Features

Unpatchable KeyWe smart lock can be easily picked

A design flaw in the KeyWe smart lock (GKW-2000D), which is mostly used for remote-controlled entry to private residences, can be exploited by attackers to gain access to the dwellings, F-Secure researchers have found.

KeyWe smart lock

To add insult to injury, in this present incarnation the lock can’t receive firmware updates, meaning that the security hole can’t be easily plugged.

About KeyWe smart lock

KeyWe smart lock is developed by the Korean company KeyWe, which raised money for it on Kickstarter.

The lock can be opened via an application (Wi-Fi, Bluetooth), an armband (NFC), through a touchpad (numeric code), or mechanically (with a regular key).

It has additional options like generating one-time guest codes, unlocking the door based on proximity, etc.

About the vulnerability and the attack

F-Secure security consultants acquired the KeyWe Smart Lock by pledging on Kickstarter.

They analyzed its hardware and firmware, as well as the hardware and firmware of the accompanying KeyWe bridge (which is used to connect the lock to a wireless network) and the code of the associated Android app.

They discovered that, while the company did implement some security protections for the lock and app (not so much the bridge), a flaw in the in-house developed key exchange protocol can be exploited to, ultimately, get the secret key needed to unlock the lock.

“The hardware needed [to perform the attack] is a board able to sniff Bluetooth Low Energy traffic. It can be bought for ~10$ and used out-of-the-box,” Krzysztof Marciniak, cyber security consultant at F-Secure, told Help Net Security.

“In terms of software, this requires additional work from the attacker – in our case a Python script was developed, but pretty much any language can be used as long as it can interact with a Bluetooth controller. It should also be mentioned that the mobile application needs to be analyzed (one needs to retrieve the key generation algorithm) in order to execute this attack.”

The user doesn’t even have to lock/unlock the door with the application for the attacker to intercept the operator password – they just need to run/open the mobile application. Once the app is run, it connects to the lock to check its status, and the password can be intercepted.

The attacker (or just the intercepting device) must be within 10-15 meters from the victim for the traffic interception to work. The recording of the traffic can later be analyzed to extract the key value needed to generate the lock-opening key.

More technical information about their research and discovery can be found here and here, but since the lock can’t receive firmware updates, the researchers decided to not to share some crucial details.

Symptoms of a larger problem

The vendor has acknowledged the issue and is working on fixing it, the researchers noted, but since the lock has no firmware upgrade functionality, already deployed locks will remain vulnerable.

“The mobile application does use Bluetooth (Smart/Low Energy), so that option is not safe either. NFC could be used to counter this attack, but it is prone to other attacks (cloning the access key [armband], intercepting the traffic with proper equipment etc.),” Marciniak told us.

“The touchpad option, however, seems to be the right fallback here. That being said, the mobile application should still be paired with a mobile device – otherwise a malicious user can pair with it without any additional owner confirmation.”

Lock owners will need to replace the lock or live with the risk. The vendor told the researchers that new iterations of the app will contain a fix for this issue and, equally important, new locks will have the firmware upgrade functionality.

One cannot say that no attention has been given to security, the researchers noted, but rolling your own in-house cryptography is always a risky proposition, and so is doing no threat modeling before design and development.

“Security isn’t one size fits all. It needs to be tailored to account for the user, environment, threat model, and more. Doing this isn’t easy, but if IoT device vendors are going to ship products that can’t receive updates, it’s important to build these devices to be secure from the ground up,” Marciniak pointed out.

He recommends consumers to consider the security implications of internet-connectivity before replacing their offline devices with online versions, and advises device vendors to perform security assessments on their products as part of their design.

Mini 4WD is an electrifying race series for makers and tinkerers

Services Name

Description

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Quam vulputate dignissim suspendisse in est ante in. Vitae congue eu consequat ac felis. Placerat duis ultricies lacus sed turpis. Diam ut venenatis tellus in metus. Pellentesque elit ullamcorper dignissim cras tincidunt lobortis feugiat vivamus. Interdum consectetur libero id faucibus nisl. Sit amet consectetur adipiscing elit duis tristique sollicitudin nibh. Morbi enim nunc faucibus a.

Benefits

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.

data protection officer

Features

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.

Our approach

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.

connect_1.1

Connect

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. 

connect step

Prepare

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. 

visualise

Visualise

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. 

implement

Implement

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. 

Why clients choose us to help with their [service name]

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.

Professional & Courteous

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.

Highly Skilled Experts

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.

Timely Delivery

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.

