System Reliability Analysis and Design

Reliability Analysis

Sfta (Software Fault Tree Analysis) software fault tree analysis.

Software Fault Tree Analysis (sfta) is a top-down software reliability analysis method ), in particular, the incidents that have a significant impact on the safety and reliability of personnel and equipment begin to trace down the causes of the incidents until the basic incidents (bottom incidents ), to determine the possible combination modes and/or probability of occurrence of software faults.

Cut Set: it is the set of basic events that cause top events. The minimum cutover set is the minimum cutover set required to cause top events.

Boolean algebra is used to obtain the minimum cut set and the shortest path set for computer software.

Fault Tree:

T = A1 + A2
= X1 X2 A3 + X4 A4
= X1 X2 (X1 + X3) + X4 (X5 + X6)
= X1 X2 A1 + X1 X2 A3 + X4 X5 + X4 X6
= X1 X2 + X4 X5 + X4 X6
Therefore, the minimum cut set is {x1, x2}, {X4, X5}, {X4, X6 }. The result shows the union of the three intersections. These three intersections are the three smallest cut sets e1 = {x1, x2}, E2 = {X4, X5}, E3 = {X4, x6 }. Equivalent diagram of the fault tree with the minimal cut set

Reference: http://www.safe001.com/anping/ap_zhishi/8_fenxi.htm

Reliability Design

• 
  
• Fault Tolerance Design Technology: Common Software Fault Tolerance technologies include:Fast Recovery Design,Program Design for version nAndRedundancy Design.
  1.Fast Recovery DesignContains several blocks with the same functions and different designs. Each block is instantly running. Once a program fails, it is replaced by a backup block.
  
  • The Block recovery method is suitable for scenarios where only one computer is available. • Redundancy is allowed only for complex and fault-prone program segments. • The design of basic blocks should be as different as possible from that of replacement blocks;

  • The acceptance test function is very important. deviations between program execution results and expected results should be detected, or outputs that can trigger security incidents should be detected and prevented.
  2.N Version DesignThe core is to design multiple modules or different versions, perform a majority vote on the operation results with the same initial conditions and the same input (to prevent a software module/version from providing incorrect services for fault tolerance ).
  3.Redundancy DesignThe idea comes from the hardware system, but it is different. The Software Redundancy Design Technology uses a variety of different paths, modules or systems with different algorithms or implementation methods as backups, replacing them in case of a fault, and maintaining the normal operation of the system.

• Error Detection Technology: When there is no need for online fault tolerance or the redundant design technology is not used, but there is a high reliability requirement, the detection design is generally used, in the case of software failure can be promptly detected and alarm, the obvious disadvantage is that the fault cannot be automatically solved.
  • Passive Error Detection-set Detection Points in several parts of the program to wait for the emergence of error signs
  

  Dog design:The watchdog technique is an effective method for controlling the running time. The watchdog is actually a timing device. When the timer is started, the watchdog is interrupted when the accumulated time reaches the specified value (that is, the dog's name ).

  • Proactive Error Detection-actively checks program status
• Error avoidance design (design to reduce complexity): Software complexity is closely related to software reliability. Software complexity is an important source of Software defects. The idea of Complexity Reduction design is to simplify the software structure on the basis of ensuring the implementation of software functions.