node-level autonomy as the basis for self-configuring, selfmaintaining and self-optimising systems is proposed. The paper concludes with a review of some of the work being undertaken towards this end.