Nexus Observatory Sign in

StoriesSystem: Resilient operations

When the links fail

Observing systems matter most on the days infrastructure fails. Designing for that day means local autonomy, stored power, store-and-forward delivery1 and networks that keep working when cut off from their core2.

Systems
Isolated operation, Delay-tolerant delivery, Stored power
Toward 2030
Resilience as a design requirement
Figures
4, systems diagram, simulator, timeline to 2030, field scene

Go to the references

01Systems diagram

Four ways a site loses the world.

Fibre cut, power out, satellite path down, core unreachable. A resilient site has an answer to each: a second path, a battery, a queue and local services. Work through the scenarios and watch what keeps flowing.

Scenario

Figure as text: A field site under four failure scenarios

A four-tier diagram: instruments, the site’s local network and power, two backhaul paths, and the operations centre.

  • Instruments: Weather, Seismic, Camera
  • Site: Local cell, Edge and queue
  • Backhaul: Fibre, Satellite
  • Operations: Operations

8 links. Scenarios: Normal operation; Fibre cut; Fibre and satellite lost; Local cell down.

02Simulator

How long the site lasts on battery.

When the grid fails, the battery and the array decide how long the site keeps measuring. Set the load and the store for your site and a cloudy week.

Figure as text: Battery autonomy of a field site

A line chart of stored energy over fourteen days with the daily balance and a reserve line.

Battery autonomy of a field site, at the starting values
Battery7200 Wh
Station load80 W
Solar array800 W
Peak sun hours1.2 h
Autonomy without sun3.0 days
Daily energy balance−1,200 Wh
Days to empty at this balance4.8 days

03Timeline to 2030

Store, carry, forward.

Delay-tolerant networking treats a missing path as normal: bundles are stored until a contact appears, then forwarded31. For observations, that means a gap in delivery, not a gap in the record. Public-safety networks add isolated operation, so a cell keeps serving local users without its core2.

Figure as text: Standards for operating while disconnected

A timeline from 2007 to 2030 with the delay-tolerant networking architecture, isolated operation for public safety and Bundle Protocol 7.

  1. 2007-04 Delay-tolerant networking architecture (RFC 4838). Store-and-forward for networks without end-to-end paths.3
  2. 2016-03 Isolated operation for public safety (3GPP TS 22.346). Cells keep serving local users without a backhaul to the core.2
  3. 2022-01 Bundle Protocol 7 (RFC 9171). The current standard for delay-tolerant delivery.1
  4. 2030 Sendai targets fall due. Including reduced disruption of basic services.4

04Field scene

The site that keeps measuring.

Back on the valley: when every link is down, the instruments still record and the edge still keeps time and order. The question for the design review is not whether the site will be cut off, but for how long it can stay useful.

Figure as text: A valley site that must keep measuring

An alpine valley with a cell tower, weather station, radar, seismic vault and edge node.

  • Edge and queue
  • Local cell
  • Seismic

Figure 1

Systems diagram

Sources reaching a sink
3 of 3
Longest route
4 hops
State
Normal operation

Figure 1. A field site under four failure scenarios. A four-tier diagram: instruments, the site’s local network and power, two backhaul paths, and the operations centre.

References

  1. RFC 9171: Bundle Protocol version 7. Internet Engineering Task Force, 2022.
  2. 3GPP TS 23.180 and TS 22.346: mission-critical services and isolated operation for public safety (IOPS). 3rd Generation Partnership Project, 2016.
  3. RFC 4838: Delay-tolerant networking architecture. Internet Engineering Task Force, 2007.
  4. Sendai Framework for Disaster Risk Reduction 2015–2030. United Nations Office for Disaster Risk Reduction, 2015.

Related stories

System: Satellite backhaul When the backhaul is a satelliteFor a remote site, the satellite is not a backup; it is the network. Since Release 17, the cellular standard includes satellites as part of the radio access…System: Edge AI inference Deciding at the edgeA model that detects smoke, flooding or a fault is most useful where the frame is captured. Running inference at the edge cuts the time to a decision and…System: Timing and PNT Time as infrastructureA 5G cell must hold its phase within 3 µs of its neighbours, ±1.5 µs of a common reference; a synchrophasor’s one per cent vector error leaves room for only…

Where this applies

Operations centreEdge compute & AIConnectivityField servicesPublic safetyCivil resilienceHumanitarianGovernmentField tour: mountain valleyField tour: arctic station