Security and CCTV — Algarve

Most of the security systems we are called to look at do work, in the sense that the cameras are recording and the alarm arms. They fail at the moment they are needed, and for reasons that were decided at design: a camera that shows someone was there but not who, an alarm that cried wolf so often it was left disarmed, a system nobody could reach from abroad when the neighbour called.

iHome designs security as part of the property’s electrical and network infrastructure — cameras, intrusion detection, access control and intercom on the same documented installation as everything else, rather than as a separate box bolted on afterwards.

Cameras: megapixels are not the specification

The question a camera has to answer is not “did something happen” but “who was that”. The measure that decides it is pixels per metre at the target — how much of the sensor lands on a face or a number plate at the distance where it matters.

The European standard for video surveillance systems, EN 62676-4, puts numbers to it, and they are the numbers worth designing against:

• Detect — around 25 px/m. Something is there.
• Observe — around 62 px/m. Characteristic details, clothing, behaviour.
• Recognise — around 125 px/m. You can tell whether you know the person.
• Identify — around 250 px/m. Sufficient for identification of a stranger.

A single high-resolution camera covering a whole driveway delivers detection across the width of it and identification nowhere. The same budget spent on a wide camera for context plus a tighter one on the gate delivers both. This is why camera positions are drawn on a plan before anything is bought: the count and the lenses follow from where identification is actually required — entrances, gates, the approach to the house — and everything else only needs to see that something moved.

Night, and the sun

Night performance is where the specification sheet and the real image diverge most.

Infrared makes a camera see, not resolve. A camera with strong IR produces a bright picture of an empty garden and a blurred smear of anyone walking through it, because holding a usable exposure in low light means a slow shutter, and a slow shutter and a moving person do not coexist. Where identification is required after dark, the honest answers are a larger sensor with a faster lens, or light — modest, well-aimed white light, which also happens to be the more effective deterrent. IR bounced back off a nearby wall, a railing or a glass pane washes out the frame entirely, which is why an IR camera under an eave often performs worse than the same camera on a pole.

The Algarve sun is the other half of the problem. A camera looking at a doorway from inside, or at a gate with the sky behind it, spends most of the day in the worst case for a sensor: a bright background and a face in shadow. Wide dynamic range handles some of it; camera orientation handles more. Pointing away from the low winter sun costs nothing at design and cannot be fixed afterwards.

False alarms are the failure mode that actually happens

Break-ins are rare. Nuisance alerts are constant, and they are what destroys a system — not because the technology stops working, but because the owner stops looking. A phone that buzzes forty times a night gets muted, and then the one alert that mattered is muted with it.

What causes them here is predictable: vegetation moving in the wind, insects drawn to infrared illuminators, spiders building webs directly across a lens, cats, headlights sweeping a boundary wall, and rain lit by IR. Pixel-based motion detection reacts to all of it.

The fix is a layer of judgement in the system rather than in the owner. Detection that classifies what it sees — person, vehicle — instead of counting changed pixels, discards the large majority. Detection zones drawn to exclude the public road and the neighbour’s palm remove most of the rest. On the intrusion side, dual-technology detectors that require both infrared and microwave confirmation before triggering, and correctly chosen pet-immune units, do the same job indoors.

Getting this right is unglamorous, iterative commissioning work over the first weeks of use. It is also the single biggest determinant of whether a system is still armed and still watched a year later.

Recording, retention and where the footage lives

Storage follows arithmetic: bitrate per camera, multiplied by cameras, multiplied by the days you want to keep. H.265 and per-camera smart encoding cut it substantially compared with older equipment, and recording on motion or object events rather than continuously cuts it further — at the cost of the seconds before the event, which is precisely why sensible systems keep a continuous low-rate stream alongside the event recordings.

Two details are worth insisting on. Recorders should use drives rated for continuous surveillance write loads rather than desktop drives, which fail early under it. And drive health should be monitored, because the classic discovery is that the disk died four months ago and nobody knew — a recorder gives no outward sign of having stopped recording.

Retention should be a deliberate decision, not the default the recorder shipped with. Keeping footage far longer than any plausible use is a liability rather than an asset.

Cameras on the internet: the part that is usually wrong

The most common serious defect we find in existing installations is a camera or recorder exposed directly to the internet through port forwarding, often with the credentials it left the factory with. Cameras are among the most attacked devices there are, they are rarely patched, and an exposed one is a way into the network and, historically, a recruit into large botnets.

Remote viewing does not require it. A properly built system puts cameras on their own isolated network segment with no route to the internet, and gives the owner access through a VPN to the property or through the manufacturer’s authenticated cloud relay. From the owner’s phone the experience is identical; from outside, the difference is total.

This is one of several reasons we treat security and the network as one design rather than two.

Intrusion detection, access and the door

An alarm system is graded by the level of threat it is designed to resist — EN 50131 defines four such grades, and the appropriate one for a property depends on what is in it and how long it stands empty, which in the Algarve is often most of the year. That decision drives detector selection, tamper protection, signalling and the standby requirement, and it should be made explicitly rather than inherited from whatever the panel supplier normally fits.

Perimeter detection — contacts on doors and windows, and detection at the glass line — has a particular advantage in a holiday property: it allows the system to be armed while the house is occupied, protecting the shell without the occupants setting off internal detectors on the way to the kitchen. Volumetric detection then covers the interior when the house is empty.

Door stations and access control are the part of the system used every single day, and the part most often specified last. A video intercom that reaches a phone abroad, an access code for the cleaner that works on Tuesdays, a gate that opens from the car — these are the functions that determine whether the household experiences the security system as useful or as an obstacle. They are also where integration pays: the same door camera on the television and the touch panel, the gate on the same keypad as the lighting.

Where it joins the rest of the house

On a property with KNX or Control4, security stops being a separate app. Arming the alarm on the way out can close the shading, drop the heating setpoint and switch off the lighting circuits that were left on. An alarm at night can bring the interior lighting up to a safe level rather than leaving the household in the dark, and an intrusion event can start recording at full rate on the relevant cameras.

None of this is exotic; all of it depends on the systems having been specified to talk to each other, which is a decision taken at the beginning of a project and expensive to retrofit between products bought separately.

Privacy and the law

Cameras on a private property in Portugal are legitimate, and they are bounded. The principles that do not change: film your own property and not the public road or a neighbour’s land, signal the presence of recording at the approaches, keep footage no longer than the purpose requires, and restrict who can see it. Coverage that spills beyond the boundary is both a legal exposure and, in practice, a source of dispute with neighbours.

We assess this at design and position cameras accordingly, including masking where a view cannot be avoided.

How a project runs

1. Risk and requirement. What needs protecting, what needs identifying, how the property is used through the year, and who needs access.
2. Design. Camera positions, lenses and pixel density on plan; detection layout and grade; cabling, power and network; where the recorder and rack live.
3. Installation. Cabling, mounting, isolated network segment, protection and labelling.
4. Commissioning. Focus and field of view verified against the design, exposure tuned day and night, detection zones and classification tuned over real use.
5. Handover. Documentation, credentials transferred to the owner, and training for whoever uses it daily.
6. Monitoring and maintenance. Drive health, device availability and remote access verified rather than assumed.

Working with iHome

We have been designing and installing electrical, network and security systems in Algarve properties since 2006, and we are an official Control4 dealer and certified KNX partner, nationally awarded for our integration work. Security is specified alongside the electrical installation and the network because that is where it either becomes part of the house or remains a set of devices nobody maintains.

Call (+351) 289 090 900 or email geral@ihome.pt to discuss a project.