Sensors generate large amounts of spatiotemporal data that\nhave to be stored and analyzed. However, spatiotemporal\ndata still lack the equivalent of a DBMS that would allow\ntheir declarative analysis. We argue that the reason for\nthis is that DBMSs have been built with the assumption\nthat the stored data are the ground truth. This is not the\ncase with sensor measurements, which are merely incomplete\nand inaccurate samples of the ground truth. Based on\nthis observation, we present Plato; an extensible DBMS for\nspatiotemporal sensor data that leverages signal processing\nalgorithms to infer from the measurements the underlying\nground truth in the form of statistical models. These models\nare then used to answer queries over the data. By operating\non the model instead of the raw data, Plato achieves significant\ndata compression and corresponding query processing\nspeedup. Moreover, by employing models that separate the\nsignal from the noise, Plato produces query results of higher\nquality than even the original measurements.
Loading....