Sliding-Window Moving Average Learning for System with Lossy Packets

Wireless technologies tend to become a core element of acquisition and sensory system and system identification process represents an important tool in many practical engineering applications. The current trend is to integrate this lossy network technologies and system identification together by having an identification element (identifier) that is able to give a good description for a system underlying dynamic when the system observations input/output data) are sent wirelessly (with lost packets). The lossy network normally sent input-output data (packets) with irregular sample periods thus introduces challenges in system identification process. This paper investigates the possibility of performing system identification with irregular sample time using first order linear system with sliding window moving average techniques. By adopting data store management approach on sliding window of data the recursive identification methods proposed are able to map the black-box system with irregular stream of sample with good level of accuracy.