A deep understanding of IT Security

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.

advice you can trust

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.

In depth training & support

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.

Other services
Contact us

Cloudy with a chance of neurons: The tools that make neural networks work

Machine learning is really good at turning pictures of normal things into pictures of eldritch horrors.

Enlarge / Machine learning is really good at turning pictures of normal things into pictures of eldritch horrors.
Jim Salter

Artificial Intelligence—or, if you prefer, Machine Learning—is today’s hot buzzword. Unlike many buzzwords have come before it, though, this stuff isn’t vaporware dreams—it’s real, it’s here already, and it’s changing your life whether you realize it or not.

A quick overview of AI/ML

Before we go too much further, let’s talk quickly about that term “Artificial Intelligence.” Yes, it’s warranted; no, it doesn’t mean KITT from Knight Rider, or Samantha, the all-too-human unseen digital assistant voiced by Scarlett Johansson in 2013’s Her. Aside from being fictional, KITT and Samantha are examples of strong artificial intelligence, also known as Artificial General Intelligence (AGI). On the other hand, artificial intelligence—without the “strong” or “general” qualifiers—is an established academic term dating back to the 1955 proposal for the Dartmouth Summer Project on Artificial Intelligence (DSRPAI), written by Professors John McCarthy and Marvin Minsky.

All “artificial intelligence” really means is a system that emulates problem-solving skills normally seen in humans or animals. Traditionally, there are two branches of AI—symbolic and connectionist. Symbolic means an approach involving traditional rules-based programming—a programmer tells the computer what to expect and how to deal with it, very explicitly. The “expert systems” of the 1980s and 1990s were examples of symbolic (attempts at) AI; while occasionally useful, it’s generally considered impossible to scale this approach up to anything like real-world complexity.

Sadly, we're not here yet.

Enlarge / Sadly, we’re not here yet.
NBCUniversal

Artificial Intelligence in the commonly used modern sense almost always refers to connectionist AI. Connectionist AI, unlike symbolic AI, isn’t directly programmed by a human. Artificial neural networks are the most common type of connectionist AI, also sometimes referred to as machine learning. My colleague Tim Lee just got done writing about neural networks last week—you can get caught up right here.

If you wanted to build a system that could drive a car, instead of programming it directly you might attach a sufficiently advanced neural network to its sensors and controls, and then let it “watch” a human driving for tens of thousands of hours. The neural network begins to attach weights to events and patterns in the data flow from its sensors that allow it to predict acceptable actions in response to various conditions. Eventually, you might give the network conditional control of the car’s controls and allow it to accelerate, brake, and steer on its own—but still with a human available. The partially trained neural network can continue learning in response to when the human assistant takes the controls away from it. “Whoops, shouldn’t have done that,” and the neural network adjusts weighted values again.

Sounds very simple, doesn’t it? In practice, not so much—there are many different types of neural networks (simple, convolutional, generative adversarial, and more), and none of them is very bright on its own—the brightest is roughly similar in scale to a worm’s brain. Most complex, really interesting tasks will require networks of neural networks that preprocess data to find areas of interest, pass those areas of interest onto other neural networks trained to more accurately classify them, and so forth.

One last piece of the puzzle is that, when dealing with neural networks, there are two major modes of operation: inference and training. Training is just what it sounds like—you give the neural network a large batch of data that represents a problem space, and let it chew through it, identifying things of interest and possibly learning to match them to labels you’ve provided along with the data. Inference, on the other hand, is using an already-trained neural network to give you answers in a problem space that it understands.

Both inference and training workloads can operate several orders of magnitude more rapidly on GPUs than on general-purpose CPUs—but that doesn’t necessarily mean you want to do absolutely everything on a GPU. It’s generally easier and faster to run small jobs directly on CPUs rather than invoking the initial overhead of loading models and data into a GPU and its onboard VRAM, so you’ll very frequently see inference workloads run on standard CPUs.

Guidemaster: The best tech that will make your home an even better place

irobot roomba 980

iRobot

We could all use a little more help around our home, and luckily now there’s a lot of tech that can lend a hand. There are a plethora of smart home devices that can do everything from lock your doors, vacuum your carpets, or keep a watchful eye over your possessions while you’re away.

Wading through the ocean of smart home tech isn’t easy—and, admittedly, much of the smart home space is not worth your time or your money. However, we’ve tried (and personally purchased) many home tech devices that actually do deliver on what they promise. These items make keeping your home how you like it much easier.

