UPES-ECS Multi-Campus Wireless Bridge (Later Phase)¶
Status: Future / later phase — not part of Phase 1. Goal: extend UPES-ECS from Bidholi to Kandoli over a rooftop point-to-point wireless link, keeping the system "LAN-only" (no public internet, no PSTN, no cloud).
Source note (Juniper.md): "site-to-site wireless connectivity from rooftop, Bidholi to Kandoli, without centralised infra." This document turns that into a plan.
1. Concept¶
BIDHOLI campus LAN KANDOLI campus LAN
┌────────────────────┐ rooftop PtP wireless ┌────────────────────┐
│ upes-ecs-pbx-01 │◄══════ Layer-2/3 ═══════►│ SIP phones / APs │
│ (Asterisk/FreePBX) │ bridge link │ ERT/fixed devices │
└────────────────────┘ └────────────────────┘
The wireless bridge makes the two campuses behave like one LAN, so existing UPES-ECS features (111, SAP-ID calling, queue, paging, conferences) work across sites without a second PBX or any cloud dependency.
2. Two possible models¶
| Model | How | When |
|---|---|---|
| A. Single PBX (bridge only) | One Asterisk at Bidholi; Kandoli devices register across the wireless link | Simplest; link must be stable + low-latency |
| B. Second PBX (resilient) | A PBX per campus, linked by inter-PBX trunk over the bridge; each site survives a link outage | More robust; more to manage |
Recommendation: start with Model A (matches "without centralised infra" and minimal added complexity); move to Model B only if the link proves unreliable or each campus needs independent survivability.
3. Link requirements¶
The wireless bridge must meet voice-grade quality end to end:
| Metric | Target |
|---|---|
| One-way latency (site-to-site) | < 150 ms |
| Packet loss | < 1% |
| Jitter | Low / stable |
| Throughput | Comfortably above peak concurrent RTP (each call ≈ 80–100 kbps) + headroom |
| Availability | Line-of-sight, weather-tolerant, monitored |
- Line of sight rooftop-to-rooftop is mandatory for PtP wireless.
- Prioritise RTP (voice) with QoS/DSCP across the link.
- Treat the link itself as a critical monitored device in Health Monitoring.
4. Numbering & identity across campuses¶
- SAP-ID identity model is unchanged — one directory spans both campuses.
- Fixed-device ranges may be split by site for clarity, e.g. Bidholi
4000–4499, Kandoli4500–4999(finalize in the numbering drill). - Emergency numbers (111/101/199, paging, conferences) stay identical at both sites.
- Paging zones extend with Kandoli-specific zones (new 70x codes).
5. Resilience considerations (Model B)¶
- Each campus PBX can answer local 111 even if the link drops.
- ERT queue membership can be site-aware so a caller reaches nearest responders first.
- Cross-site conference/paging degrade gracefully when the link is down (local-only).
- Config + backups replicated to both sites.
6. Security¶
- The wireless link carries only UPES-ECS LAN traffic — encrypted (WPA3/enterprise or link-layer encryption on the PtP radios) and firewalled to allowed subnets.
- No public exposure introduced by the bridge; it is an internal campus-to-campus link.
- Guest traffic never traverses the emergency bridge.
7. Rollout¶
1. Survey line-of-sight Bidholi ↔ Kandoli rooftops
2. Install + align PtP radios; measure latency/loss/jitter/throughput
3. Bridge the two LANs; verify cross-site SIP registration + 2-way audio
4. Pilot Model A: Kandoli devices on the Bidholi PBX
5. Cross-site test: 111, SAP-ID calling, paging, conference, recording
6. If link reliability is marginal → move to Model B (PBX per site + trunk)
7. Add link monitoring to the Health Dashboard
8. Out of scope (still)¶
Even multi-campus, UPES-ECS stays LAN/wireless-LAN only: no public internet dependency, no PSTN, no cloud PBX, no cellular reliance. The wireless bridge is an extension of the campus LAN, not an internet link.
9. TBD¶
Rooftop survey + line-of-sight confirmation · PtP radio model/vendor · link throughput vs. expected concurrent calls · Model A vs B decision · site-split numbering · QoS support on the radios and switches.