Dealing with blocking in supervisory control of discrete event systems

A study is made of the supervisory control of discrete-event systems where blocking is unavoidable or nonblocking solutions do not yield good performance. The authors consider strategies to improve the performance of a given blocking supervisor. Performance is characterized in terms of a satisfying measure (how much of the legal marked behavior is achieved) and a blocking measure (how often the execution of legal unmarked strings is blocked). Techniques are presented for improving each of these conflicting measures.<<ETX>>