It hasn’t taken long for the question of security to rise to the top the list of concerns about the Internet of Things. If you are going to open up remote control interfaces for the things that assist our lives, you have to assume people will be motivated to abuse them. As cities get smarter, everything from parking meters to traffic lights are being instrumented with the ability to remotely control them. Manufacturing floors and power transmission equipment are likewise being instrumented. The opportunities for theft or sabotage are hard to deny. What would happen, for example, if a denial of service attack were launched against a city’s traffic controls or energy supply?
Privacy is a different, but parallel concern. When you consider that a personal medical record is worth more money on the black market than a person’s credit card information, you begin to realize the threat. The amount of personal insight that could be gleaned if everything you did could be monitored would be frightening.
The problem is that the Internet of Things greatly expands the attack surface that must be secured. Organizations often have a hard enough time simply preventing attacks on traditional infrastructure. Add in potentially thousands of remote points of attack, many of which may not be feasible to physically protect, and now you have a much more complex security equation.
The truth is that it won’t be possible to keep the Internet of Things completely secure, so we have to design systems that assume that anything can be compromised. There must be a zero trust model at all points of the system. We’ve learned from protecting the edges of our enterprises that the firewall approach of simply controlling the port of entry is insufficient. And we need to be able to quickly recognize when a breach has occurred and stop it before it can cause more damage.
There are of course multiple elements to securing the Internet of things, but here are four elements to consider:
1) “Things” physical device security – in most scenarios the connected devices can be the weakest link in the security chain. Even a simple sensor that you may not instinctively worry about can turn into an attack point. Hackers can use these attack points to deduce private information (like listening in on a smart energy meter to deduce a home occupant is away), or even to infiltrate entire networks. Physical device security starts with making them tamper-resistant. For example, devices can be designed to become disabled (and data and key wiped) when their cases are opened. Software threats can be minimized with secure booting techniques that can sense when software on the devices has been altered. Network threats can be contained by employing strong key management between devices and their connection points.
Since the number of connected things will be extraordinarily high, on boarding and bootstrapping security into each one can be daunting. Many hardware manufacturers are building “call home” technology into their products to facilitate this, establishing a secure handshake and key exchange. Some manufacturers are even using unique hardware-based signatures to facilitate secure key generation and reduce spoofing risk.
2) Data security – data has both security and privacy concerns, so it deserves its own special focus. For many connected things, local on-device caching is required. Data should always be encrypted, preferably on the device prior to transport, and not decrypted until it reaches it’s destination. Transport layer encryption is common, but if data is cached on either side of the transport without being encrypted, then there are still risks. It is also usually a good idea to insert security policies that can inspect data to ensure that it’s structure and content is what should be expected. This discourages many potential threats, including injection and overflow attacks.
3) Network security – beyond securing the transmission of data, the Internet of things needs to be sensitive to the fact that it is exposing data and control interfaces over a network. These interfaces need to be protected by bi-lateral authentication, and detailed authorization policies that constrain what can be done at each side of the connection. Since individual devices cannot always be physically accessed for management, remote management is a must, enabling new software to be pushed to devices, but this also opens up connections that must be secured. In addition, policies needs to be defined at the data layer to ensure that injection attacks are foiled. Virus and attack signature recognition is equally important. Denial of service type attacks also need to be defensed, which can be facilitated by monitoring for unusual network activity and providing adequate buffering and balancing between the network and back end systems.
4) Detecting and isolating breaches – despite the best efforts of any security infrastructure, it is impossible to completely eliminate breaches. This is where most security implementations fail. The key is to constantly monitor the environment down to the physical devices to be able to identify breaches when they occur. This requires the ability to recognize what a breach looks like. For the Internet of things, attacks can come in many flavors, including spoofing, hijacking, injection, viral, sniffing, and denial of service. Adequate real-time monitoring for these types of attacks is critical to a good security practice.
Once a breach or attack is detected, rapid isolation is the next most important step. Ideally, breached devices can be taken out of commission, and remotely wiped. Breached servers can be cut off from sensitive back end systems and shut down. The key is to be able to detect problems as quickly as possible and then immediately quarantine them.
Outside of these four security considerations, let me add two more that are specifically related to privacy. Since so much of the Internet of things is built around consumer devices, the privacy risks are high. Consumers are increasingly back lashing against the surveillance economy inherent in many social networking tools, and the Internet of things threatens to take that to the next level.
Opt in – Most consumers have no idea what information is being collected about them, even by the social tools they use every day. But when the devices you use become connected, the opportunities for abuse get even worse. Now there are many great reasons for your car and appliances and personal health monitors to be connected, but unless you know that your data is being collected, where the data is going, and how it is being used, you are effectively being secretly monitored. The manufacturers of these connected things need to provide consumers with a choice. There can be benefits to being monitored, like discounted costs or advanced services, but consumers must be given the opportunity to opt in for those benefits, and understand that they are giving up some personal liberties in the process.
Data anonymization – when data is collected, much of the time, the goal is not to get specific personal information about an individual user, but rather to understand trends and anomalies that can help improve and optimize downstream experiences. Given that, organizations who employ the Internet of things should strive to remove any personally identifying information as they conduct their data analysis. This practice will reduce the number of privacy exposures, while still providing many of the benefits of the data.
The Internet of things requires a different approach to security and privacy. Already the headlines are rolling in about the issues, so it’s time to get serious about getting ahead of the problem.