Hi-tech approach keeps hackers out of  smart homes

Hi-tech approach keeps hackers out of smart homes

Having designed a hardware-based firewall that sits right between your home WiFi hotspot and the cloud, Californian CUJO LLC. is turning IoT security on its head. Built around a Cavium dual core 1GHz processor with 1GB of DDR SRAM and 4GB of Flash, the bowl-shaped device performs deep packet inspections on the transiting traffic and applies a proprietary and adaptive machine-learning algorithm to identify and notify the end-user about abnormal data behaviour or stop a hack right in its track.
By eeNews Europe

Share:

In effect, CUJO is acting as a gateway between your local devices and your connection to the Internet, thwarting virtual intrusions such as malware, viruses and hacking techniques which would necessarily transit via your WiFi or Ethernet connection. The device is not only looking for known signatures, like an antivirus, but also learning from the users’ typical data behaviour to prevent what could be seen as abnormal traffic behaviour.

"Existing solutions typically only react to existing threats, but we apply machine learning algorithms to offer a proactive solution that keeps improving". Then the learnings performed on all the meta data collected can be applied to all CUJO protected homes.

But any machine learning algorithm needs to mature to offer the best protection, so how was this tested and for how long?

"Certainly. More data gives us more tools to deliver better results. We can develop more accurate and more elaborate algorithms" told us CUJO’s CEO, Einaras Gravrock, "We have been building and testing algorithms since early this year, though our team has engineers who have done this for much of their careers".

"We’ve tried to hack up to 40 different devices tied to one hub, we’ll extend that to more and will also use the services of 3rd party companies to test different hacking scenarios before we ship our devices. But so far, we haven’t seen a limit to the number of devices which could be protected".

So what would be an abnormal behaviour CUJO could be picking on?

"Say you have a smart device that pulls data files from your PC to Australia", Gravrock puts as an example.

The CEO didn’t want to say more on the type of data patterns CUJO may be screening for, nor did he want to reveal the computational requirements of their machine learning algorithm.

"We would rather not discuss the qualitative or quantitative aspects of our cloud right now. Since we launched we have received a lot of interest from companies that we compete with. Therefore we will try to be very careful when discussing technology" he said.

In the long term, if encryption became the norm across all IoT devices, would Cujo still be relevant (being deprived from deep packet inspection)?

"I doubt that we are anywhere near to a future where all IoT devices are hack-proof by default. Too many variables, too many integrations between devices, too many device makers who don’t have the resources to stay protected. Unfortunately this is the environment that we are in for the foreseeable future" replied Gravrock.

If so, installing this sort of self-adaptive firewall may be your best bet. But beyond scrutinizing your data flows, won’t there always be social engineering threats?

"Education is of paramount importance and that’s why we are educating our customers about such risks. It depends of the scenario, but at the end of the day, our goal is to make it unreasonably difficult to penetrate a single home", the CEO concluded.

The company’s business model is similar to that of an antivirus company, offering annual subscription services for continuous protection (USD 89 annually on top of the initial device purchase). CUJO is expected to ship during the first quarter of 2016.

Get your CUJO on indiegogo

Visit the company’s page at https://www.getcujo.com/

Smart2.0
10s