The Indian Canvas #83: Who keeps a factory's regulator camera feed running?

An order from the Commission for Air Quality Management now requires certain polluting factories in the National Capital Region to point cameras at their fuel-feeding points and stream the video live to regulators. Installing a camera is a one-off job. The order also makes the factory responsible for keeping the feed running and for reporting long outages with a fix. This issue asks whether that second job could be a business: who would pay, how it would run, what it would cost to find out, and how to pilot it. It is a hypothesis. I found no evidence yet that feeds fail often enough to pay for a service. I have not spoken to a factory owner, an installer or an official.

What the order says

According to a PIB press release of 16 October 2026, CAQM has extended the requirement for Pan-Tilt-Zoom cameras to all categories of industry under its Statutory Direction No. 98. The release lists the covered units as the 17 categories of highly polluting industries identified by the Central Pollution Control Board, Red Category (medium and large) air-polluting industries, food and textile units with boilers or thermic fluid heaters, and metal units with furnaces, in the NCR. I read the press release, not the order itself.

As the release describes it, cameras must cover fuel-feeding points, gas meters where gas is used, stack emission and continuous-emission-monitor sampling points, and fume extraction areas, without significant blind spots. Feeds must stream continuously by RTSP, a standard video-streaming protocol, to servers of the CPCB and the state pollution control boards or the Delhi committee. Units must provide power backup for cameras and connectivity, keep the system working, restore it promptly after a fault, and report any prolonged disruption to the board with the corrective action taken. The deadline in the release is 30 November 2026. The release says the order replaces one dated 5 October. Check CAQM for later changes before relying on that date.

The stated reason: the Commission's flying squads found units using fuels outside the approved list. The Trity Enviro, a compliance consultancy, makes the same point and notes that a unit can switch back to an approved fuel before an announced inspection.

What I could and could not establish

The deadline is not the only pressure. A regional officer of the Haryana pollution board told The Tribune that over 100 red-category units in Bahadurgarh were being notified and that action would follow after 30 November. The Gurgaon Chamber of Commerce and Industry had asked for more time, saying the work involves choosing camera locations and angles, equipment, cabling, electrical and network work, and commissioning, and that vendors and technical teams may be hard to find. The small-business federation FISME has raised the cost of cameras, connectivity and upkeep for smaller units, according to KNN India.

I could not find: how many units are covered, how many already have cameras, what the receiving portals require, whether a third party may monitor a unit's stream, what "prolonged" means in the disruption rule, or any data on how often such feeds drop. Without that last one I cannot say the problem is real.

Not empty: the install side is crowded

Compliance consultancies such as Trity Enviro and others are already publishing explainers and inviting enquiries, and installers look like the obvious route for a unit. I did not review their service terms, so I do not know whether annual maintenance is bundled. If it is, a separate uptime service has less room. That is the first thing to find out.

The idea in one paragraph

A shared "feed desk" for an industrial cluster. It watches whether each unit's cameras are online and whether the stream is up, calls a technician from a pool the units share when it is not, and keeps a dated log of each outage, cause and fix that the unit can show the board. Payer: the unit, or the cluster's industry association paying on behalf of its members. Missing piece: outside monitoring of stream health, shared field technicians, and a log in a form a board can read. India adaptation: industrial areas already have associations that negotiate common services, so the buyer can be one body speaking for many small units. The unit stays legally responsible under the order. The service does not change that.

How it would run, stage by stage

1. Choose a cluster

My choice, not a sourced rule: an industrial area with an association willing to sponsor a pilot, and several units that fall under the order. Whether a given unit is covered is for its own consultant or the pollution board to say. I would not guess.

2. Map each unit to the order

For each unit, list the points the release names: fuel-feeding points, gas meters, stack and sampling points, fume extraction. Record where cameras sit, how they are powered and how they connect. This is also where the camera field-of-view problem shows up, since the release says the view must cover the points without significant blind spots.

3. Watch health, not footage

My design choice: the desk checks whether a camera is online, whether the stream is reaching its destination, whether the image is clear at night, and whether power and the network link are up. It does not watch what the cameras see. Whether a third party can check delivery to the regulator's server at all depends on how the CPCB portals work, which I could not find. If the only way to verify delivery is the regulator's own portal, the desk can only watch the unit side of the link.

4. Power and connectivity

The order requires adequate power backup and continuous connectivity. Options to test: a battery backup on the camera line, a second network link, or a shared arrangement per block of units. I priced none of them. The pilot should record every outage by cause: power, network, camera, blocked view, or misaim.

5. A shared technician pool

One team covers several units, which is cheaper than each unit keeping a technician only if outages are not all at once. A winter storm or a grid failure could take down many feeds together. The response target should come from the pilot, not from me.

6. The outage log

For each event: time down, time restored, cause, fix, who did it. The release says prolonged disruption must be reported with the corrective action, and does not say how long counts as prolonged. Ask the board for its definition, and keep the log in whatever form it accepts.

7. Money in, money out

A monthly fee per camera or per unit, or a lump sum from an association. The service pays technicians, tools and software. GST on the fee is a question for an accountant. I put no price on it.

8. The cost ledger, with no invented numbers

These are the lines to measure:

• Monitoring software and the network checks it needs.

• Technicians: hours on call, travel and spare parts.

• Spare cameras, power units and network gear held in stock.

• Site surveys and onboarding of each unit.

• Insurance and any legal advice on responsibility, data access and the regulator's portal rules.

• Sales and association time, because the buyer is a committee.

• Time with no outages, which is revenue with little work and which pays for the weeks with many.

9. Pricing logic

The fee has to cover technician cover for the worst weeks, not the average ones, plus tools, software and a margin, and still be smaller than what a unit would spend on its own annual-maintenance contract. I do not know that contract price. The pilot should ask five units what they pay or are quoted.

A pilot with gates

Gate 0, questions only. Ask the state pollution board and the CPCB what the receiving portal needs and whether a third party can see stream health. Ask them what counts as a prolonged disruption. Ask a lawyer who is responsible for a feed that fails while a service is paid to watch it. Ask two installers whether maintenance is bundled. If the answers close the door, stop.

Gate 1, measure before building. With three to five units that have installed cameras, and with their consent, log feed up and down time for six weeks and classify each outage by cause. If feeds almost never drop, the idea is dead. If most outages come from power, the answer may be a battery, not a service.

Gate 2, a concierge desk. One association, a spreadsheet and a phone group, one technician, one cluster, through a winter month. Set the success tests before starting, such as outages fixed within a window agreed with the units, and whether units renew. I do not invent those numbers.

Gate 3, grow only if Gate 2 passes. A second cluster, then software.

How it could fail

Installers may bundle upkeep, leaving no gap. The regulator may show each unit its own feed status on a portal, which would make the desk redundant. The order may change again or be enforced loosely. The Red-category part of the order covers medium and large units, which may have their own maintenance teams. Outages may be rare, or caused by power alone. Winter faults may arrive all at once and overwhelm a shared team. A unit may blame the desk for a missed report, even though the order puts the duty on the unit. And a buyer who is a committee can take months to say yes.

Sources and photo credit

PIB press release on the PTZ camera order, KNN India on FISME and the order, Hindustan Times on the chamber request, The Tribune on Bahadurgarh notices, Trity Enviro (consultancy) on the order. Cover photo: a security camera on a pole against a blue sky by Turquo Cabbit (@turquocabbit) on Unsplash, free to use under the Unsplash License. It gives no location, so it is illustrative only, not an Indian factory. This issue has no chart: I have no data series I trust.