An Architectural Overview of a Software Supervisor

T. Savor and R.E. Seviora ([email protected], [email protected]) Bell Canada Software Reliability Laboratory University of Waterloo Waterloo, Ontario, Canada N2L 3G1 Abstract Speci cation-based software supervision is an approach to the automatic run-time detection of failures. The supervisor is a unit that monitors the inputs and outputs of a target software system and reports discrepancies between speci ed and observed behaviors as failures. Two fundamental issues arise in supervision. The rst is due to speci cation non-determinisms. The major di culty being that the supervisor is required to consider all legitimate behavioral alternatives to avoid erroneous failure reports. In some cases, this results in signi cant supervisor time and space complexities. The second issue deals with continuation of supervision after a failure was detected. The major problem is that after a failure is detected, the supervisor no longer has de nite knowledge about the speci cation state of the target system which can result in in spurious failure reports. This paper describes a software supervisor architecture for target systems whose speci cations are expressed in communicating extended nite state machine-based formalisms. The architecture includes several independent mechanisms, each tailored to resolving a speci c ambiguity occurring within the software supervisor. Included are mechanisms for resolving speci cation non-determinism and identifying the state of the target system after a failure is detected.