UPES-ECS Risk Register & Known Gaps¶
An honest list of what can make UPES-ECS fail, weaken it, or block go-live β with severity, impact, and a recommended fix. Some items require a university decision (marked β), not just engineering. Review this at every phase gate.
Severity: π΄ Critical (can defeat the system's purpose) Β· π High Β· π‘ Medium Β· βͺ Low
Decisions applied β 2026-07-04¶
| Risk | Decision | Result |
|---|---|---|
| R1 Power | System deploys via a self-powered disaster-response van + corner repeaters (doc 23) | β Resolved for field mode (van power autonomy). UPS still recommended for fixed campus mode. |
| R2 Mobile inbound | Battery best-practices documented (doc 24); fixed phones for critical answer points | β Mitigated. Future app solves fully. |
| R3 After-hours | "Whoever is available"; staffing handled off-system | β Out of scope / accepted by university. |
| R4 Surge | Surge / Declared Incident posture added to ERT SOP Part J | β Documented. |
| R5 Location | Out of scope now; future UPES VoIP app broadcasts coordinates to the aggregated workflow | π΅ Deferred with a plan. |
| R6 Silent calls | Keep on 111 + silent-call protocol (ERT SOP Part I); future app panic button | β Decided. No separate line. |
| R7 Cleartext media | Future security enhancement β not a current concern | π΅ Deferred. |
| R8 Single PBX | Fine for now; van doubles as mobile failover (doc 23 Β§7) | β Mitigated + plan noted. |
The entries below keep the full detail; the table above is the current status.
A. Mission-level risks¶
R1 π΄ β Power dependency contradicts the mission¶
- Gap: Spec assumes power is available; the system's whole value is working when networks fail β but disasters often cut power too. No UPS in Phase 1.
- Impact: PBX / switch / router / APs lose power β 111 is dead exactly during a real disaster.
- Fix: Promote UPS/battery backup for core infra (server, switch, router, key APs) from "later" to Phase 1. Size for the realistic outage window; monitor UPS as a critical device. Add PoE + UPS for the APs covering ERT/critical zones.
- Owner: IT + university (budget). Decision needed before go-live.
R2 π΄ Mobile inbound (ERT callback) unreliable¶
- Gap: LAN-only forbids push notifications, so locked iOS/Android phones often drop the SIP registration and won't ring on an ERT callback. The Missed-Call Recovery SOP depends on callback.
- Impact: ERT calls a missed caller back β phone never rings β critical follow-up lost.
- Fix: Prefer fixed phones as callback/answer targets; tune PJSIP qualify/keepalive; train users on background/battery settings; document mobile inbound as best-effort, outbound (dial 111) as reliable. Consider a small on-device keepalive.
- Owner: IT + ERT Lead.
R3 π΄ β After-hours / zero-staff coverage undefined¶
- Gap: "Same path day and night" + "min 2 responders," but no on-call/night model. Empty queue at night β voicemail with nobody watching.
- Impact: A 3 AM emergency escalates to an unwatched voicemail.
- Fix: Define a 24/7 on-call rota; night ERT phone(s) always registered; queue-zero raises an alert that someone is actually on-call to see; document the night dispatch chain.
- Owner: ERT Lead + university. Decision needed.
R4 π΄ Mass-casualty surge not modeled¶
- Gap: One queue + few operators + ring-all can't absorb dozens of simultaneous 111 calls in a real event.
- Impact: System saturates; real reports blocked; callers dumped to voicemail en masse.
- Fix: Define a "Declared Incident" surge posture β paging-first to push instructions, pull in extra operators, overflow callers to a triage voicemail with location capture, dedup same-event reports. Capacity-test toward realistic surge, not just 2β5 calls.
- Owner: ERT Lead + IT.
R5 π Location blindness for mobile callers¶
- Gap: SAP ID identifies the person, not their physical location. Panicked/injured/lost callers can't always state it.
- Impact: ERT can't dispatch to the right place fast.
- Fix: Put "where are you" first in the answer script (already in SOP β reinforce); rely on fixed phones for location-anchored calls; explore AP-association β coarse zone mapping and a caller location field later.
- Owner: ERT Lead + IT.
R6 π β Voice-only excludes silent / unable-to-speak emergencies¶
- Gap: No path for someone who can't talk (active threat, injury, speech/hearing impairment).
- Impact: The people in the most dangerous situations can't signal.
- Fix: Define a silent-call protocol (dial 111, stay silent β still recorded, located by extension, treated as live/critical, ERT trained to respond to silence); plan a future text/panic-button channel and accessibility support.
- Owner: ERT Lead + university (policy).