Not all of the home tech we recommend falls into the large and nebulous category of “the Internet of Things,” either—some are kitchen appliances, home speakers, gaming accessories, and other devices that most people primarily use in the home in order to make that space feel more like our own. Some after a lot of lived-in testing time, here’s all of the home tech that we think would make great gifts this holiday season.

Note: Ars Technica may earn compensation for sales from links on this post through affiliate programs.

Philips Hue lights

Philips Hue color smart light bulbs.

Enlarge / Philips Hue color smart light bulbs.
Philips

One of the easiest ways to start making your home smarter is with smart light bulbs and Philips’ Hue line are a good option. First, you can get white or color bulbs—while most will be happy with plain, ol’ white, color bulbs can be fun if you want to add personality to a room with color-changing light scenes.

Second, all Hue bulbs connect to a bridge that comes with most Hue starter packs. The bridge helps the lights communicate with each other and with your home Wi-Fi, which is how you control them. Using the Hue mobile app, you can turn on and off individual lights or entire rooms lights, dim them to your liking, and set schedules. You can have all the lights in your home come on before you arrive home from work, so you’re not walking into a dark house.

Third, Hue light bulbs connect to a bunch of other smart home systems like Works with Alexa, IFTTT, Apple HomeKit, the Google Assistant, and more. That means you can control your lights using voice commands or other smart commands that you customize. Not only are Hue lights an easy and affordable way to get into smart home tech, but they also make the lights in your home even more convenient to control on a regular basis.

Philips Hue White and Color starter set product image

Philips Hue White and Color starter set

(Ars Technica may earn compensation for sales from links on this post through affiliate programs.)

Zojirushi rice cooker

Zojirushi rice cooker.

Enlarge / Zojirushi rice cooker.
Zojirushi

I make a lot of rice and I’ve gone through at least two rice cookers in the process. After my last $25 rice cooker broke on me, I decided to invest in the Zojirushi NS-TSC10 Micom rice cooker and—this is not hyperbole—it’s changed my cooking life. Gone are the days of burnt or undercooked rice as Zojirushi’s magical machine has propelled me into a world where all kinds of rice are cooked to perfection every single time.

I attribute this to actually reading the directions that come with the rice cooker. If you do this and follow the instructions, washing the rice before cooking and using the proper settings on the cooker itself, everything made in this machine will be tasty. In addition to rice, Zojirushi’s machine comes with a steaming basket for steaming vegetables and other foods, and it even has a cake setting.

But the machine truly shines make rice. You don’t have to guess how much water to include as the interior pot has indicators for that, and you don’t have to guess cooking times either. The machine senses how much rice and water you put into the pot and automatically sets the cooking time. All you have to do is wait for it to play a cute little jingle as soon as your rice is done and then experience rice heaven. I’ll never go back to a cheap rice cooker again, and I implore anyone who eats a lot of rice to consider a Zojirushi machine.

Zojirushi NS-TSC10 rice cooker product image

Zojirushi NS-TSC10 rice cooker

(Ars Technica may earn compensation for sales from links on this post through affiliate programs.)

5G won’t change everything, or at least probably not your things

Artist's impression of millimeter-wave 5G speeds.

Enlarge / Artist’s impression of millimeter-wave 5G speeds.
Aurich Lawson / Getty

The long-touted fifth generation of wireless communications is not magic. We’re sorry if unending hype over the world-changing possibilities of 5G has led you to expect otherwise. But the next generation in mobile broadband will still have to obey the current generation of the laws of physics that govern how far a signal can travel when sent in particular wavelengths of the radio spectrum and how much data it can carry.

For some of us, the results will yield the billions of bits per second in throughput that figure in many 5G sales pitches, going back to early specifications for this standard. For everybody else, 5G will more likely deliver a pleasant and appreciated upgrade rather than a bandwidth renaissance.

That doesn’t mean 5G won’t open up interesting possibilities in areas like home broadband and machine-to-machine connectivity. But in the form of wireless mobile device connectivity we know best, 5G marketing has been writing checks that actual 5G technology will have a lot of trouble cashing.

A feuding family of frequencies

The first thing to know about 5G is that it’s a family affair—and a sometimes-dysfunctional one.

Wireless carriers can deploy 5G over any of three different ranges of wireless frequencies, and one of them doesn’t work anything like today’s 4G frequencies. That’s also the one behind the most wild-eyed 5G forecasts.

Millimeter-wave 5G occupies bands much higher than any used for 4G LTE today—24 gigahertz and up, far above the 2.5 GHz frequency of Sprint, hitherto the highest-frequency band in use by the major US carriers.

