Node Mobility

If nodes change their location over time, they have to update their location estimates frequently in order to avoid inaccuracies resulting from using outdated location estimates. Moreover, node movement during the measurement of parameters needed for location computation can cause inaccuracies in the estimated location.

The time $t_\mathrm{update}$ between successive location updates usually equals the time $t_\mathrm{turn}$ required for one rotation of the lighthouse. Thus, the update frequency $1/t_\mathrm{update}$ can be increased by decreasing $t_\mathrm{turn}$. However, there is an easy way to double the update frequency when using rotating mirrors for beam generation, because the beams are reflected to both sides of the lighthouse as depicted by the dashed laser beams in Figure 5. Thus, we actually have two ``virtual'' wide beams we can use for location estimation, effectively doubling the update frequency.

If a node moves during measurement of $t_\mathrm{beam}$ (i.e., after detection of the first beam and before detection of the second beam), the obtained value of $t_\mathrm{beam}$ will be incorrect. Additional errors are caused by the node moving between measurements of $t_\mathrm{beam}$ of the three lighthouses.

There are two ways to detect and reject faulty location estimates resulting from node movement during measurement. The first compares two or more consecutive position estimates and rejects them if they differ by more than a small threshold. The second approach uses accelerometers to detect movement during measurement. Accelerometers can also be used to estimate node movement (velocity, direction) during measurements of $t_\mathrm{beam}$. The obtained values can be used to correct $t_\mathrm{beam}$, such that correct location estimates can also be obtained during node movement. In fact, the Smart Dust prototypes developed at Berkeley already contain such sensors.