Title | ||
---|---|---|
The Simple Reliable Monitor: A Formalisation of the Concept of a Safe Software Monitor |
Abstract | ||
---|---|---|
The (safety) monitor concept has been proposed as a method of bounding brilliant and creative software with something a little more provable in order to keep one's feet on the ground and planes in the air. In this paper the monitor concept is extended to the simple, reliable (SR) monitor The SR monitor may be seen as an alternative to n-version programming. The nth version of the system is the most sophisticated the n-1 version is functionally similar, but more reliable, etc. At any time, M versions (N, N-1, N-2,,,) of the system are running, with a lower version providing the ultimate outputs only if the postconditions of the higher version fail. More than this, from the point of view of failure of the most sophisticated version, one is running a lesser version which would have the advantage of giving a correct answer, albeit with some performance loss. This would seem a far move philosophically correct way to provide a fallback. By the use of subtype and inheritance concepts, plus pre and post condition refinement, the authors suggest how to set up classes of monitors for problems. One is able to reason about the added complexity of a more sophisticated solution or conversely the ability of a simpler solution to act as a monitor for the more complex solution. A system process byproduct is that, as the classes develop, more accurate estimates of costs and better knowledge of the uncertainties in the development of sophisticated strategies for the problem solution can be obtained.(1) |
Year | Venue | DocType |
---|---|---|
1995 | AUSTRALIAN COMPUTER JOURNAL | Journal |
Volume | Issue | ISSN |
27 | 1 | 0004-8917 |
Citations | PageRank | References |
0 | 0.34 | 0 |
Authors | ||
2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Mark Pfeiffer | 1 | 5 | 2.37 |
John Leaney | 2 | 186 | 25.20 |