UPES-ECS Master Implementation Plan¶
System: UPES Emergency Communication System (UPES-ECS) Build: FreePBX (web admin over Asterisk), one local server, LAN-only Goal: the smallest working emergency line first, then layer everything else on top.
The whole system reduces to one MVP: Dial 111 → ERT phone rings → call is recorded → if unanswered it goes to voicemail and becomes a reviewable incident. Everything else layers on without changing that core.
Phase map¶
| Phase | Name | Outcome | Effort |
|---|---|---|---|
| 0 | Foundations | Two phones can call each other over Wi-Fi | ½–1 day |
| 1 | Emergency core (MVP) | Dial 111 works end-to-end — go/no-go milestone | 2–4 days |
| 2 | Roles & access | Student calling + locked-down emergency features | 2–3 days |
| 3 | Coordination | Paging, conference rooms, dispatch, availability | 3–5 days |
| 4 | Operational hardening | Health checks, backup/restore, SOP + first drill | 2–4 days |
| 1.5 / 2 | AI line 101 | AI triage that always falls back to 111 | later |
| Later | Multi-campus | Bidholi ↔ Kandoli rooftop wireless bridge | later |
Each phase is independently usable — you can stop after Phase 1 and already have a real emergency line.
Phase 0 — Foundations¶
Do:
1. Install Ubuntu Server LTS (or Debian stable) on the existing server; hostname upes-ecs-pbx-01.
2. Assign a static IP (mandatory) and record subnet/gateway.
3. Install FreePBX (official distro or on top of Asterisk).
4. Confirm the LAN path: Wi-Fi client isolation OFF (or a voice VLAN) so phones can reach the PBX for SIP + RTP.
5. Set local name resolution pbx.upes.lan / sip.upes.lan (or document the static IP fallback).
6. Create 2 test PJSIP extensions + enable echo test 198.
Exit criteria: two phones on campus Wi-Fi register and call each other; 198 echo works.
Phase 1 — Emergency core (MVP) ← the milestone¶
Do:
1. Provision ~10 SAP-ID extensions (CSV import) in the correct contexts.
2. Create queue ert_emergency_queue — ring all available, 20s timeout, min 2 agents.
3. Point 111 → recording starts immediately → queue.
4. Build the escalation chain: queue → ERT Lead 4101 (20s) → backup group (4300+4200+warden/admin, ring-all 20s) → Emergency Voicemail.
5. Configure MixMonitor recording on the 111 flow; naming ERT-YYYYMMDD-0001_...wav.
6. Create Missed Emergency Incident handling (Critical, Pending Review).
7. Create 199 drill line mirroring 111 but with no real dispatch, logs DRILL-ONLY.
Exit criteria (all must pass): dial 111 → ERT answers → recorded; unanswered 111 → voicemail → missed incident appears; 199 works and dispatches nothing real.
Phase 2 — Roles & access¶
Do:
1. Create the 7 contexts (ctx_student … ctx_admin) per the Role Matrix.
2. Enable student-to-student / internal calling (not recorded).
3. Provision fixed devices in 4000–4999 (Medical 4200, Security 4300, ERT desks 41xx).
4. Lock paging, conference, recordings, voicemail review to emergency roles.
5. Enforce: no anonymous SIP, unique ≥12-char passwords, LAN-only registration.
Exit criteria: a student account can call 111 + another student but is denied paging/conference/recording; fixed devices show correct caller ID.
Phase 3 — Coordination¶
Do:
1. Paging 700–799 (ConfBridge/Page), PIN on 700, all attempts logged.
2. Conference rooms 9000–9004 with PINs; 9000 recorded when active.
3. Dispatch workflow: warm transfer + three-way bridge; blind transfer restricted.
4. Responder availability: *45/*46 pause/resume; queue reflects busy/paused/offline.
Exit criteria: authorized paging + 9000 work; unauthorized attempts are blocked and logged; transfers and pause/resume behave per SOP.
Phase 4 — Operational hardening¶
Do:
1. Health checks (script/CLI first, dashboard later) — see Health Checklist.
2. Backups: git repo upes-ecs-config, daily config + pre-change snapshots, encrypted sensitive data — see Backup & Restore.
3. Tested restore (config restore under 1 hour target).
4. Train ERT on the SOP; run the first drill; capture a post-drill review.
Exit criteria: health check reports correctly; a restore test passes; ERT completes a drill successfully.
Later phases¶
- AI line 101 (19-AI-101-Design.md) — AI-first triage on its own number, always escalates/falls back to 111. 111 never depends on AI.
- Multi-campus (20-Multi-Campus-Wireless.md) — rooftop point-to-point wireless links Bidholi ↔ Kandoli while staying "LAN-only."
TBD items to collect from UPES IT (blocking go-live)¶
| Item | Needed for |
|---|---|
| Router / switch / AP models | Capacity, VLAN/QoS, client-isolation behaviour |
| Server specs + OS confirmation | Capacity, backup planning |
| Static server IP / subnet / gateway | SIP client config |
| Campus Wi-Fi SSID | Setup guides |
| Client-isolation status | SIP/RTP connectivity |
| Allowed subnets | Firewall / access rules |
| Final ERT roster | Queue membership |
| Final fixed-phone locations | Numbering + provisioning |
| Recording-retention policy | Legal/admin compliance |
| Go-live approval authority | Sign-off |
Success criteria (Phase 1 go/no-go)¶
Mobile SIP registration over Wi-Fi · SAP-ID → SAP-ID calling · any authenticated user can call 111 · 111 reaches ERT queue · escalation works · voicemail works · recording works for 111 · student calls not recorded · restricted features denied · health check passes (script level) · backup/restore config test passes · 199 drill safe · SOP understood in a responder drill. All must pass before go-live.
Rollback triggers: 111 fails, queue unavailable, recording fails, PBX unstable, Wi-Fi can't carry calls, unauthorized-access risk, or no available ERT coverage.