Extremely Low-Power Edge Connected Devices

Information and data gathering are closely linked. Many systems are based on gathering sensor data from multiple nodes and processing centrally (cloud based). Since accessibility to increasing amounts of data leads to better decisions, this puts an increasing processing pressure on the cloud. Coupled with better and more capable sensors which are now available, data gathering has grown and accelerated into almost all applications. While it is a clear expectation that this increased amount of data will improve outcomes, it is also clear that the increasing rate of sensor data must be processed at the same rate. The computation of local data remotely creates a bottleneck to the cloud resulting in long latency. Decisions may arrive back too late to determine the best course of action. Furthermore, remote servers must share their resources according to a strategy that may not be beneficial to the critical task being controlled. Untimely computation breakdown may make critical computations difficult or completely unavailable if out-of-range highlighting the need to provide computing intelligence and decision making on the edge. Recently, edge processing has been proposed and may be the only reasonable answer. With sufficient computing capability it can provide decisions with local data quickly bypassing the latency of a cloud connection. Even in the larger context where cloud computing is required, local computation preprocesses the data resulting in better utilization of the edge-cloud transmission link. As an illustration of this type of capability, an asset tracking demonstration with real hardware was generated at ON Semiconductor. This tracking system utilizes Bluetooth tag transmitters on each asset and multiple receiving antennas connected in a network detecting multiple angle-of-arrival (AoA) from each tag. The demonstrator system determines the tag location from these measurements.