At those frequencies, 5G can send data with fiber optic speeds and latency—1.2 Gbps of bandwidth and latency from 9 to 12 milliseconds, to cite figures from an early test by AT&T. But it can’t send them very far. That same 2018 demonstration involved a direct line of sight and only 900 feet of distance from the transmitter to the test site.

Those distance and line-of-sight hangups still persist, although the US carriers that have pioneered millimeter-wave 5G say they’re making progress in pushing them outward.

“Once you get enough density of cell sites, this is a very strong value proposition,” said Ashish Sharma, executive vice president for IoT and mobile solutions at the wireless-infrastructure firm Inseego. He pointed in particular to recent advances in solving longstanding issues with multipath reception, when signals bounce off buildings.

There are a lot of "5G" stock images available. Some of them are more optimistic than others. This is one of the more optimistic ones.

Enlarge / There are a lot of “5G” stock images available. Some of them are more optimistic than others. This is one of the more optimistic ones.
Photographer is my life / Getty

Reception inside those buildings, however, remains problematic. So does intervening foliage. That’s why fixed-wireless Internet providers using millimeter-wave technology like Starry have opted for externally placed antennas at customer sites. Verizon is also selling home broadband via 5G in a handful of cities.

Below millimeter-wave, wireless carriers can also serve up 5G on mid- and low-band frequencies that aren’t as fast or responsive but reach much farther. So far, 5G deployments outside the US have largely stuck to those slower, lower-frequency bands, although the industry expects millimeter-wave adoption overseas to accelerate in the next few years.

“5G is a little more spectrally efficient than 4G, but not dramatically so,” mailed Phil Kendall, director of the service provider group at Strategy Analytics. He added that these limits will be most profound on existing LTE spectrum turned over to 5G use: “You are not going to be able to suddenly give everyone 100Mbps by re-farming that spectrum to 5G.”

And even the American carriers preaching millimeter-wave 5G today also say they’ll rely on these lower bands to cover much of the States.

For example, T-Mobile and Verizon stated early this year that millimeter-wave won’t work outside of dense urban areas. And AT&T waited until it could launch low-band 5G in late November to start selling service to consumers at all; the low-resolution maps it posted then show that connectivity reaching into suburbs.

Sprint, meanwhile, elected to launch its 5G service on the same 2.5GHz frequencies as its LTE, with coverage that is far less diffuse than millimeter-wave 5G. Kendall suggested that this mid-band spectrum will offer a better compromise between speed and coverage: “Not the 1Gbps millimeter-wave experience but certainly something sustainable well in excess of 100Mbps.”

The Federal Communications Commission is working to make more mid-band spectrum available, but that won’t be lighting up any US smartphones for some time.

(Disclosure: I’ve done a lot of writing for Yahoo Finance, a news site Verizon owns.) 

Video: How Oddworld solved its narrative problems with mind control

[embedded content]
Video shot by Sean Dacanay, edited by Jeremy Smolik. Click here for transcript.

Some games entice you into playing them with loud marketing campaigns, sexualized cover art, or the promise of ludicrous over-the-top violence. But then there are games like Lorne Lanning’s Oddworld series—games that don’t lead with muscle- or bikini-clad heroes and defy easy categorization. Games like Oddworld tempt you into playing by promising a different kind of experience. There are guns and violence, sure, but the setting is strange, the plot is filled with gray, and the hero—well, Abe isn’t exactly sexy, or really even, you know, human.

But players who gave the original Oddworld a chance back in 1997 found themselves stumbling through a unique and fascinating world that was equal parts surprising and subversive, and the series has gone on to acquire legitimate cult-success status. With the approaching release of Oddworld: Soulstorm in 2020, we thought it was a good time to pay a visit to Lorne Lanning and his team at Oddworld Inhabitants, and talk about our favorite meat processing factory worker and his long journey from design notebook to screen.

“Write what you know,” they say…

We interviewed Lanning at the Emeryville, CA headquarters of Oddworld Inhabitants, the studio he co-founded with Sherry McKenna in 1994. For Oddworld fans, the office was a magical place, stuffed with the kind of memorabilia that amasses over more than two decades of game design. Lanning walked us through his journey to become a game creator, starting from his poor beginnings in what sounds like an unstable family. He got into video games because his father had a job at Coleco, and Lanning thought gaming would be a good way to meet girls.

