News Your mesh Wi-Fi can now double as a home security system — Gamgee uses home Wi-Fi networks for intruder detection

Admin

Administrator
Staff member
Regarding the last full paragraph:

"While this project is fascinating and seems to hold great promise, there are several things to be wary of. Firstly, we always advise readers to be careful with their money, as backing a crowdfunder isn’t the same as purchasing a product outright. Secondly, it is much easier to promise a great product than to deliver one, so third-party reviews are essential here. The Gamgee Wi-Fi Home Alarm seems to be a product that would be particularly susceptible to detection glitches, as it simply promises so much precision and detail. Thirdly, even the AI giants aren’t getting it right all the time, far from it. AI hallucinations and blatant errors seem to be commonplace in 2024, and accuracy would be paramount for a system such as this."

The paragraph should be in bold print. [I added bold here.]
 
I expect it'll suffer from blind spots and false positives (i.e. failing to recognize an occupant that it should). I'd be more worried about the latter than the former, since an intruder can't stay only in the blind spots.

I find it intriguing that someone decided to productize this tech, but if they can't get the false-positive rate low enough without causing too many false-negatives (i.e. missed detections of unknown individuals), then it'll fail. Imagine the bad press it would get from a break-in that it fails to detect!

Another risk would seem to be Wi-Fi jammers. The base station better know how to recognize when it's being hit with a jamming attack and it better have a hard-wired internet connection to inform you of that.
 
Last edited:
The article said:
even the AI giants aren’t getting it right all the time, far from it. AI hallucinations and blatant errors seem to be commonplace in 2024, and accuracy would be paramount for a system such as this."
Hallucinations apply to generative AI, which this isn't. Yes, the accuracy concerns are definitely a fair concern to raise (as I mentioned), but given how much press AI topics are getting these days, you'd hope the author would be more familiar with the jargon and use it correctly.

Furthermore, there's really no comparison between the type of deep learning models used in such a detector and LLMs. Pointing to hallucinations by LLMs is like talking about rocket launch failures in the context of discussing airplane safety. That's about how far off the mark it is.
 
Last edited:
In some says using the word hallucinations might be better in some ways when dealing with a non technical audience even though it really isn't correct. It is surprising how well know the concept of AI hallucination is by the general public...seems to be trendy on social media. Lately any software is called "AI" and since it is seeing things that do not exist it is kinda correct.

In the end if it scares off a few non technical people who throw money at stupid stuff maybe a valid misrepresentation.
 
One possible advantage to this system would be WIFI jamming, that's becoming more common in break-ins to disable cameras, being useless.
I'm sure wi-fi jamming would disable this system, also.

If you merely wanted an alert when a wi-fi jammer is in use, jammer detection is a feature that could be integrated into a standard wi-fi router.
 
I'm sure wi-fi jamming would disable this system, also.

If you merely wanted an alert when a wi-fi jammer is in use, jammer detection is a feature that could be integrated into a standard wi-fi router.
This system is looking for 'devations' from normal WIFI signals. Jamming expressly causes deviations in the signal. It will be the same as blowing an air-horn in a microphone field. Blatantly obvious.
 
  • Like
Reactions: dalef
This system is looking for 'devations' from normal WIFI signals.
Not just any deviations, but structured ones that it recognizes as being caused by people in the home. You don't want some kids playing outside to set it off, nor do you want a piece of furniture being out of place to cause false alarms. So, it not only needs to be quite sensitive, but also very selective.

Jamming expressly causes deviations in the signal.
And it also disrupts communication with the mesh extenders, rendering them pointless. But, you don't even need a system like this to detect a jammer. I'm sure a run-of-the-mill wi-fi router could do it, were it designed to.
 
Last edited:
@bit_user

The grey shaded text in Post #4 is not mine. I did not write that text.

I was only referencing the entire paragraph from the end of article which includes the sentence attributed to me.

I did apply bold font to the entire paragraph which I noted in Post #2.

The objective being my belief that the risks regarding crowd-funding, etc. should have been high-lighted all that much more.

Just keeping the record straight.
 
  • Like
Reactions: dalef and bit_user
@bit_user

The grey shaded text in Post #4 is not mine. I did not write that text.

