Algorithms for collective motion of smart dust
As an example, we have considered the collective movement of motes towards a target located in a portion of the Martian surface that extends over a range of several kilometers. The motes move in wind-driven hops by a process called saltation. It involves the random lofting of a grain into the ambient wind flow where it becomes entrained, later falling to ground: whereupon the process repeats.
Sand (and by implication smart dust) may also become electrically charged in a Martian environment. The simulation of this problem has features that correspond to classical studies of electron transport/trapping in semiconductor devices near rough interfaces.
We use Monte Carlo simulation of the mote dynamics coupled to a wind profile computed from the Navier- Stokes equations: for fixed drag coefficients the results are standard. We have computed 2D and 3D transport for smart dust motes that alter their drag coefficients, for example, by a shape change, when a favourable entrainment to the turbulent wind moves the motes towards the target. The induced correlation of the motion with the target relies on inertial effects (transient response) and exploits time-dependent fluctuations in the wind pattern. At low Reynolds numbers (which would apply to NEMS motes of sizes in the 10 micron range acting in a small region) the motion corresponds to adaptive diffusion or adaptive Brownian motion. At the high Reynolds numbers considered here, the wind flow is turbulent with speeds of the order of 3 – 20 m s-1. Collective motion into a star shape is illustrated below in frames from a Monte Carlo study.
The results indicate that switchable drag coefficients provide a mechanism for self-control of mote motion. The use of sensory information produces self-organising or anti- thermodynamic behaviour. Collective effects are possible, for example, if the target represents the centre of mass of the swarm. The latter may be sensed from the mean signal received periodically from the host swarm. In our experimental programme the mote surface profile is altered by using low-power electrical stimulation of an outer electro-active polymer layer.
We have devised wireless networking models for smart dust swarms that lead to a mode in which the swarm acts a phased-array of wireless emitters designed to swing a strong communications signal towards an orbiting mothership.