Lanning’s ambitions weren’t aimed at the small screen—he had his eyes set on making movies. To pay the bills, he took a job at TRW Aerospace, where he worked on anti-missile defense systems (it was the 1980s, and Reagan’s Strategic Defense Initiative boondoggle was in full swing). His exposure to soul-crushing bureaucracy and supplier management formed the basis for many of the Brazil-seque ideas later presented in the Oddworld games.

But it’s the time Lanning spent at Rhythm and Hues Studios that had the biggest effect on Oddworld—at least the series’ collective look and feel. Working on visual effects set him on the path of visualizing game design in terms of cinema—not just how things were framed on screen, but also the discipline and budgeting style of the movie industry. When Lanning and McKenna (a fellow Rhythm and Hues alum) eventually started their own studio in the 90s, they approached their game and their character designs in the way Hollywood does. This obviously is de rigeur in 2019, but in 1995 when work on Oddworld started, it was most definitely not the industry norm.

The problem

When designing the first Oddworld game, Lanning and his team had to confront an annoying reality of game design—there are only so many ways to interact with the world in a side-scrolling action game, and a lot of those ways involve shooting stuff. And one of the immovable design goals of Oddworld was that protagonist Abe would go through the entire game without being armed—not because of any kind of political stance against guns, but because having Abe unarmed increases the character’s vulnerability in a world that’s already overwhelmingly hostile. A gun would provide an easy solution to many of the game’s problems, and where’s the fun in that?

It took some time to work out a solution, but Lanning the other designers decided that characters like Yoda don’t need guns to solve problems. They instead infused the game with a pastiche of mysticism drawn from a number of different sources, which gave Abe his secret weapon: the ability to possess other characters, including bad guys with guns. This let them then design in some puzzles involving shooting, which the player can solve by finding a bad guy, taking over his body, and having the bad guy shoot his way through the puzzle. To prevent the player from picking up the bad guy’s gun after the puzzle is solved, NPCs violently explode after possession.

An Odd(world) legacy

This video ended up being extremely long in the rough cut because Lanning gave us so much great interview material. We had to trim out quite a bit, but we’ll be producing an extended version if there’s enough interest in this video. There are several rabid Oddworld fans here at the Ars Orbiting HQ, and this video, like several others in the War Stories series, was a passion project with a lot of emotion invested in it (not to mention some custom voiceover lines performed by Lanning just for us!). We hope you enjoy watching it as much as we enjoyed making it.

Dealmaster: All the best Cyber Monday 2019 tech deals happening right now

Dealmaster: All the best Cyber Monday 2019 tech deals happening right now

Enlarge (credit: Ars Technica)

Greetings, Arsians! The Dealmaster is back with a jumbo-sized roundup of hand-picked Cyber Monday tech deals. To be candid, many of the discounts we’re seeing on Black Friday’s sister holiday are similar to what we saw last week. We do have some new offers worth noting, including new Nintendo Switch bundles, a Sonos speaker sale, monitor deals, and more. Still, for online tech deals, Cyber Monday is primarily a good opportunity to capitalize on the discounts you may not have checked out on Black Friday.

Not that this is a bad thing. Even after sorting through all the junk on offer, we’ve still got a boatload of worthwhile deals that cover just about every type of tech gadget. As we did with our Black Friday deals list, we’ve highlighted a few deals we particularly like ahead of our full rundown. You can see it all below.

(P.S. Because the Dealmaster is a company guy, he’ll note that Ars has its own Cyber Monday sale: New Ars Pro++ subscribers can get discounts on a YubiKey 5c or YubiKey 5 NFC device.)

Read 12 remaining paragraphs | Comments

How neural networks work—and why they’ve become a big business

How neural networks work—and why they’ve become a big business

Enlarge (credit: Aurich Lawson / Getty)

The last decade has seen remarkable improvements in the ability of computers to understand the world around them. Photo software automatically recognizes people’s faces. Smartphones transcribe spoken words into text. Self-driving cars recognize objects on the road and avoid hitting them.

Underlying these breakthroughs is an artificial intelligence technique called deep learning. Deep learning is based on neural networks, a type of data structure loosely inspired by networks of biological neurons. Neural networks are organized in layers, with inputs from one layer connected to outputs from the next layer.

Computer scientists have been experimenting with neural networks since the 1950s. But two big breakthroughs—one in 1986, the other in 2012—laid the foundation for today’s vast deep learning industry. The 2012 breakthrough—the deep learning revolution—was the discovery that we can get dramatically better performance out of neural networks with not just a few layers but with many. That discovery was made possible thanks to the growing amount of both data and computing power that had become available by 2012.

Read 48 remaining paragraphs | Comments