Question bank — Safety-critical standards — DO-178C (airborne software), IEC 61508, ISO 26262

False — yeh opposite hain. DAL A aviation ka deadliest level hai (hull loss), jabki ASIL A automotive ka least safety-critical level hai; ASIL D wala deadly hai.

False — yeh ulta hai. SIL 4 demand karta hai $PF{D}_{avg}$ $[10^{-5},10^{-4})$ mein, jo SIL 1 ke $[10^{-2},10^{-1})$ se choti (zyada stringent) failure probability hai.

False — yeh ek process standard hai. Yeh sirf auditable evidence require karta hai ki ek disciplined lifecycle (objectives, independence, coverage) follow kiya gaya; residual bugs phir bhi exist kar sakte hain.

True — ISO 26262 generic parent IEC 61508 ka automotive adaptation hai, jo "zyada danger ⇒ zyada rigor" logic reuse karta hai lekin quantitative SIL ki jagah $S,E,C$-derived ASIL use karta hai.

True — yeh dikhane ke liye ki har condition independently outcome flip karti hai, tumhe poori decision ko TRUE aur FALSE dono exercise karna padta hai, isliye decision coverage woh subset hai jo MC/DC force karta hai.

False — QM ka matlab hai koi safety-specific mandate nahi, lekin normal quality-management engineering practices phir bhi apply hoti hain; yeh "no safety integrity requirement" hai, "no engineering" nahi.

True — kyunki $PF{D}_{avg}={\lambda }_{DU}{T}_1/2$ $T_1$ mein linear hai, interval ko aadha karne se average failure probability bhi aadhi ho jaati hai.

False — DAL A par bhi sirf objectives ka ek subset independence require karta hai; objectives ki sankhya aur jinhe independence chahiye dono criticality ke saath badhte hain, lekin sab nahi.

Error count aur criterion mein hai. MC/DC ko har condition ko independently outcome flip karna require karta hai jabki doosre fixed rahen; do conditions ke liye yeh $n+1=3$ tests hote hain, 4 nahi, aur sirf "har value dekha" condition coverage hai, MC/DC nahi.

Error band edge mein hai. SIL 3 hai $10^{-4}\le PFD<10^{-3}$; $2\times 10^{-3}$ ki value $10^{-3}$ se upar hai aur isliye SIL 2 mein aati hai, SIL 3 mein nahi.

Error mapping mein hai. MC/DC DAL A ka coverage criterion hai; DAL B ko decision coverage chahiye aur DAL C ko sirf statement coverage — teeno mein se sabse sasta.

Error derivation methods ko confuse karne mein hai. ASIL har hazardous scenario ke liye Severity, Exposure, aur Controllability se compute hota hai; yeh koi fixed attribute nahi jo tum kisi part ke liye lookup karo.

Error ASIL decomposition ko ignore karta hai. D requirement ko do sufficiently independent elements mein split kiya ja sakta hai (jaise B(D)+B(D) ya A+C), har ek lower rigor par develop kiya gaya, agar koi common-cause failure unhe link na kare.

Error ise arbitrary treat karne mein hai. $1/2$ test interval par linearly-growing failure probability ko average karne se aata hai — 0 se ${\lambda }_{DU}{T}_1$ tak badhne wali line ka mean uska midpoint hota hai.

Error domains ke across rigor compare karne mein hai. Yeh standards alag consequences ke liye alag scales aur authorities use karte hain; "mass-produced" certification model change karta hai (self-cert + audit vs per-aircraft), koi universal strictness ranking nahi.

Kyunki verification cost rigor ke saath roughly exponentially badhti hai, isliye expensive effort sirf wahan lagao jahan failure consequences severe hain — safety ka ek 80/20, jo Functional Safety mein bhi dikh ta hai.

Decision coverage tab bhi pass ho sakti hai jab poora outcome toggle ho jabki kisi condition ka individual effect kabhi isolate na ho; MC/DC force karta hai ki har condition demonstrably result control kare, compound Booleans ke andar chhupe faults expose karte hue — dekho Code Coverage Metrics.

Ek rarely-acting safety function (jaise emergency shutdown) characterise hoti hai is baat se ki woh jab bulaayi jaaye tab kaam kare, isliye probability-per-demand fit hoti hai; ek continuously-relied-on function ko operation ke har ghante mein dangerous failures se better measure kiya jaata hai.

Yeh kisi bhi electrical/programmable safety system ke liye domain-independently likha gaya tha, aur automotive engineers ne ise ISO 26262 mein specialise kiya rather than underlying functional-safety philosophy ko reinvent karne ke bajaye.

Independence self-review ke blind spots remove karta hai — jo insaan bug likhta hai woh usually same galat mental model se re-test karta hai, isliye ek alag reviewer common-cause reasoning errors todta hai, jo Fault Tolerance & Redundancy se redundancy ka idea hai.

Risk sirf iti nahi hai ki outcome kitna bura hai; ek rare situation ($E$) ya jo driver easily control kar sake ($C$) real-world danger kam karta hai chahe severity ($S$) high ho — teeno ko combine karna, FMEA & Hazard Analysis ki tarah, harm ke true exposure ka estimate karta hai.

QM — agar koi injury ho hi nahi sakti, toh integrate karne ke liye koi safety hazard nahi hai, isliye situation kitni bhi frequent ya uncontrollable ho koi safety integrity level assign nahi hota.

Yeh $0$ ki taraf approach karta hai, kyunki $PF{D}_{avg}={\lambda }_{DU}{T}_1/2\to 0$; constant re-testing ka matlab hai ek dangerous undetected fault essentially kabhi latent nahi — ek idealised limit, practice mein unreachable.

Jab ${\lambda }_{DU}t$ ab small nahi raha (lambe intervals ya high failure rates), true $1-e^{-\lambda t}$ straight line se neeche curve karta hai, isliye simplified formula $PFD$ ko overestimate karta hai aur linear model replace karna padta hai — linked to Failure Rate & Reliability.

Safety ke liye kuch mandated nahi — DAL E ka matlab hai "safety par koi effect nahi" agar yeh fail ho, isliye koi structural coverage objective apply nahi hota; yeh sirf normal quality expectations ke hisaab se engineer kiya jaata hai.

$PFD=10^{-4}$ (SIL 3/SIL 4 edge) par RRF exactly $10^4=10,000$ hai; isse zyada reduction chahiye woh SIL 4 territory mein jaata hai.

Nahi — shared power supply ek common-cause failure hai jo dono elements ko ek saath down kar sakti hai, independence precondition violate karte hue, isliye lower rigor par split credible nahi hai.