A Correct by Construction Fault Tolerant Voter for Input Selection of a Control System
Arif Ali AP, Jasine Babu, Deepa Sara John · 2025
Read A Correct by Construction Fault Tolerant Voter for Input Selection of a Control System by Arif Ali AP, Jasine Babu, Deepa Sara John (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-22236.txt.
Introduction
Safety-critical control systems, like those in aircraft, read several redundant sensors and must pick which inputs to trust. Arif Ali AP, Jasine Babu, and Deepa Sara John design the voting unit for that job in this 2025 paper. A voter selects a reliable input from the redundant sources, guided by rules that detect and isolate faulty units. Their work covers the full path: formal requirements, design, verification, and synthesis of a generic voter for a measurement system with any number of redundant channels.
The approach is correct by construction, carried out with the Rocq theorem prover. Instead of testing a finished design for errors, the team builds the design inside a proof environment, so the verified properties hold by the way it was made. The voter is generic rather than tied to one aircraft system, since the redundancy count is a parameter.
Catalog record
| CREATOR | Arif Ali AP, Jasine Babu, Deepa Sara John |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 71 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-22236
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A Correct by Construction Fault Tolerant Voter for Input Selection of a Control System Arif Ali AP, Jasine Babu, Deepa Sara John Abstract Safety-critical systems use redundant input units to improve their reliability and fault tolerance. A voting logic is then used to select a reliable input from the redundant sources. A fault detection and isolation rules help in selecting input units that can participate in voting. This work deals with the formal requirement formulation, design, verification and synthesis of a generic voting unit for an N N -modular redundant measurement system used for control applications in avionics systems. The work follows a correct-by-construction approach, using the Rocq theorem prover. Department of Computer Science and Engineering, Indian Institute of Technology Palakkad, [email protected] Department of Computer Science and Engineering, Indian Institute of Technology Palakkad, [email protected] ISRO Inertial Systems Unit, Indian Space Research Organization, Kerala, [email protected] The authors acknowledge Saarang S and Mis Ab VP, former undergraduate students of IIT Palakkad, for their involvement in preliminary experiments with the Coq implementation of the voter, and Sandra S, visiting student at IIT Palakkad, for testing support. The authors also thank Murali Krishnan, Professor, National Institute of Technology Calicut, for his useful feedback on the initial draft. ccs Theory of computation -¿ Logic -¿ Logic and verification, Theory of computation -¿ Logic -¿ Automated reasoning keywords Fault Tolerant System Design, Formal Verification, Correct by Construction, Input Selection, Interactive Theorem Proving † † runningauthor: A. Ali AP and J. Babu and D. S. John † † copyright: Arif Ali AP and Jasine Babu and Deepa Sara John † † funding: A part of this work is funded by Indian Space Research Organisation (ISRO) RESPOND project RES-IISU-2022-013 titled ‘Formal Analysis and Verification of Redundancy Management Logic for Navigation Processor used in Man-rated Launch Vehicle’. 1 Introduction Redundancy management is used in safety-critical systems to improve their reliability and fault tolerance [ 9 , 10 , 6 ] . Examples of some well known domains of safety critical applications include avionics control, automated vehicles and nuclear reactor control systems [ 15 , 14 ] . Since the reliability guarantee provided by testing is insufficient, such systems are suitable candidates for formal verification [ 4 ] . From the late 1980s, usage of formal verification for the design of fault-tolerant systems [ 1 ] was supported by agencies such as NASA, which work in the domain of safety-critical applications [ 31 , 9 , 28 ] . Owre et al. [ 23 ] mention the formal verification of fault-tolerant architectures as a major motivation for the development of the PVS theorem prover. Interactive consistency and convergence [ 19 ] , clock synchronization [ 29 , 21 ] , and redundancy management [ 9 ] are some well-studied subproblems of fault-tolerant system design for which formal verification has been attempted. There are also works on the verification of timing specifications of fault-tolerant systems [ 27 , 30 , 13 , 32 ] . Most of the works on formal verification of fault tolerant systems [ 19 , 29 , 5 , 24 , 16 , 17 ] discuss only verification of the formal model and not the code synthesis. There is a recent work by Rahli et al. [ 26 ] that discusses Rocq based synthesis of a code implementing a Byzantine fault tolerant state machine replication protocol based on message passing. Our work deals with the formal modelling, verification and construction of an input selection unit for a synchronous N -modular redundant measurement system used for control applications in avionics systems. Though the redundant input units work synchronously and measure the same physical quantity, there can be differences in the measured value due to the noises in the measurement system (see Section 2.2 ). Moreover, some of these input units may have transient faults due to factors such as cosmic radiations [ 22 ] and vibrations [ 11 ] , or permanent faults due to hardware failures [ 7 ] . Hence, it becomes necessary to apply an input selection logic to the values obtained from the redundant input units and produce a single output value corresponding to the physical quantity. In this work, we follow a correct-by-construction approach for synthesis of the input selection unit which uses a noise-resilient voting algorithm for the selection. Henceforth, we refer to the input selection unit as the voter unit . The interesting application scenarios are when the system is running continuously, where in every cycle of operation, each input unit produces a measurement. In many real scenarios, the incremental change in the physical quantity being measured (such as distance to an object, angular velocity, atmospheric pressure, etc. measured in an avionics control unit) from one cycle to the next is small and smooth. In each cycle, the voter unit has to select one of the input units as a prime unit . The selection should ensure a reliable output value by discarding the measurements having transient errors. In addition, permanently faulty units should be identified and isolated, so that they do not corrupt the feed to the controller. To do this effectively, the input selection of each cycle has to also take into account some information from the past cycles. If the measurement domain is \mathcal{D} , the number of input units is N and the information based on the past t cycles is used for the calculations, then the voting unit essentially calculates a function f:\mathcal{D}^{tN}\mapsto\mathcal{D} in every cycle. In a more general situation, some self-reported parameters from the input units are also required for the decision making. It is also desired that, along with the selected measurement value, some reliability parameters of the output are also supplied to the controller. The reader may refer to Figure 1 which depicts an illustration of such a system. The formal verification of such systems becomes challenging due to the large size of the underlying state space [ 4 , Section 1.3.1] . Data Data Input Unit 1 Input Unit N Input Selection Unit Controller (unit status info, measured value) (unit status info, measured value) (output value, reliability parameters) \dots \dots preserved data Figure 1: In each cycle, the input selection unit has to select an appropriate measured value and feed it to the controller, along with some parameters indicating the reliability of the output. Between consecutive cycles, the change in the physical quantity being measured is assumed to be small. Preserved data stores some information from past cycles and it is used for fault handling and output generation. The output value could be retained from the previous cycle, when the current cycle measurement is not reliable. One aspect of transient resilience of the system is to ensure that measurements with transient errors are discarded, without getting fed to the controller. On the other hand, measurements are usually taken from redundant units that are physically separated and there could be minor deviations in the measurements made by them. Hence, it is desirable to avoid frequent switching of the prime unit due to transient failures, to maintain better stability. Although these two aspects of transient resilience may seem conflicting, the assumption that the incremental variations in the measured quantity between consecutive cycles is