This paper describes a smart-card based solution that further extends the concept of static access control list by performing real-time analysis over user's access patterns, thus allowing early detection of abnormal card usages. A list of irregular behaviors are identified and classified to formulate a logical model that quantifies abnormal usage behaviors. Based on this model, a real-time suspicious action detection prototype is developed, allowing rapid alert and reaction towards irregular behaviors. Finally, example scenarios on some applications on electronic commerce are discussed.