Wireless Sensor Gathering Data during Long Time

Recently, one of promising markets utilizing wireless sensors is disaster protection SoC. Sensors are located at some positions and gather environmental data periodically. The volume of data accumulated during long time can be valuable but the limit of local storage should also be considered. From the observation, we devise a local storage management algorithm in wireless sensor nodes. Nodes just store gathered data locally, not deliver them to the base node. At first, local storage is divided into two parts: Correct Data(CD) and Differential Data(DD). CD is used to store data in normal form. DD is an array of N bits object data and each row can contain several bit values - differential data. That is, if gathered data is almost similar to one in the previous period, the differential value is calculated and stored into DD. And the current and the previous data are connected via index field in CD. In addition, a query processing algorithm is also proposed based on local data. In this way, storage efficiency can be obtained while data accuracy is preserved. Via experimental works, we come to know only 40-50% storage area is needed compared with the case that only gathered data is stored.