Beam Generation

In order to keep the hardware and energy overhead on the Smart Dust nodes small, the beam must be easily detectable. Furthermore, the system should work with high accuracy even if the base station is far away (tens of meters, say) from the nodes. Therefore we decided to use a laser-based approach. As mentioned above, the beam should be as wide as possible in order to keep inaccuracies small. In order to achieve this, we use two lasers to create the outline of a parallel beam as depicted in the upper half of Figure 4. This makes no difference to a single wide beam, since we are only interested in the edges of the beam (i.e., change from ``dark'' to ``light'' and vice versa) in order to measure $t_\mathrm{beam}$ and $t_\mathrm{turn}$.

Due to the narrow laser beams, the ``virtual'' parallel beam generated this way can only be seen from a single plane, however. In order to ensure that the beam can be seen from any point in the northern hemisphere of the lighthouse without defocusing the lasers, the laser beams have to scan this space in some way. The lower half of Figure 4 depicts two ways to achieve this. The first approach uses a small mirror mounted on a rotating axle under an angle of $45^\circ$. By pointing the laser at this mirror, the reflected rotating beam describes a plane. With commercial off the shelf technology we can easily achieve a rotation frequency of about 300Hz. The second approach uses a small deflectable MEMS mirror similar to the one used as part of the corner cube retroreflector (CCR). The MEMS mirror presented in [7], for example, operates at 35kHz and achieves a deflection angle of $25^\circ$. A laser beam pointed at such a mirror can thus sweep over an angle of $50^\circ$ at a frequency of 35kHz.

Based on this approach, a lighthouse consists of a (slowly) rotating platform, on which two semiconductor laser modules and two rotating (or deflectable) mirrors are mounted. However, as mentioned at the beginning of Section 4.2, it is next to impossible to assemble all the pieces such that the resulting ``virtual'' wide beam is almost parallel. Therefore, we have to come up with a model which describes an imperfect but realistic system. The model discussed below is based on rotating mirrors, since we used this approach in our prototype implementation of the system. However, the model equally applies to a system based on deflectable mirrors.