Perimeter security guide
MEMS, fibre optic, taut wire or video analytics: how to choose
There is no best perimeter intrusion detection system, only one that suits the boundary you have. We sell MEMS and fibre optic detection, so we have an interest in two of the four columns below — which is exactly why the table states where ours are the wrong choice, and names the two categories we do not sell at all.
Most perimeter projects go wrong at specification rather than installation. A technology gets chosen for the perimeter the buyer wishes they had — well tensioned, well lit, uniform, with power everywhere — and is then installed on the one they actually have.
So start from the boundary. Its construction, its length, whether power and cable can reach along it, whether there are sightlines, and what has to happen when an alarm fires: those five answers narrow the field faster than any feature comparison.
The four technologies side by side
| MEMS fence sensor | Fibre optic sensing | Taut wire | Video analytics | |
|---|---|---|---|---|
| Detection principle | An accelerometer mounted on the fence senses vibration travelling through the fabric. | The sensing fibre is itself the detector — disturbance alters the light returning along it. | Tensioned wires on dedicated posts; sensors detect displacement or a change in tension. | Cameras plus software that classifies objects and movement within the scene. |
| What it detects | Climbing, cutting, drilling and impact on the fence or wall it is fixed to. | Disturbance of the cable and of the fence it is attached to. Distributed acoustic sensing additionally resolves the acoustic signature. | Cutting, spreading or climbing — anything that moves the wires. | Approach and loitering before the boundary is touched, and what the intruder is. |
| Typical false-alarm triggers | Wind, especially on a poorly tensioned fence; heavy rain; nearby vehicle traffic; animals on or along the line. | Wind and rain on the cable; ground-borne vibration; movement of the cable itself where it is loosely secured. | Few from weather, which is the main reason it is still specified. Thermal expansion and ground settlement show up as drift rather than alarms. | Changing light, headlights, shadows, rain or fog on the lens, moving foliage, and insects drawn to infrared illuminators at night. |
| Supported surfaces | Fencing, brick and masonry walls. Not rock walls. | Anywhere the cable can be run or buried — fence, wall or trench. | Its own posts. It is the barrier as well as the sensor. | Not surface-dependent, but needs sightlines and adequate illumination. |
| Power and cabling along the line | Powered nodes and a bus cable running the length of the perimeter. | No power and no electronics on the line; all active equipment sits at the interrogator. | Power to each sensor post. | Power and network to every camera. |
| Locating an event | To the individual node, where each node is separately addressed. | Intensity-based sensing: to a sensing loop. Distributed acoustic sensing: to a position along the cable. | To the zone between two sensor posts. | To the camera's field of view, with an image of what caused it. |
| Where it is the wrong choice | No fence or wall to mount on; a boundary so slack it cannot be tensioned; runs long enough that cabling every node is impractical. | Short perimeters where one interrogator cannot be justified; intensity-based sensing where you need to know where along the line it happened. | Retrofitting to an existing fence; sites unwilling to carry the mechanical maintenance; boundaries that are irregular or frequently altered. | As the only means of detection where sightlines are blocked, fog is common or lighting is poor. It watches the approach; it does not feel the fence. |
| From Gandeevan | Prachir N100 node and M100 master — available. | Abhedya L100 is in field trials; Abhedya D100 is in design. | Not offered. | Not offered as a sensor. Sanchalan links your existing CCTV to perimeter events. |
Start from the boundary you already have
What is the boundary made of? Fencing, brick and masonry suit a fence-mounted sensor. Rock walls do not. A boundary that is a line on a map rather than a physical barrier suits fibre in a trench or cameras, not a fence sensor.
How long is it? A few hundred metres of fence is comfortable for discrete nodes on a bus. Kilometres of pipeline or boundary is where distributed fibre sensing earns its cost, because one interrogator covers the whole run from one end.
Can power and cable reach along it? If not, the choice narrows sharply: fibre optic sensing puts no electronics and no power on the line at all.
Are there sightlines and light? Video analytics tells you who is approaching before the boundary is touched, which no vibration sensor can do. It needs to be able to see, which rules it out as a sole means of detection on foggy or unlit sites.
What happens when an alarm fires? If somebody has to assess before responding, detection on its own is not enough; you need the camera view that goes with the event. That is an integration question, and it is usually where the cost that was not budgeted appears.
Where each one is the wrong choice
The table says this per technology; it is worth saying in prose too, because this is the part most comparisons leave out.
A MEMS fence sensor is wrong where there is nothing suitable to mount it on, where the fence is too slack to be tensioned, or where the run is long enough that cabling and powering every node becomes the dominant cost. It is also wrong on rock walls, which we exclude explicitly.
Fibre optic sensing is wrong on a short perimeter, where a single interrogator cannot be justified against discrete sensors. Intensity-based fibre sensing is also wrong where you need to know where along the line the disturbance happened, as distinct from that it happened — that is what distributed acoustic sensing adds, at considerably greater cost.
Taut wire is wrong as a retrofit to an existing fence, since it brings its own posts and is the barrier as well as the sensor. It is also wrong where nobody will carry out the periodic retensioning it needs, or where the boundary is irregular or gets altered often. Where those conditions do not apply, its weather immunity is genuinely hard to beat.
Video analytics is wrong as the only means of detection where sightlines are blocked, fog is common or lighting is poor. It watches the approach; it does not feel the fence. Used alongside a fence sensor rather than instead of one, it is the fastest way to turn an alarm into a decision.
A combination is usual, not a sales tactic
Most sites that are happy with their perimeter run two layers: something that detects contact with the boundary, and something that lets an operator see what caused it. The sensor says where and what kind; the camera says who.
That only works if the two are actually joined. Sanchalan exists for this: cameras are associated with perimeter zones, and when an event fires the footage around it is kept as a separate clip, already linked to the event that caused it. The alternative — two systems, two screens and an operator hunting for the right timestamp — is the usual reason recorded evidence turns out to be missing when it is needed.
What we sell, and what we do not
Prachir, our MEMS fence sensing system, is available today. Abhedya L100, intensity-based fibre optic detection, is in field trials, and we are not quoting a general availability date until those are complete. Abhedya D100, the distributed acoustic sensing system, is in design — there is no hardware to trial. If your requirement is live, deploy what exists rather than holding a site open for a product that does not.
We do not sell taut wire and we do not sell video analytics as a detection technology. If either is the right answer for your site, that is the answer, and we would rather tell you than sell you something that underperforms on it.