Packet Loss Prediction for a Motion-Capture (mocap) System
Mentor:Gustavo Vejarano, Associate Professor Electrical Engineering Department , Loyola Marymount University
Reliable wireless communication is important for performing daily life activities because of the increasing amount of electronics that use this technology. Our goal is to develop an affordable and portable mocap system that uses three types of non-optical sensors: an accelerometer, gyroscope, and magnetometer. Such systems have myriad applications, including to health and sports professions, where individuals such as physical therapists, athletes and prosthetic professionals require a convenient and reliable record of human movement. While testing our mocap system, we found that the sensors were not able to transmit information without interruption, causing erroneous joint angle calculations due to packet loss. Our current investigation is to predict packet loss, which will improve the accuracy of our measurements. Using a token passing protocol, sensors attached to the limbs of a person transmit information to a receiver PC, which reconstructs the limbs’ motion in real time. This allows the sensors to take turns for transmission, making it easier to detect when a sensor was not able to transmit successfully. The inclusion of a packet loss counter and RSSI (received signal strength indicator) in each transmission provides a method of measuring communication performance, enabling us to use algorithmic and mathematical analysis to identify patterns that assist in the prediction of packet loss. Through mocap system testing, we can determine the ability of our algorithms to predict or flag the packet loss. Moreover, to test accuracy, we compare our measurements with those obtained by optical sensors, which are used as a proxy for the true values.