Your alarm clock jars you awake. You stumble to the kitchen, fire up your coffee maker, grab some milk from the fridge, and pour yourself a bowl of cereal. You turn down the thermostat before you head to your car. You park your sedan in your usual spot in the garage at work, and you enter your office building by waving your badge at the door. Finally, you drop into your chair and fire up your computer.
A mundane story, one millions of people trudge through every day with only minor variations. But here’s the question: how many Internet-connected devices did you interact with between opening your eyes and logging in? Let’s see: alarm clock, coffee maker, fridge, thermostat, your automobile, all the stop lights, traffic cameras, toll transceivers, and in-road traffic sensors on your commute, and finally your badge and the door. OK, maybe your household appliances aren’t on the Internet yet. Give them a few years.
Now ask yourself: how many of those net-connected doodads are secure? The answer: none of them. Every device on this list is woefully unprotected from various attacks, and to make matters worse, many of them might contain confidential information ripe for the picking. And if all that weren’t sufficiently disconcerting, the vendors of such miscellany aren’t particularly motivated to make them secure – even if they knew how to do it properly. Which they don’t. Nevertheless, we blindly forge ahead, building out the Internet of Things (IoT), as though the security issues will somehow resolve themselves. Just how worried should we be?
The Bad and the Ugly – but None of the Good
This tale of woe begins with Radio Frequency Identification (RFID) tags. These innocuous tags appear in everything from product packaging to airport tarmac equipment to passports to, yes, your security badge. And as you would expect from the tone of this ZapFlash, RFID tags are dead simple to hack. They come in two flavors: passive and active. The passive ones need no power source; they simply respond when the right signal gets close enough to them. No encryption, no authentication, no nothing. Anyone with the right device (which you can easily obtain over the Internet, of course) can read your tag simply by getting their snooping device close enough to it. Have you ever walked down the street with your security badge, or through an airport with your passport? Has anybody ever passed within a few feet of you? Stupid questions, right?
So, how do the best RFID security minds recommend protecting your RFID tags from compromise? Put them in protective sleeves. And no, wrapping your passport in aluminum foil won’t do. You need a special Faraday cage sleeve. But even if you manage to keep your RFID tags in an effective sleeve, all a hacker has to do is wait till you take it out. Recommending a sleeve to protect the IoT from attack is about as effective as climbing under school desks was at surviving a Cold War nuke.
Surely the technology in our increasingly cyber-aware automobiles is more secure than your run of the mill RFID tag, right? Sorry, no. Today’s cars have fifty or more tiny computers called electronic control units that control all aspects of the vehicle’s function. These units communicate with each other via a Controller Area Network (CAN). As vehicle manufacturers increasingly provide Internet access to their autos, hackers can easily access the CAN remotely – and with it, all the functions of the car. Brakes. Steering. Engine. Everything down to the radio.
There are two primary modes of protection the car manufacturers are implementing to prevent hackers from using these weaknesses to steal cars, kill targeted individuals, or simply wreak havoc. First, CAN protocols are proprietary. And second, the manufacturers are keeping all the details secret.
Neither technique, of course, provides any true measure of security, as researchers proved at a recent DefCon conference. Secrets are virtually impossible to keep in today’s Facebooked world. Also keep in mind, any authorized repair shop will have a diagnostic machine that interfaces with the CAN. If a hacker doesn’t want to bother reverse engineering the proprietary protocol directly, they can simply get their hands one of those machines and hack that.
Why the IoT is so Hard to Secure
There are both business and technical reasons why the IoT is so difficult to secure. On the technical side, the core problem is that the tried-and-true technologies we use to secure traditional interactions with the Internet just don’t work well – if they work at all. To use Public Key Infrastructure (PKI) technology, for example, each endpoint must be able to store digital keys and run encryption and decryption algorithms, conduct sophisticated handshakes to establish secure SSL connections, etc. However, many IoT nodes like the passive RFID tags simply don’t have the electrical power, storage, or processing power necessary to tackle even the simplest of PKI tasks.
Secondly, a large part of the IoT approach involves machine-to-machine (M2M) communication. In other words, sensors and other IoT endpoints talk to each other, instead of talking to a server somewhere. If your smart thermostat tells your dishwasher when to run, that communication might be running over your home Wi-Fi or perhaps Bluetooth or some other local network protocol that doesn’t require traffic to actually go over the Internet. And not only does it go without saying that Wi-Fi and Bluetooth protocols are shockingly easy to hack, but how are the two communicating nodes supposed to know that the information coming from the other is authorized? Essentially, any kind of M2M interaction requires a certain level of trust, only we have no way of providing that trust in the first place, or revoking it should a breach occur. How will your dishwasher know someone has hacked your thermostat?
In fact, the two examples above provide special cases of a broader problem: the IoT gives us no way to control permissions. Let’s say you figure it’s a good idea for said thermostat to Tweet certain information so it’s easy for you to monitor your home while you’re away. If a hacker compromises the thermostat, they automatically get your Twitter login – and you no longer have any way to control your Tweets.
The final challenge I’ll consider here (keeping in mind there are sure to be dozens of others) is the fact that devices on the Internet must have IP addresses – and in many cases, IoT sensors wouldn’t work properly behind firewalls. They must have public IP addresses that anyone can access. And if someone can access them, then someone will. Ever heard of Shodan? It’s a tool for finding IP addresses for random devices, including baby monitors, Webcams, security systems, and all manner of other bric-a-brac. How would you like a hacker to compromise your baby monitor? It’s happened before, and it’ll happen again.
Scanning random IP addresses, however, is only practical for the familiar IPv4 space. As we move to IPv6, there will be so many possible addresses that scanning them at random will be much more difficult. This advantage, however, is weaker than you might think. First, it simply presents an interesting challenge to enterprising hackers out there. How long will it take for a Shodan 2.0 to be IPv6 compatible? Secondly, IPv6 can actually make it more difficult for an organization with many IoT sensors to secure them (assuming they have any idea how to do so in the first place), because IPv6 makes it more difficult for an authorized party to scan for them as well. And if you don’t know what devices and sensors you have, you can’t control, manage, or secure them.
Such technical issues, of course, aren’t the whole story. On the business side, the problems are even more slippery. There is no agreement on how or even whether to address IoT security. Few countries have any regulation requiring companies to implement security in their devices. And there’s no market pressure forcing such vendors to get their act together. We, the customers, have simply grown too complacent. If we won’t pay more for secure automobiles and refrigerators, then rest assured no company will bother to go through the trouble to secure them.
The ZapThink Take
You were hoping I had some slick, imaginative approach for solving these issues, right? Sorry to disappoint. But rather than throwing our collective hands in the air, dumping all our devices down the garbage chute, and moving to a cave on Borneo somewhere, we must realize that the only way we’ll ever solve this riddle is by taking an entirely different perspective on securing technology.
We cannot impose security from the outside onto each sensor. It’s simply too easy for hackers to get a hold of them and defeat whatever mechanism we’ve put in place. Instead, the sensors themselves must be inherently secure. Only when a hacker can break open a sensor, reverse engineer it as well as the communication protocols it uses, and still not be able to hack into it or use it to hack into something else will we finally be able to sleep at night. Solve this challenge and I promise you, you’ll be very, very rich.