I was only referencing the entire paragraph from the end of article which includes the sentence attributed to me.
Apologies. Fixed.

Note how I use a [quote="The article"] tag, when I'm quoting the text. Just a suggestion. I also sometimes wrap a quote in [indent] tags, if I'm citing it for some reason. I know the WYSIWYG editor struggles with indentation, but such problems can be avoided by staying in markup mode.

Just keeping the record straight.
Thanks! Corrections are always welcome!
: )
 
Reads "AI" in product description.

Ah, another grift.
Okay, then tell us how you'd design a highly-accurate, wi-fi based person detection and recognition system.

On a related note, it might interest you to know that after decades of R&D, conventional face recognition systems are functionally extinct. All face recognition systems in wide usage are now based on deep learning, simply because it's that much better than anything else even the smartest PhD's, backed by many $Millions in funding, have tried.

But, go ahead. Since signal processing of wi-fi is so much easier and more intuitive than clear images in visible light, you should have no trouble designing an effective solution that's free of any "AI" technologies.
\s
 
Gonna be funny when they try to bring their tech to the US, etc. and patent infringement suits start popping up outta the blue.

To my knowledge, this technology has been used for marketing, for years.
 
The technology presented by Gamgee, which uses Wi-Fi for intruder detection, does not allow for the verification of alarm authenticity. Currently, there are alarm systems that detection combined with video verification to ensure that alarms are legitimate. This dual approach provides a visual confirmation, significantly reducing false alarms and enhancing the reliability of the security system. Without such verification, the new Wi-Fi-based system may struggle to distinguish between real threats and false positives.
 
  • Like
Reactions: bit_user
The technology presented by Gamgee, which uses Wi-Fi for intruder detection, does not allow for the verification of alarm authenticity. Currently, there are alarm systems that detection combined with video verification to ensure that alarms are legitimate. This dual approach provides a visual confirmation, significantly reducing false alarms and enhancing the reliability of the security system. Without such verification, the new Wi-Fi-based system may struggle to distinguish between real threats and false positives.
Nobody is keeping you from also having a camera and separately doing verification, though. IMO, that seems like a necessity.

Yes, it'd be better if they were integrated with the wi-fi detection, so you could see exactly which person the wi-fi system through was unfamiliar. But, a well-chosen camera location should probably be able to see the intruder at some point, if there really is one.
 
Okay, then tell us how you'd design a highly-accurate, wi-fi based person detection and recognition system.
Why would I ever want to do that?

Why would anyone sane want to use consumer grade Wi-Fi system which is usually accesible from outside using some crappy vendor cloud system or even TR069 from the ISP, or poorly secured by default to act as important security infrastructure in their home, without power backup to boot?

There are dedicated alarm systems and sensors which are tamper-proof, have backup power using batteries and are properly secured and wired to raise an alarm with your security provider if you are not at home.

Of the top of my head you could disable this simply by:

1. Cutting power
2. Performing DoS by sending authenticate packets with bogus encryption keys
3. Hacking it remotely (even over Wi-Fi itself)

Bonus points -- if you gain access to it then you can also spy on the very people who were supposed to be protected by it.

Being able to do something doesn't automatically mean it's a good idea.
 
There are dedicated alarm systems and sensors which are tamper-proof, have backup power using batteries and are properly secured and wired to raise an alarm with your security provider if you are not at home.
What's interesting about this is that it can conceivably work while people are at home! For instance, let's say you leave your young children at home with a babysitter. If it were trained on the babysitter, it could give you an alert if anyone else showed up, such as either an unexpected friend of the babysitter's or maybe even home invaders.

Of the top of my head you could disable this simply by:

1. Cutting power
2. Performing DoS by sending authenticate packets with bogus encryption keys
3. Hacking it remotely (even over Wi-Fi itself)
That's if you even knew the home had such a system. Given the alleged popularity of wi-fi jammers, I think a lot of burglars lack the skills, equipment, time, and planning to cut power to a home. Besides which, they would have to assume the router wasn't on a UPS, if all they did was cut the power.

Bonus points -- if you gain access to it then you can also spy on the very people who were supposed to be protected by it.
That's true of any network-connected security system.