Software Fault Tolerance: An Aspect Oriented Approach

摘要

Software fault tolerance demands additional tasks like error detection and recovery through executable assertions, exception handling, diversity and redundancy based mechanisms. These mechanisms do not come for free; rather they introduce additional complexity to the core functionality. This paper presents light weight error detection and recovery mechanisms based on the rate of change in signal or data values. Maximum instantaneous and mean rates are used as plausibility checks to detect erroneous states and recover. These plausibility checks are exercised in a novel aspect oriented software fault tolerant design framework that reduces the additional logical complexity. A Lego NXT Robot based case study has been completed to demonstrate the effectiveness of the proposed design framework.