Radio-Frequency Identification (RFID) technology has brought tremendous benefits for business processing, especially in incorporating RFID data into Supply Chain Management (SCM). With the explosion of RFID data, there has raised important problem how to mine valuable information from tremendous and potentially infinite volumes of RFID data. Complex Event Processing (CEP) technology used to process RFID data has attracted increasing attention. However, the current researches usually focus on centralized CEP architecture, which requires greater bandwidth and computational capability, and Lacks of robustness and scalability because of single point failure or network break. This paper first proposes a novel distributed CEP architecture, which spreads centralized CEP tasks load over multiple communicating stations by a space-based communication paradigm over a distributed message broker architecture based on Jini network. A distributed complex event detection algorithm based on Master-workers pattern is proposed. The results of experiment show that the performance of our approach is remarkable in the large-scale RFID applications.