Comparing Local and Remote Sensor Data Processing for Shape Detection Systems

Low power systems equipped with sensors that recognize users and objects are increasingly used today. These devices need to take actions according to the information resulting from the processed data. Sensed data processing is an intensive task and often needs more computational power than that available on the system connected to the sensor. This paper analyzes the opportunity of running the computational intensive tasks remotely in a remote High Performance Computing (HPC) center compared to processing data directly on low power sensor platform and still providing a real time response. The platform used is Raspberry Pi and the sensor is the Raspberry Pi Camera Module. Various connectivity methods to the HPC center will be tested such as wired, Wi-Fi and cellular HSDPA.