R7 π Cleartext SIP/RTP on shared campus Wi-Fi¶
- Gap: Plain UDP media is sniffable; inconsistent with "recordings encrypted at rest."
- Impact: Emergency call audio interceptable on the LAN.
- Fix: TLS + SRTP at least for ERT/fixed devices; evaluate for all clients (Linphone supports it). Weigh against setup complexity.
- Owner: IT.
R8 π Single PBX = single point of failure¶
- Gap: One server; restore target 1 hour = up to an hour with no emergency line.
- Impact: Hardware/disk/corruption outage removes 111 entirely.
- Fix: Keep a pre-imaged spare/warm-standby server ready to swap; later, a second PBX (see Multi-Campus Model B). Test the swap.
- Owner: IT.
B. Security & compliance¶
R9 π SIP registration abuse / DoS not hardened¶
- Fix: fail2ban + registration rate-limiting; disable unused Asterisk modules; concrete LAN firewall rules; lock FreePBX GUI to the management subnet; protect the 111 queue from flooding (per-account call limits, prank-call rate tracking).
- Owner: IT.
R10 π β Data-protection compliance unnamed¶
- Gap: SAP-ID-linked audio + identity = sensitive personal data; docs defer to "university policy" without naming a regime.
- Fix: Map recording/retention/access to India DPDP Act 2023 (and any UGC/university rules): lawful basis, notice, access logging, breach handling, retention limits. Get legal sign-off.
- Owner: University + IT. Decision needed.
R11 π‘ Retention automation + recording verification¶
- Fix: Cron-based 90-day auto-delete with an audit log and preservation flag; periodically verify recordings are audible, not merely present.
- Owner: IT.
C. Operational & organizational¶
R12 π Bus factor / ownership¶
- Gap: If the one admin who built FreePBX leaves, nobody can run it.
- Fix: Train a backup admin; keep runbooks current; define a RACI (who owns config, roster, approvals, backups, health).
- Owner: IT + university.
R13 π‘ No cost / bill of materials¶
- Fix: Produce a BOM + budget: IP phones, APs, PoE switch, UPS units, spare server, cabling, headsets. FreePBX is free; hardware is not.
- Owner: IT + university.
R14 π‘ Prank / false-alarm culture on a campus¶
- Fix: Track repeat false callers; SOP for warnings/escalation; keep 111 always reachable but log abuse; account suspension after review (already in Role Matrix β add a "cry wolf" tracking note).
- Owner: ERT Lead + IT.
R15 βͺ Documentation aids missing¶
- Fix: Glossary (done); one-page ERT desk quick-card and student quick-card; a RACI table.
- Owner: IT (docs).
D. Risk summary table¶
| ID | Risk | Sev | Status | Fix owner |
|---|---|---|---|---|
| R1 | Power/UPS dependency | π΄ | β Resolved (van) | IT + university |
| R2 | Mobile inbound callback | π΄ | β Mitigated (doc 24) | IT + ERT |
| R3 | After-hours coverage | π΄ | β Accepted/out of scope | ERT + university |
| R4 | Mass-casualty surge | π΄ | β Documented (SOP Part J) | ERT + IT |
| R5 | Location blindness | π | π΅ Deferred (future app) | ERT + IT |
| R6 | Voice-only / silent calls | π | β Decided (stay on 111) | ERT + university |
| R7 | Cleartext SIP/RTP | π | π΅ Deferred (future sec) | IT |
| R8 | Single PBX SPOF | π | β Mitigated (van failover) | IT |
| R9 | SIP abuse / DoS | π | β Documented (doc 26) | IT |
| R10 | DPDP compliance | π | βΈοΈ Out of scope (deferred by decision) | University + IT |
| R11 | Retention automation | π‘ | β Scripted (retention-cleanup.sh) | IT |
| R12 | Bus factor / RACI | π | β Documented (doc 27) | IT + university |
| R13 | Cost / BOM | π‘ | β οΈ Open (deferred by request) | IT + university |
| R14 | Prank / false alarms | π‘ | β Documented (doc 26 Β§6) | ERT + IT |
| R15 | Doc aids | βͺ | β Done (glossary + quick-cards) | IT |
E. Recommended actions before go-live¶
- Resolve the five β decisions (R1 power, R3 night coverage, R6 silent-call policy, R10 compliance) with the university β these are not engineering choices.
- Promote UPS (R1) into Phase 1 infrastructure.
- Add TLS/SRTP for ERT/fixed devices (R7) and basic hardening (R9).
- Write the surge posture (R4) into the ERT SOP.
- Prepare a warm-standby server (R8) and train a backup admin (R12).
None of these block starting the build β the Phase 1 MVP is still correct. They block declaring the system production-ready and trustworthy for real emergencies.